The Variety of Integrable Killing Tensors on the 3-Sphere

Integrable Killing tensors are used to classify orthogonal coordinates in which the classical Hamilton-Jacobi equation can be solved by a separation of variables. We completely solve the Nijenhuis integrability conditions for Killing tensors on the sphere S³ and give a set of isometry invariants for...

Повний опис

Збережено в:

Бібліографічні деталі
Дата:	2014
Автор:	Schöbel, K.
Формат:	Стаття
Мова:	English
Опубліковано:	Інститут математики НАН України 2014
Назва видання:	Symmetry, Integrability and Geometry: Methods and Applications
Онлайн доступ:	https://nasplib.isofts.kiev.ua/handle/123456789/146598
Теги:	Додати тег Немає тегів, Будьте першим, хто поставить тег для цього запису!
Назва журналу:	Digital Library of Periodicals of National Academy of Sciences of Ukraine
Цитувати:	The Variety of Integrable Killing Tensors on the 3-Sphere / K. Schöbel // Symmetry, Integrability and Geometry: Methods and Applications. — 2014. — Т. 10. — Бібліогр.: 46 назв. — англ.

Репозитарії

Digital Library of Periodicals of National Academy of Sciences of Ukraine

id	nasplib_isofts_kiev_ua-123456789-146598
record_format	dspace
spelling	nasplib_isofts_kiev_ua-123456789-1465982025-02-10T01:25:30Z The Variety of Integrable Killing Tensors on the 3-Sphere Schöbel, K. Integrable Killing tensors are used to classify orthogonal coordinates in which the classical Hamilton-Jacobi equation can be solved by a separation of variables. We completely solve the Nijenhuis integrability conditions for Killing tensors on the sphere S³ and give a set of isometry invariants for the integrability of a Killing tensor. We describe explicitly the space of solutions as well as its quotient under isometries as projective varieties and interpret their algebro-geometric properties in terms of Killing tensors. Furthermore, we identify all Stäckel systems in these varieties. This allows us to recover the known list of separation coordinates on S³ in a simple and purely algebraic way. In particular, we prove that their moduli space is homeomorphic to the associahedron K₄. I would like to express my gratitude to Robert Milson for his motivation and the inspiring discussions about my findings. I would also like to thank Alexander P. Veselov for pointing out the link between my solution and moduli spaces of stable curves. Finally, I would like to thank the anonymous referees, who helped to improve the paper considerably with their comments and additional references. 2014 Article The Variety of Integrable Killing Tensors on the 3-Sphere / K. Schöbel // Symmetry, Integrability and Geometry: Methods and Applications. — 2014. — Т. 10. — Бібліогр.: 46 назв. — англ. 1815-0659 2010 Mathematics Subject Classification: 53A60; 14H10; 14M12 DOI:10.3842/SIGMA.2014.080 https://nasplib.isofts.kiev.ua/handle/123456789/146598 en Symmetry, Integrability and Geometry: Methods and Applications application/pdf Інститут математики НАН України
institution	Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection	DSpace DC
language	English
description	Integrable Killing tensors are used to classify orthogonal coordinates in which the classical Hamilton-Jacobi equation can be solved by a separation of variables. We completely solve the Nijenhuis integrability conditions for Killing tensors on the sphere S³ and give a set of isometry invariants for the integrability of a Killing tensor. We describe explicitly the space of solutions as well as its quotient under isometries as projective varieties and interpret their algebro-geometric properties in terms of Killing tensors. Furthermore, we identify all Stäckel systems in these varieties. This allows us to recover the known list of separation coordinates on S³ in a simple and purely algebraic way. In particular, we prove that their moduli space is homeomorphic to the associahedron K₄.
format	Article
author	Schöbel, K.
spellingShingle	Schöbel, K. The Variety of Integrable Killing Tensors on the 3-Sphere Symmetry, Integrability and Geometry: Methods and Applications
author_facet	Schöbel, K.
author_sort	Schöbel, K.
title	The Variety of Integrable Killing Tensors on the 3-Sphere
title_short	The Variety of Integrable Killing Tensors on the 3-Sphere
title_full	The Variety of Integrable Killing Tensors on the 3-Sphere
title_fullStr	The Variety of Integrable Killing Tensors on the 3-Sphere
title_full_unstemmed	The Variety of Integrable Killing Tensors on the 3-Sphere
title_sort	variety of integrable killing tensors on the 3-sphere
publisher	Інститут математики НАН України
publishDate	2014
url	https://nasplib.isofts.kiev.ua/handle/123456789/146598
citation_txt	The Variety of Integrable Killing Tensors on the 3-Sphere / K. Schöbel // Symmetry, Integrability and Geometry: Methods and Applications. — 2014. — Т. 10. — Бібліогр.: 46 назв. — англ.
series	Symmetry, Integrability and Geometry: Methods and Applications
work_keys_str_mv	AT schobelk thevarietyofintegrablekillingtensorsonthe3sphere AT schobelk varietyofintegrablekillingtensorsonthe3sphere
first_indexed	2025-12-02T11:37:08Z
last_indexed	2025-12-02T11:37:08Z
_version_	1850396301706919936
fulltext	Symmetry, Integrability and Geometry: Methods and Applications SIGMA 10 (2014), 080, 48 pages The Variety of Integrable Killing Tensors on the 3-Sphere Konrad SCHÖBEL Institut für Mathematik, Fakultät für Mathematik und Informatik, Friedrich-Schiller-Universität Jena, 07737 Jena, Germany E-mail: konrad.schoebel@uni-jena.de Received November 14, 2013, in final form July 15, 2014; Published online July 29, 2014 http://dx.doi.org/10.3842/SIGMA.2014.080 Abstract. Integrable Killing tensors are used to classify orthogonal coordinates in which the classical Hamilton–Jacobi equation can be solved by a separation of variables. We completely solve the Nijenhuis integrability conditions for Killing tensors on the sphere S3 and give a set of isometry invariants for the integrability of a Killing tensor. We describe explicitly the space of solutions as well as its quotient under isometries as projective varieties and interpret their algebro-geometric properties in terms of Killing tensors. Furthermore, we identify all Stäckel systems in these varieties. This allows us to recover the known list of separation coordinates on S3 in a simple and purely algebraic way. In particular, we prove that their moduli space is homeomorphic to the associahedron K4. Key words: separation of variables; Killing tensors; Stäckel systems; integrability; algebraic curvature tensors 2010 Mathematics Subject Classification: 53A60; 14H10; 14M12 Contents 1 Introduction 2 1.1 The problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Prior results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.3 Method and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.4 Generalisations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 1.5 Structure of the article . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2 Properties of algebraic curvature tensors 10 2.1 Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2 The action of the isometry group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.3 Aligned algebraic curvature tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.4 Diagonal algebraic curvature tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.5 The residual action of the isometry group . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3 Solution of the algebraic integrability conditions 16 3.1 Reformulation of the first integrability condition . . . . . . . . . . . . . . . . . . . . . . . 16 3.2 Integrability implies diagonalisability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 3.3 Solution of the second integrability condition . . . . . . . . . . . . . . . . . . . . . . . . . 19 4 The algebraic geometry of the KS-variety 21 5 Stäckel systems 25 5.1 Young tableaux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 5.2 Commuting Killing tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 mailto:konrad.schoebel@uni-jena.de http://dx.doi.org/10.3842/SIGMA.2014.080 2 K. Schöbel 6 Geometric constructions of integrable Killing tensors 32 6.1 Special Killing tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 6.2 Killing tensors of Benenti type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 6.3 Benenti systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 6.4 Extension of integrable Killing tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 7 Separation coordinates 40 7.1 Eigenvalues of special conformal Killing tensors . . . . . . . . . . . . . . . . . . . . . . . . 40 7.2 Killing tensors and separation coordinates on S2 . . . . . . . . . . . . . . . . . . . . . . . 42 7.3 The classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 7.4 The classification space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 References 46 1 Introduction 1.1 The problem Given a partial differential equation, it is a natural problem to seek for coordinate systems in which this equation can be solved by a separation of variables and to classify all such coor- dinate systems. It is not surprising, that for classical equations this problem has a very long history and is marked by the contributions of a number of prominent mathematicians, begin- ning with Gabriel Lamé in the first half of of the 19th century [29]. For example, consider the Hamilton–Jacobi equation 1 2 gij ∂S ∂xi ∂S ∂xj + V (x) = E (1.1) on a Riemannian manifold. We say that it separates (additively) in a system of coordinates x1, . . . , xn if it admits a solution of the form S(x1, . . . , xn; c) = n∑ i=1 Si(xi; c), det ( ∂2S ∂xi∂cj ) 6= 0, depending on n constants c = (c1, . . . , cn). If this is the case, the initial partial differential equation decouples into n ordinary differential equations. We call the corresponding coordinate system separation coordinates. It turns out that the Hamilton–Jacobi equation in the form (1.1) separates in a given ortho- gonal coordinate system if and only if it also separates for V ≡ 0 and if the potential V satisfies a certain compatibility condition. In fact, the Hamilton–Jacobi equation with V ≡ 0 plays a key role for separation of variables in general. The reason is that for orthogonal coordinates the (additive) separation of this equation is a necessary condition for the (multiplicative) separa- tion of other classical equations such as the Laplace equation, the Helmholtz equation or the Schrödinger equation. Sufficient conditions can be given in the form of compatibility conditions for the curvature tensor and for the potential. The classification of orthogonal separation coordinates for the Hamilton–Jacobi equation thus leads to a variety of ordinary differential equations. These are well-known equations that define special functions appearing all over in mathematics and physics – such as Bessel functions, Le- gendre polynomials, spherical harmonics or Mathieu functions, to name just a few. In particular, these functions serve as bases for series expansions in the explicit solution of boundary value problems on domains bounded by coordinate hypersurfaces. Separation coordinates are intimately related to the existence of rank two Killing tensors that satisfy a certain integrability condition. The Variety of Integrable Killing Tensors on the 3-Sphere 3 Definition 1.1. A symmetric tensor Kαβ on a Riemannian manifold is a Killing tensor if and only if ∇(γKαβ) = 0, (1.2) where ∇ is the Levi-Civita connection of the metric g and the round parenthesis denote complete symmetrisation of the enclosed indices. Depending on the context, we will consider a Killing tensor either as a symmetric bilinear form Kαβ or alternatively as a symmetric endomorphism Kα β = gαγKγβ. Definition 1.2. We say a diagonalisable endomorphism K is integrable, if around any point in some open and dense set we find local coordinates such that the coordinate vectors are eigenvectors of K. The above geometric definition of integrability can be cast into a system of partial differential equations, involving the Nijenhuis torsion of K defined by Nα βγ = Kα δ∇[γK δ β] +∇δKα [γK δ β] . (1.3) The following result was proven by Nijenhuis for endomorphisms with simple eigenvalues [37], but is true in general [41]. Proposition 1.3. A diagonalisable endomorphism field Kα β on a Riemannian manifold is integrable if and only if it satisfies the Nijenhuis integrability conditions 0 = N δ [βγ gα]δ, (1.4a) 0 = N δ [βγKα]δ, (1.4b) 0 = N δ [βγKα]εK ε δ , (1.4c) where the square brackets denote complete antisymmetrisation in the enclosed indices. At the end of the 19th century, Paul Stäckel showed that any system of orthogonal separation coordinates gives rise to what we now call a Stäckel system. And in 1934, Luther P. Eisenhart proved the converse. This established a one-to-one correspondence between orthogonal separa- tion coordinates and Stäckel systems in the following sense. Definition 1.4. A Stäckel system on an n-dimensional Riemannian manifold is an n-dimensional space of integrable Killing tensors which mutually commute in the algebraic sense12. Theorem 1.5 ([13, 45]). On a Riemannian manifold, there is a bijective correspondence between orthogonal separation coordinates and Stäckel systems. Benenti later proved that a Stäckel system can be represented by a single integrable Killing tensor with simple eigenvalues. More precisely: Theorem 1.6 ([3]). 1. Every Stäckel system contains a Killing tensor with simple eigenvalues. 1Eisenhart’s original definition of a Stäckel system was more restrictive, but is equivalent to the one given here [3]. 2There is an alternative notion of “commuting” for Killing tensors, namely with respect to the Poisson bracket, which is different but closely related for a Stäckel system. Indeed, one can replace the condition that the Killing tensors in a Stäckel system are integrable by the condition that they mutually commute under the Poisson bracket [3]. In this article the term “commuting” will always be used in the algebraic sense. 4 K. Schöbel 2. Conversely, every integrable Killing tensor with simple eigenvalues is contained in some Stäckel system. 3. A Stäckel system is uniquely determined by such a Killing tensor. In particular, the separation coordinates can be recovered from a generic representative in a Stäckel system by solving the eigenproblem and integrating the distributions normal to the eigenvectors. Although initially introduced to classify separation coordinates, Stäckel systems play a role far beyond separation of variables, as they constitute an important class of completely integrable dynamical systems. The construction and classification of Stäckel systems in constant curvature is, for instance, equivalent to the construction and classification of Gaudin magnets [21, 27, 28]. Stäckel systems also arise from Killing–Yano towers and in black hole integrable models [14]. Furthermore, there exists a natural quantisation of Killing tensors such that the quantum version of Stäckel systems are the maximal commutative subalgebras of second order symmetries of the Schrödinger operator [5]. The Nijenhuis integrability conditions (1.4) are homogeneous algebraic equations in K and ∇K, which are invariant under isometries. Separation of variables therefore naturally leads to the following problem. Problem 1.7. Determine the projective variety K(M) of all integrable Killing tensors on a given Riemannian manifold M . That is, solve explicitly the Nijenhuis integrability conditions (1.4) for Killing tensors (1.2). Moreover, find natural isometry invariants characterising the integrability of a Killing tensor. Of course, there are many examples of separation coordinates and in constant curvature we even have a complete classification. In principle this gives a description of the corresponding Stäckel systems and integrable Killing tensors, although in practice they are often obtained by intricate limiting processes. Nevertheless, this only yields a description of K(M) as a set, whereas in Problem 1.7 we seek to elucidate the natural algebraic geometric structure of this set. Note that in dimension two the Nijenhuis integrability conditions are void, so that K(M) is simply the projective space of Killing tensors. This means that Problem 1.7 is trivial in dimension two once the space of Killing tensors is known. However, to the best of our knowledge, already in dimensions three nothing is known about the set K(M) as a projective variety, apart from trivial cases, not even for Euclidean space R3 or the sphere S3 – the two examples for which Eisenhart already derived the complete list of separation coordinates. Moreover, an explicit solution of the equations (1.4) has been considered intractable by various experts in this domain [15]. Although there exists no general construction for Stäckel systems, there is one which yields a large family of Stäckel systems. This family is interesting in itself, because it appears under different names in a number of different guises in the theory of integrable systems: as “New- tonian systems of quasi-Lagrangian type” [40], as “systems admitting special conformal Killing tensors” [8], as “cofactor systems” [31], as “bi-Hamiltonian structures” [6, 16], as “bi-quasi- Hamiltonian systems” [9, 10], as “L-systems” [4] or as “Benenti systems” [7]. We follow the latter nomenclature of Bolsinov and Matveev, because their description is best suited to our context. This brings us to the following question. Question 1.8. Do Benenti systems parametrise the entire variety of integrable Killing tensors? Moreover, by construction, the set of Benenti systems is not only invariant under isometries, but even under the projective group. So, if the answer to Question 1.8 is “no” and the isometry group is a proper subgroup of the projective group, the following question arises. Question 1.9. Is the variety of integrable Killing tensors invariant under the projective group? The Variety of Integrable Killing Tensors on the 3-Sphere 5 Obviously, there are three groups acting on the variety K(M) of integrable Killing tensors: 1. R, acting by addition of multiples of the metric g. 2. The isometry group Isom(M). 3. R∗, acting by multiplication. The actions 1 and 3 result from the fact that adding the identity to an endomorphism or multiplying it by a constant does not alter the eigenspaces and hence does not affect integrability. The actions 1 and 2 commute and the action 3 descends to the quotient. Hence the essential information about integrable Killing tensors is encoded in the projectivisation of the quotient K(M)/ ( Isom(M)× Rg ) . (1.5) This leads to the following problem. Problem 1.10. Determine explicitly the quotient (1.5), which describes integrable Killing ten- sors modulo obvious symmetries. Recall that orthogonal separation coordinates are in bijective correspondence with Stäckel systems. Thus, in view of the classification problem for orthogonal separation coordinates, one is interested in linear subspaces of commuting integrable Killing tensors, rather than in the integrable Killing tensors themselves. This leads to the following problem. Problem 1.11. Determine explicitly the space of all Stäckel systems on a given Riemannian manifold M (modulo isometries). That is, find all maximal linear subspaces of commuting Killing tensors in K(M). The separation coordinates can be recovered from an integrable Killing tensor with simple eigenvalues, because the coordinate hypersurfaces are orthogonal to the eigenvectors. This leads to the final step in the determination of separation coordinates via integrable Killing tensors and Stäckel systems. Problem 1.12. Find a (canonical) representative in each Stäckel system for which it is possible to solve the eigenvalue problem and which has simple eigenvalues. Deduce the classification of orthogonal separation coordinates and give natural isometry invariants. In this article we will completely solve Problems 1.7, 1.10, 1.11 and 1.12 for the three- dimensional sphere S3 equipped with the round metric and give answers to Questions 1.8 and 1.9. This is done without relying on any computer algebra. One of the reasons we choose S3 is that it has the simplest isometry group among the 3-dimensional constant curvature manifolds. Our solution will reveal diverse facets of the problem of separation of variables, related to repre- sentation theory, algebraic geometry, geometric invariant theory, geodesic equivalence, Stasheff polytopes or moduli spaces of stable algebraic curves. 1.2 Prior results In his fundamental paper Eisenhart also derived a list of orthogonal separation coordinates for R3 and S3 [13]. Separation coordinates on S3 were also classified in [38]. These results have been generalised by Kalnins and Miller, who gave an iterative diagrammatical procedure for the construction of orthogonal separation coordinates on spaces of constant curvature in any dimension [17, 22]. Furthermore, Eisenhart computed the Stäckel systems on R3 in their respective separation coordinates. For S3 the quantum point of view was adopted in [23]. 6 K. Schöbel In dimension two virtually everything is known. Koenigs classified all spaces with at least three independent Killing tensors. Apart from constant curvature spaces these are four so- called Darboux spaces and eleven multiparameter families of Koenigs spaces [26]. For Darboux spaces the separation coordinates have been classified in [19, 20] and for Koenigs spaces this is straightforward. Darboux spaces have analogs in higher dimensions [1], which can be obtained from constant curvature spaces via a Stäckel transform, and for the three dimensional version it is also straightforward to find all separation coordinates. Finally, superintegrability theory has led to the construction of non-conformally flat systems which are separable [18]. Despite these advances, the topology and the geometry of the space of integrable Killing tensors or its quotient by isometries has never been studied for any non-trivial case. Probably the only advance towards an algebraic geometric study of these spaces has been the construction of isometry invariants for Killing tensors which discriminate the different orthogonal separation coordinates in three dimensional Euclidean space [15]. This approach had not become viable until recently, because it relies on an extensive use of modern computer algebra. The algorithm, which depends on an a priori knowledge of the separation coordinates, involves the solution of a linear system of approximately 250,000 equations in 50,000 unknowns and the complexity of the algorithm renders it impractical in dimensions greater than three [11]. Nothing similar is known for the sphere or hyperbolic space. We are not aware of any applications of the resulting invariants to elucidate the structure of the resulting quotient. 1.3 Method and results Our method is based on a purely algebraic description of the vector space of Killing tensors on constant curvature manifolds in combination with the corresponding algebraic equivalent of the Nijenhuis integrability conditions. We consider the standard 3-sphere, isometrically embedded as the unit sphere in a 4-dimensional Euclidean vector space (V, g): S3 = {x ∈ V : g(x, x) = 1} ⊂ V. The vector space of Killing tensors K on S3 is then naturally isomorphic to the vector space of algebraic curvature tensors on the ambient space V [34]. These are (constant) tensors Ra1b1a2b2 on V having the symmetries of a Riemannian curvature tensor. The isomorphism is explicitly given by Kx(v, w) := Ra1b1a2b2x a1xa2vb1wb2 , x ∈ S3, v, w ∈ TxS3, (1.6) and equivariant with respect to the natural actions of the isometry group O(V, g) on both spaces. Under this correspondence the Nijenhuis integrability conditions (1.4) for a Killing tensor are equivalent to the following algebraic integrability conditions for the associated algebraic curvature tensor [41]: a2 b2 c2 d2 gijR i b1a2b2R j d1c2d2 = 0, (1.7a) a2 b2 c2 d2 a1 b1 c1 d1 gijgklR i b1a2b2R j k a1 c1R l d1c2d2 = 0. (1.7b) Here the Young symmetrisers on the left hand side denote complete antisymmetrisation in the indices a2, b2, c2, d2 respectively complete symmetrisation in the indices a1, b1, c1, d1. These homogeneous algebraic equations define the space of integrable Killing tensors as a projective variety. The Variety of Integrable Killing Tensors on the 3-Sphere 7 The key to the solution of the algebraic integrability conditions is a reinterpretion in terms of the decomposition of algebraic curvature tensors into irreducible representations of the isometry group – the selfdual and anti-selfdual Weyl part, the trace-free Ricci part and the scalar part. Under the action of the isometry group these components can be brought to a certain normal form, which simplifies considerably if the algebraic integrability conditions are imposed. We regard an algebraic curvature tensor on V as a symmetric endomorphism on Λ2V . In this sense we will say that an algebraic curvature tensor R is diagonal if Rijkl = 0 unless {i, j} = {k, l}. In dimension four a diagonal algebraic curvature tensor can be parametrised as R0α0α = wα + tα + s 12 , Rβγβγ = wα − tα + s 12 , (1.8) where (α, β, γ) denotes any cyclic permutation of (1, 2, 3). In this parametrisation the wα are the eigenvalues of the self-dual Weyl part, which turn out to be the same as those of the anti- self-dual Weyl part, the tα parametrise the eigenvalues of the trace free Ricci tensor and s is the scalar curvature. In a first step we prove that diagonal algebraic curvature tensors define a (local) slice for the action of the isometry group on integrable Killing tensors. Theorem 1.13 (integrability implies diagonalisability). Under the action of the isometry group any integrable Killing tensor on S3 is equivalent to one with a diagonal algebraic curvature tensor. The first algebraic integrability condition is identically satisfied on diagonal algebraic cur- vature tensors and the second reduces to a single algebraic equation on the diagonal entries. This yields a description of integrable Killing tensors with diagonal algebraic curvature tensor as a projective variety, equipped with the residual action of the isometry group. We call this variety the Killing–Stäckel variety (KS-variety). Theorem 1.14 (integrability in the diagonal case). A Killing tensor on S3 with diagonal alge- braic curvature tensor (1.8) is integrable if and only if the associated matrix∆1 −t3 t2 t3 ∆2 −t1 −t2 t1 ∆3  (1.9) has zero trace and zero determinant, where ∆1 := w2 − w3, ∆2 := w3 − w1, ∆3 := w1 − w2. This defines a projective variety in P4 which carries a natural S4-action, given by conjuga- ting (1.9) with the symmetries of the regular octahedron in R3 with vertices ±ei and adjacent faces oppositely oriented. Two diagonal algebraic curvature tensors are equivalent under the isometry group if and only if their associated matrices (1.9) are equivalent under this S4-action. Combining Theorems 1.13 and 1.14 yields a solution to Problem 1.10: The projectivisation of the quotient (1.5) is isomorphic to the quotient of the KS-variety by its natural S4-action. Moreover, in the course of the proof of both theorems we establish isometry invariants for the integrability of a Killing tensor. Proposition 1.15 (isometry invariants). Let W and T be the Weyl and Ricci part of the alge- braic curvature tensor corresponding to a Killing tensor on S3 ⊂ V , regarded as endomorphisms on Λ2V . Then this Killing tensor is integrable if and only if 1) tr [W,T]2 = 0, 2) tr(∗W2) = tr(∗W3) = 0, where ∗ denotes the Hodge operator on Λ2V and 3) the endomorphisms I, W and W2 −T2 are linearly dependent. 8 K. Schöbel With this result the check whether a given Killing tensor on S3 is integrable becomes algo- rithmically trivial. Näıvely, one can find the Stäckel systems by solving the commutator equation [K, K̃] = 0 for a fixed integrable Killing tensor K. But a far simpler way is to exploit the algebraic geometry of the KS-variety. Recall that Stäckel systems are linear spaces of dimension n and always contain the metric. We show that the Killing tensors in a Stäckel system have simultaneously diagona- lisable algebraic curvature tensors. Therefore Stäckel systems on S3 correspond to certain pro- jective lines in the KS-variety. In general, the projective lines in a projective variety constitute themselves a projective variety, called the Fano variety. That is, we are looking for a subvariety of the Fano variety of the KS-variety. By definition, the KS-variety is a linear section of the variety of 3×3 matrices with vanishing determinant, the so called full determinantal variety. Fortunately, the Fano variety of a full determinantal variety is well understood just in the case we need here. So we only have to use the commutator equation to check which projective lines correspond to Stäckel systems, rather than to actually solve it. In order to do so, we derive an algebraic equivalent of the commutator equation in terms of the corresponding algebraic curvature tensors. This entails a detailed description of the algebraic geometry of the KS-variety and the Stäckel systems therein. In particular, we can relate the algebraic geometric properties of the KS-variety to the geometric properties of the corresponding integrable Killing tensors. For this, we remark that there is a natural way how to extend (integrable) Killing tensors from an embedded sphere Sm ⊂ Sn to the ambient sphere Sn for m < n. Theorem 1.16. For M = S3 the projectivisation of the quotient (1.5) is isomorphic to the quotient of the KS-variety by its natural action of S4. Furthermore: 1. The singularities of the KS-variety correspond to extensions of the metric on S2 ⊂ S3 respectively on S1 ⊂ S3. 2. Stäckel systems determine projective lines in the KS-variety which consist of matrices (1.9) annihilating a given vector. 3. Projective planes in the KS-variety that consist of matrices (1.9) annihilating a fixed vector correspond to extensions of (integrable) Killing tensors from S2 to S3. 4. The projective plane of antisymmetric matrices in the KS-variety corresponds to the pro- jective space of special Killing tensors. These are Killing tensors K of the form K = L− (trL)g, where L is a special conformal Killing tensor3. Based on the algebraic description of Stäckel systems we derive an algebraic description of Benenti systems as well and answer Question 1.8. The answer is “almost”, in the following sense. On one hand, there are Stäckel systems on S3, which are not Benenti systems. So the answer is clearly “no”. But on the other hand these are exactly the extensions of Killing tensors from S2 and for S2 the answer is easily seen to be “yes”. From the algebraic description of Benenti systems we derive the following parametrisation of K(S3), which is equivariant under the isometry group. This solves Problem 1.7. Theorem 1.17 (equivariant parametrisation). The algebraic variety K(S3) is the closure of the set of algebraic curvature tensors of the form λ2h7 h+ λ0g 7 g, h ∈ Sym2(V ), λ0, λ2 ∈ R. 3Cf. Definition 6.1. The Variety of Integrable Killing Tensors on the 3-Sphere 9 Here 7 denotes the Kulkarni–Nomizu product4 and 1 2g 7 g is the algebraic curvature tensor of the metric. That is, K(S3) is the closure of the projective cone from the metric over the image of the Kulkarni–Nomizu square h 7→ h7h, a map from symmetric tensors to algebraic curvature tensors. Remark that the apex lies itself in this image. Since the set of algebraic curvature tensors of the form h7h is invariant under the projective group PGL(V ) of S3, the answer to Question 1.9 is “almost” as well. The reason is that although K(S3) is not projectively invariant, this “invariant part” still contains the essential part of each Stäckel system unless it is an extension from S2. But again, on S2 the answer is easily seen to be “yes”. Our answers to Questions 1.8 and 1.9 indicate that one should think of the space of integrable Killing tensors on the sphere as composed of strata, where the non- generic strata come from lower dimensional spheres. Then each stratum is generated by Benenti systems and invariant under the projective group in the respective dimension. As a simple consequence of Theorem 1.16 we get a canonical representative in each Stäckel system. Corollary 1.18. Each Stäckel system on S3 contains an essentially unique special Killing ten- sor. If the eigenvalues of the corresponding special conformal Killing tensor are simple, they define the respective separation coordinates [8]. Moreover, every special conformal Killing tensor L on Sn is the restriction of a constant symmetric tensor L̂ on the ambient space V [32]. This is the reason why we can solve the eigenvalue problem for integrable Killing tensors explicitly. In this manner we get a purely algebraic classification of separation coordinates, which solves Problem 1.12. The details will be given later. A solution to Problem 1.11 can be derived from the description of Stäckel systems in the KS- variety. This gives the classification space for separation coordinates on S3 modulo isometries. Theorem 1.19 (moduli space of separation coordinates). The space of Stäckel systems on S3 modulo isometries is homeomorphic to the quotient of the blow-up of P2 in the four points (±1 :±1 :±1) ∈ P2 under the natural action of the symmetry group S4 of a tetrahedron in R3 inscribed in a cube with vertices (±1,±1,±1). The fundamental domain of the S4-action is a pentagon bounded by the non-generic sep- aration coordinates. In fact, the multiplicities of the eigenvalues of the constant symmetric tensor L̂ on V identify this pentagon to the associahedron (or Stasheff polytope) K4 [46]. These observations suggests a relation between separation coordinates on spheres and moduli spaces of stable curves of genus zero with marked points [12, 24, 25]. Indeed, our thorough analysis of the case S3 presented here has led to the following generalisation of Theorem 1.19 [42]. Theorem 1.20. The Stäckel systems on Sn with diagonal algebraic curvature tensor form a smooth projective variety isomorphic to the real Deligne–Mumford–Knudsen moduli space M̄0,n+2(R) of stable genus zero curves with n+ 2 marked points. 1.4 Generalisations Most of our results are stated for constant curvature manifolds of arbitrary dimension. What impedes a straightforward generalisation to higher dimensional spheres is that in the proof of diagonalisability we make essential use of the Hodge decomposition which only exists in dimension four. Theorem 1.20 together with a proof of the following conjecture would show, that the classification space for separation coordinates on Sn is the associahedron Kn+1. 4Cf. Definition 6.6. 10 K. Schöbel Conjecture 1.21. Under the action of the isometry group any integrable Killing tensor on Sn is equivalent to one with a diagonal algebraic curvature tensor. In Lorentzian signature the Hodge star squares to minus one. A generalisation to hyperbolic space therefore naturally requires complexification. But over the complex numbers, all signa- tures are equivalent. This indicates that one should solve the algebraic integrability conditions over the complex numbers. The variety of complex solutions then contains not only the solution for hyperbolic space, but for all (non-flat) pseudo-Riemannian constant curvature spaces as real subvarieties, singled out by appropriate choices of a real structure. In dimension three there is an elegant way to do this using spinors and the Petrov classification of curvature tensors [35]. The main complication comes from the fact that diagonalisability does not hold anymore in the complexified case. A challenge remains without doubt a non-trivial algebraic geometric classification of inte- grable Killing tensors and orthogonal separation coordinates on some non-constant curvature manifold. This is out of scope of common, purely geometric techniques. Note that the Nijenhuis integrability conditions (1.4) can always be rewritten as homogeneous algebraic equations of degree two, three and four on the space of Killing tensors, invariant under the isometry group action on Killing tensors. This endows the set of integrable Killing tensors on an arbitrary Riemannian manifold with the structure of a projective variety, equipped with an action of the isometry group. Our approach can therefore be extended to non-constant curvature as well. Natural candidates are semi-simple Lie groups, homogeneous spaces or Einstein manifolds. No- tice that S3 falls in either of these categories. Taking our solution for S3 as a roadmap, we propose the following three-stage procedure: 1. Determine the space of Killing tensors and identify it explicitly as a representation of the isometry group (as in [34] for constant curvature manifolds). 2. Translate the Nijenhuis integrability conditions for Killing tensors into purely algebraic integrability conditions (as in [41] for constant curvature manifolds). 3. Solve the algebraic integrability conditions, study the algebraic geometric properties of the resulting variety and derive the complete classification of orthogonal separation coordinates (as in this article for S3). We consider the present work as a proof of concept that this algebraic geometric approach to the problem of separation of variables is viable. 1.5 Structure of the article The article is organised as follows: In Section 2 we examine some properties of algebraic cur- vature tensors that will be used to solve the algebraic integrability conditions in Section 3. We then describe the algebraic geometry of this solution in Section 4 and identify the Stäckel sys- tems therein in Section 5. In Section 6 we give various geometric constructions for integrable Killing tensors and interpret them algebraically. In the last section we demonstrate how the classification of separation coordinates can be recovered within our framework. 2 Properties of algebraic curvature tensors Algebraic curvature tensors on a vector space V are four-fold covariant tensors Ra1b1a2b2 having the same algebraic symmetries as a Riemannian curvature tensor: antisymmetry: Rb1a1a2b2 = −Ra1b1a2b2 = Ra1b1b2a2 , (2.1a) pair symmetry: Ra2b2a1b1 = Ra1b1a2b2 , (2.1b) Bianchi identity: Ra1b1a2b2 +Ra1a2b2b1 +Ra1b2b1a2 = 0. (2.1c) The Variety of Integrable Killing Tensors on the 3-Sphere 11 Given a scalar product g on V , we can raise and lower indices. The symmetries (2.1a) and (2.1b) then allow us to regard an algebraic curvature tensor Ra1a2b1b2 on V as a symmetric endomor- phism Ra1b1a2b2 on the space Λ2V of 2-forms on V . Since we will frequently change between both interpretations, we denote endomorphisms by the same letter in boldface. 2.1 Decomposition In the special case where dimV = 4, the Hodge star operator “∗” defines a decomposition Λ2V = Λ2 +V ⊕ Λ2 −V of Λ2V into its ±1 eigenspaces. We can therefore write an algebraic curvature tensor R and the Hodge star as block matrices R = ( W+ T∓ T± W− ) + s 12 ( I+ 0 0 I− ) , ∗ = ( +I+ 0 0 −I− ) (2.2a) with the 3×3-blocks satisfying W t + = W+, W t − = W−, T∓ = T t±, (2.2b) trW+ + trW− = 0, (2.2c) trW+ − trW− = 0. (2.2d) Here I+ and I− denote the identity on Λ2 +V respectively Λ2 −V . The conditions (2.2b) assure symmetry, condition (2.2c) says that s = 2 tr R and condition (2.2d) is a reformulation of the Bianchi identity (2.1c). Indeed, the Bianchi identity is equivalent to the vanishing of the antisymmetrisation of Ra1b1a2b2 in all four indices. In dimension four this can be written as εa2b2a1b1Ra1b1a2b2 = 0 or εa2b2a1b1R a1b1 a2b2 = 0, where εa2b2a1b1 is the totally antisymmetric tensor. Notice that εa2b2a1b1 is nothing else than the Hodge star operator and hence tr(∗R) = 0, (2.3) where tr : End(Λ2V )→ R is the usual trace. Now (2.2d) is (2.3) applied to R in (2.2). The space of algebraic curvature tensors is an irreducible GL(V )-representation and (2.2) gives a decomposition of this representation into irreducible representations of the subgroup SO(V, g) when dimV = 4. As the notation already suggests, we can relate these components to the familiar Ricci decomposition R = W + T + S of an algebraic curvature tensor R into • a scalar part Sa1b1a2b2 := s 12(ga1a2gb1b2 − ga1b2gb1a2) given by the scalar curvature s = ga1a2gb1b2Ra1b1a2b2 , 12 K. Schöbel • a trace free Ricci part Ta1b1a2b2 = 1 2(Ta1a2gb1b2 − Ta1b2gb1a2 − Tb1a2ga1b2 + Tb1b2ga1a2) given by the trace free Ricci tensor Ta1a2 = gb1b2Ra1b1a2b2 − s 4ga1a2 , • and a Weyl part, given by the totally trace free Weyl tensor W := R− T − S. It is not difficult to check that [∗,T] = 0 and hence W = ( W+ 0 0 W− ) , T = ( 0 T∓ T± 0 ) , S = s 12 ( I+ 0 0 I− ) . (2.4) We see that W+ and W− are the self-dual and anti-self-dual part of the Weyl tensor. Implicitly, the above interpretation also provides an isomorphism Sym0(V ) ∼=−→ Hom ( Λ2 +V,Λ 2 −V ) , T 7→ T±, (2.5) between trace-free Ricci tensors and homomorphisms from self-dual to anti-self-dual 2-forms. 2.2 The action of the isometry group The Killing tensor equation (1.2) as well as the integrability conditions (1.4) are invariant under the action of the isometry group. In other words, the variety of integrable Killing tensors is invariant under isometries. This will allow us later to put the algebraic curvature tensor of an integrable Killing tensor on S3 to a certain normal form and to solve the algebraic integrability conditions for this particular form. Recall that the isometry group of Sn ⊂ V is O(V ) and that SO(V ) is the subgroup of orientation preserving isometries. We first examine the induced action of SO(V ) on algebraic curvature tensors of the form (2.2). The standard action of SO(V ) on V induces a natural action on Λ2V which is the adjoint action of SO(V ) on its Lie algebra under the isomorphism so(V ) ∼= Λ2V. Now consider the following commutative diagram of Lie group morphisms for dimV = 4: Spin(V ) ∼=−−−−→ Spin ( Λ2 +V ) ×Spin ( Λ2 −V )y2:1 y2:1 y2:1 π : SO(V ) 2:1−−−−→ SO ( Λ2 +V ) × SO ( Λ2 −V ) ↪→ SO ( Λ2V ) U 7−→ (U+ , U−) 7→ ( U+ 0 0 U− ) . Here the double covering π in the second row is induced from the exceptional isomorphism in the first row via the universal covering maps (vertical). Under the induced isomorphism π∗ of Lie algebras, so(V ) ∼= so ( Λ2 +V ) ⊕ so ( Λ2 −V ) , The Variety of Integrable Killing Tensors on the 3-Sphere 13 the adjoint action of an element U ∈ SO(V ) corresponds to the adjoint actions of U+ ∈ SO(Λ2 +V ) and U− ∈ SO(Λ2 −V ). Since Λ2 +V has dimension three, the Hodge star operator gives an isomor- phism so ( Λ2 +V ) ∼= Λ2Λ2 +V ∼= Λ2 +V under which the adjoint action of U+ corresponds to the standard action on Λ2 +V and similarly for U−. The Hodge decomposition completes the above isomorphisms to a commutative diagram π∗ : so(V ) ∼=−−−−→ so ( Λ2 +V ) ⊕ so ( Λ2 −V )y∼= y∼= y∼= Λ2V ∼=−−−−→ Λ2 +V ⊕ Λ2 −V, which shows that the natural action of U ∈ SO(V ) on Λ2V is given by( U+ 0 0 U− ) ∈ π ( SO(V ) ) ⊂ SO ( Λ2V ) . (2.6) Hence the natural action of U on End(Λ2V ) is given by conjugation with this matrix. Restricting to algebraic curvature tensors we get: Proposition 2.1 (action of the isometry group). An element U ∈ SO(V ) acts on algebraic curvature tensors in the form (2.2) by conjugation with (2.6), i.e. via W+ 7→ U t+W+U+, W− 7→ U t−W−U−, T± 7→ U t−T±U+, (2.7) where π(U) = (U+, U−) is the image of U under the double covering π : SO(V )→ SO ( Λ2 +V ) × SO ( Λ2 −V ) . We can describe the twofold cover π explicitly in terms of orthonormal bases on the spaces V , Λ2 +V and Λ2 −V . It maps an orthonormal basis (e0, e1, e2, e3) of V to the orthonormal bases (η+1, η+2, η+3) of Λ2 +V and (η−1, η−2, η−3) of Λ2 −V , defined by η±α := 1√ 2 (e0 ∧ eα ± eβ ∧ eγ) (2.8) for each cyclic permutation (α, β, γ) of (1, 2, 3). From this description also follows that kerπ = {±I}. The action of O(V ) on algebraic curvature tensors is now determined by the action of some orientation reversing element in O(V ), say the one given by reversing the sign of e0 and preserving e1, e2 and e3. This element maps η±α to −η∓α. Hence its action on algebraic curvature tensors in the form (2.2) is given by conjugation with( 0 −I −I 0 ) respectively by mapping W+ 7→W−, T± 7→ T∓. (2.9) 14 K. Schöbel 2.3 Aligned algebraic curvature tensors W+ and W− are symmetric and hence simultaneously diagonalisable under the action (2.7) of SO(V ). On the other hand, the singular value decomposition shows that T± is also diagona- lisable under this action, although in general not simultaneously with W+ and W−. The following lemma gives a criterion when this is the case. Lemma 2.2. Let W+ and W− be symmetric endomorphisms on two arbitrary Euclidean vector spaces Λ+ respectively Λ− and suppose the linear map T± : Λ+ → Λ− satisf ies T±W+ = W−T±. Then there exist orthonormal bases for Λ+ and Λ− such that W+, W− and T± are simultaneously diagonal with respect to these bases. Proof. It suffices to show that we can chose diagonal bases for W+ and W− such that the matrix of T± has at most one non-zero element in each row and in each column, for the desired result can then be obtained by an appropriate permutation of the basis elements. The above condition implies that T± maps eigenspaces ofW+ to eigenspaces ofW− with the same eigenvalue. Without loss of generality we can thus assume that Λ+ and Λ− are eigenspaces ofW+ respectivelyW− with the same eigenvalue. But then W+ and W− are each proportional to the identity and therefore diagonal in any basis. In this case the lemma follows from the singular value decomposition of T±. � Lemma 2.3. The following conditions are equivalent for an algebraic curvature tensor (2.2): 1) T±W+ = W−T±, 2) [W,T] = 0, 3) W and T are simultaneously diagonalisable under SO(Λ2V ). Proof. The equivalence of 1 and 2 follows from (2.4). For the equivalence of 2 and 3 it suffices to note that W and T are both symmetric and hence diagonalisable. � The following terminology is borrowed from general relativity. Definition 2.4. We say that an algebraic curvature tensor is aligned, if it satisfies one of the equivalent conditions in Lemma 2.3. By Lemma 2.3, for any aligned algebraic curvature tensor we can find an orthonormal basis of V such that W+ = w+1 0 0 0 w+2 0 0 0 w+3  , W− = w−1 0 0 0 w−2 0 0 0 w−3  , T± = t1 0 0 0 t2 0 0 0 t3  (2.10a) with w+1 + w+2 + w+3 = 0, w−1 + w−2 + w−3 = 0. (2.10b) To simplify notation we agree that henceforth the indices (α, β, γ) will stand for an arbitrary cyclic permutation of (1, 2, 3). Changing the basis in Λ2V from (2.8) to ei ∧ ej , 0 6 i < j 6 4, we obtain the independent components of an aligned algebraic curvature tensor: R0α0α = w+α + w−α 2 + tα + s 12 , Rβγβγ = w+α + w−α 2 − tα + s 12 , R0αβγ = w+α − w−α 2 . (2.11) The Variety of Integrable Killing Tensors on the 3-Sphere 15 The trace free Ricci tensor of this aligned algebraic curvature tensor is diagonal and given by T00 = tα + tβ + tγ , Tαα = tα − tβ − tγ , tα = T00 + Tαα 2 . (2.12) This is nothing else than the restriction of the isomorphism (2.5) to diagonal tensors. 2.4 Diagonal algebraic curvature tensors Definition 2.5. We call an algebraic curvature tensor on V diagonal in a basis ei if it is diagonal as an element of End(Λ2V ) with respect to the associated basis ei ∧ ej , i < j. We call it diagonalisable if it is diagonalisable under the adjoint action of SO(V ) on Λ2V , i.e. under the subgroup π ( SO(V ) ) ⊂ SO(Λ2V ). Of course, being a symmetric endomorphism on Λ2V , an algebraic curvature tensor is always diagonalisable under the full group SO(Λ2V ). Proposition 2.6 (diagonalisability criterion). An algebraic curvature tensor is diagonalisable if and only if it is aligned and W+ has the same characteristic polynomial as W−. Proof. Consider an aligned algebraic curvature tensor and suppose that W+ has the same characteristic polynomial as W−. Then we can find a transformation (2.7) such that, as 3 × 3 matrices, W+ = W−. The condition W−T∓ = T±W+ then implies that T± can be simultaneously diagonalised with W+ = W−. This means we can assume without loss of generality that w+α = w−α =: wα for α = 1, 2, 3 in (2.10), so that (2.11) reads R0α0α = wα + tα + s 12 , R0αβγ = 0, Rβγβγ = wα − tα + s 12 (2.13a) with wα subject to w1 + w2 + w3 = 0. (2.13b) Obviously this algebraic curvature tensor is diagonal. On the other hand, any diagonal algebraic curvature tensor can be written in the form (2.13), because s = 2(R0101 +R2323 +R0202 +R3131 +R0303 +R1212), tα = R0α0α −Rβγβγ 2 , wα = R0α0α +Rβγβγ 2 − s 12 . � 2.5 The residual action of the isometry group For later use we need the stabiliser of the space of diagonal algebraic curvature tensors under the action of the isometry group. Such tensors are of the form (2.10) with w−α = w+α and are invariant under the transformation (2.9). Hence it is sufficient to consider only the subgroup SO(V ) of orientation preserving isometries. Let U ∈ SO(V ) with π(U) = (U+, U−) be an element preserving the space of diagonal algebraic curvature tensors under the action (2.7). Then U+ ∈ SO(Λ2 +V ) and U− ∈ SO(Λ2 −V ) preserve the space of diagonal matrices on Λ2 +V respectively Λ2 −V . In the orthogonal group, the stabiliser of the space of diagonal matrices under conjugation is the subgroup of signed permutation matrices. They act by permuting the diagonal elements in disregard of the signs. In particular, in SO(3) the stabiliser subgroup of the space of diagonal matrices on R3 is isomorphic 16 K. Schöbel to S4. Under the isomorphism S4 ∼= S3 n V4, the permutation group S3 is the subgroup of (unsigned) permutation matrices and acts by permuting the diagonal elements, whereas the Klein four group V4 is the subgroup of diagonal matrices in SO(3) and acts trivially. Using the above, it is not difficult to see that the stabiliser of the space of diagonal algebraic curvature tensors under the action (2.7) is isomorphic to S4 × kerπ, with kerπ = {±I} acting trivially. Under the isomorphism S4 ∼= S3nV4, the factor S3 acts by simultaneous permutations of (w1, w2, w3) and (t1, t2, t3) and the factor V4 by simultaneous flips of two signs in (t1, t2, t3). 3 Solution of the algebraic integrability conditions This section is dedicated to proving Theorems 1.13 and 1.14. 3.1 Reformulation of the first integrability condition In the same way as for the Bianchi identity we can reformulate the first integrability condi- tion (1.7a) as Rib1 a2b2 εa2b2c2d2R c2d2 id1 = 0 or r(R ∗R) = 0, (3.1) where r : End(Λ2V ) −→ End(V ), Eij kl 7→ Eij kj (3.2) denotes the Ricci contraction. Notice that E := R ∗ R is symmetric, but does in general not satisfy the Bianchi identity. A proof of the following lemma can be found in [43]. Lemma 3.1. The kernel of the Ricci contraction (3.2) is composed of endomorphisms E satis- fying ∗E∗ = Et, tr E = 0. Applying this to the symmetric endomorphism E := R ∗R = (W+ + s 12I+)2 − T∓T± (W+ + s 12I+)T∓ −T∓(W− + s 12I−) T±(W+ + s 12I+) T±T∓ − (W− + s 12I−)2−(W− + s 12I−)T±  , in (3.1) and using (2.2c) we get W+T∓ = T∓W−, (3.3a) trW 2 + = trW 2 −. (3.3b) This shows: Proposition 3.2. The first algebraic integrability condition for an algebraic curvature tensor R = W + T with Weyl part W and (not necessarily trace free) Ricci part T is equivalent to [W,T] = 0, tr ( ∗W2 ) = 0. In particular, an algebraic curvature tensor satisfying the first algebraic integrability condition is aligned. The Variety of Integrable Killing Tensors on the 3-Sphere 17 As a consequence of Proposition 2.6 we get: Corollary 3.3. A diagonalisable algebraic curvature tensor satisfies the first algebraic integra- bility condition. 3.2 Integrability implies diagonalisability The aim of this subsection is to prove Theorem 1.13. In view of Propositions 2.6 and 3.2 it is sufficient to show that if an algebraic curvature tensor satisfies the algebraic integrability conditions, then W+ and W− have the same characteristic polynomial. We will first prove that the first algebraic integrability condition implies that W+ has the same characteristic polynomial either as +W− or as −W−. We then prove that the latter contradicts the second algebraic integrability condition. We saw that an algebraic curvature tensor which satisfies the first algebraic integrability con- dition is aligned and thus of the form (2.10) in a suitable orthogonal basis. The first integrability condition in the form (3.3) then translates to w2 +1 + w2 +2 + w2 +3 = r2, w2 −1 + w2 −2 + w2 −3 = r2 (3.4a) for some r > 0 and w+αtα = tαw−α. (3.4b) If we regard (w+1, w+2, w+3) and (w−1, w−2, w−3) as vectors in R3, then each equation in (2.10b) describes the plane through the origin with normal (1, 1, 1) and each equation in (3.4a) the sphere of radius r centered at the origin. Hence the solutions to both equations lie on a circle and can be parametrised in polar coordinates by a radius r > 0 and angles ϕ+, ϕ− as w+α = r cos ( ϕ+ + α2π 3 ) , w−α = r cos ( ϕ− + α2π 3 ) . (3.5) Assume now that W+ and W− have different eigenvalues. Then we have T± = 0. Indeed, if tα 6= 0 for some α, then (3.4b) shows that w+α = w−α for this index α. But then (3.5) implies that W+ and W− have the same eigenvalues, which contradicts our assumption. Hence T± = 0. Without loss of generality we will set s = 0 in what follows, because integrability is indepen- dent of the scalar curvature s. This follows from the fact that the Killing tensor corresponding to the scalar component in the Ricci decomposition is proportional to the metric and thus to the identity when regarded as an endomorphism. But adding a multiple of the identity only alters the eigenvalues, not the eigendirections, and therefore has no effect on integrability. After these considerations, the independent components (2.11) of our algebraic curvature tensor are R0α0α = w+α + w−α 2 = Rβγβγ , R0αβγ = w+α − w−α 2 . (3.6) Choosing a1 = b1 = c1 = d1 = 0 in the second integrability condition (1.7b) for this algebraic curvature tensor yields gijgkl a2 b2 c2 d2 Ri0a2b2Rj0k0Rl0c2d2 = ∑ α=1,2,3 a2 b2 c2 d2 Rα0a2b2Rα0α0Rα0c2d2 = 0, since only terms with i = j = k = l 6= 0 contribute to the contraction. For (a2, b2, c2, d2) = (0, 1, 2, 3) this is gives∑ Rα00αRα0α0Rα0βγ = ∑ (R0α0α)2R0αβγ = 0, 18 K. Schöbel where the sums run over the three cyclic permutations (α, β, γ) of (1, 2, 3). We conclude that∑ α=1,2,3 (w+α + w−α)2(w+α − w−α) = 0. Substituting (3.5) into this equation yields, after some trigonometry, r3 cos2 ϕ+ − ϕ− 2 sin ϕ+ − ϕ− 2 sin 3 2 (ϕ+ + ϕ−) = 0. Substituting the solutions r = 0, ϕ+ = ϕ− + kπ, ϕ+ = −ϕ− + k 2π 3 , k ∈ Z of this equation back into (3.5) shows that the set {w−1, w−2, w−3} is either equal to {w+1, w+2, w+3} or to {−w+1,−w+2,−w+3}. The first case is excluded by assumption and we conclude that W+ and −W− have the same eigenvalues. Consequently we can choose an orthonormal basis of V such that w+α = −w−α =: wα. (3.7) Then the only remaining independent components of the algebraic curvature tensor (3.6) are R0αβγ = wα. (3.8) Now consider the second integrability condition (1.7b) for (a2, b2, c2, d2) = (0, 1, 2, 3), written as a1 b1 c1 d1 g ijgklEib1ld1Rja1kc1 = 0 (3.9) with Eib1ld1 := 1 4 a2 b2 c2 d2 Rib1a2b2Rld1c2d2 . (3.10) The product Rib1a2b2Rld1c2d2 is zero unless {i, b1, a2, b2} = {l, d1, c2, d2} = {0, 1, 2, 3}. Since {a2, b2, c2, d2} = {0, 1, 2, 3}, we have Eib1ld1 = 0 unless {i, b1, l, d1} = {0, 1, 2, 3}. The symmetries Eb1ild1 = −Eib1ld1 = Eib1d1l, Eib1ld1 = Eld1ib1 (3.11) then imply that the only independent components of (3.10) are E0αβγ = w2 α. (3.12) Now consider the full contraction of (3.9): ga1b1gc1d1 a1 b1 c1 d1 E i k b1 d1Ria1kc1 = 0. Since the Ricci tensor of (3.8) is zero, this is results in EijklRijkl + EijklRilkj = 0. Using the symmetries (3.11) of Eijkl and those of Rijkl, we can always permute a particular index to the first position: E0jklR0jkl + E0jklR0lkj = 0. The Variety of Integrable Killing Tensors on the 3-Sphere 19 And since both Eijkl and Rijkl vanish unless {i, j, k, l} = {0, 1, 2, 3}, we get∑( E0αβγR0αβγ + E0γβαR0γβα ) + ∑( E0αβγR0γβα + E0γβαR0αβγ ) = 0. Here again, the sums run over the three cyclic permutations (α, β, γ) of (1, 2, 3). Substitu- ting (3.8) and (3.12) yields∑( w3 α + w3 γ ) − ∑( w2 αwγ + w2 γwα ) = 0. With (2.13b) the second term can be transformed to∑( w2 αwγ + w2 γwα ) = ∑( w2 βwα + w2 βwγ ) = ∑ w2 β(wα + wγ) = − ∑ w3 β, implying w3 1 + w3 2 + w3 3 = 0. By (2.13b) and Newton’s identity w3 1 + w3 2 + w3 3 = 3w1w2w3 − 3(w1w2 + w2w3 + w3w1)(w1 + w2 + w3) + (w1 + w2 + w3) 3 this is equivalent to w1w2w3 = 0. Using (2.13b) once more, this shows that {w1, w2, w3} = {−w, 0,+w} for some w ∈ R. But then (3.7) implies that {w+1, w+2, w+3} = {w−1, w−2, w−3}, which contradicts our assumption that the eigenvalues of W+ and W− are different. Consequently W+ and W− have the same eigenvalues. This finishes the proof of Theorem 1.13: Any integrable Killing tensor on S3 has a diagonalisable algebraic curvature tensor. 3.3 Solution of the second integrability condition It remains to prove Theorem 1.14. In the preceding subsections we showed that an algebraic curvature tensor which satisfies the algebraic integrability conditions is diagonal in a suitable orthogonal basis of V and that any diagonal algebraic curvature tensor satisfies the first of the two algebraic integrability conditions. We therefore now solve the second integrability condi- tion (1.7b) for a diagonal algebraic curvature tensor in the form (2.13). To this aim consider the tensor gijgklRib1a2b2Rja1kc1Rld1c2d2 (3.13) appearing on the left hand side of (1.7b). Suppose it does not vanish. Then we have b1 ∈ {a2, b2} and d1 ∈ {c2, d2}, so without loss of generality we can assume b1 = b2 and d1 = d2. Then only terms with i = j = a2 and k = l = c2 contribute to the contraction and hence {a1, a2} = {c1, c2}. If now a2 = c2, then (3.13) vanishes under complete antisymmetrisation in a2, b2, c2, d2. This means that the left hand side of (1.7b) vanishes unless {a1, b1, c1, d1} = {a2, b2, c2, d2}. Note that a2, b2, c2, d2 and therefore a1, b1, c1, d1 have to be pairwise different due to the antisymmetrisation. For dimV = 4 in particular, (1.7b) reduces to a sole condition, which can be written as det 1 R0101 +R2323 R0101R2323 1 R0202 +R3131 R0202R3131 1 R0303 +R1212 R0303R1212  = 0. 20 K. Schöbel Substituting (2.13) yields det 1 w1 w2 1 − t21 1 w2 w2 2 − t22 1 w3 w2 3 − t23  = 0. (3.14) The form of this equation allows us to give an isometry invariant reformulation of the algebraic integrability conditions, which does not rely on diagonalising the algebraic curvature tensor. This also entails the isometry invariants in Proposition 1.15. Proposition 3.4. The algebraic integrability conditions for an algebraic curvature tensor with Weyl part W and (not necessarily trace free) Ricci part T are equivalent to [W,T] = 0, (3.15a) tr ( ∗W2 ) = tr ( ∗W3 ) = 0 (3.15b) and the linear dependence of W, W2 −T2 and the identity matrix. Proof. This is a consequence of Theorem 1.13 and Proposition 2.6, together with Corollary 3.3 and the second integrability condition in the form (3.14). Indeed, (3.15a) is the definition of alignedness for R and tr(∗Wk) = 0 is equivalent to trW k + = trW k −. For k = 1 this equation is the Bianchi identity in the form (2.2d). Hence (3.15b) means that W+ and W− have the same characteristic polynomial. The linear dependence of W, W2 − T2 and the identity matrix is equivalent to (3.14). � For t1 = t2 = t3 = 0 the left hand side of (3.14) is the Vandermonde determinant. Therefore (w1 − w2)t 2 3 + (w2 − w3)t 2 1 + (w3 − w1)t 2 2 + (w1 − w2)(w2 − w3)(w3 − w1) = 0. (3.16) This proves the first part of Theorem 1.14, for this equation can be written in the form det w2 − w3 −t3 t2 t3 w3 − w1 −t1 −t2 t1 w1 − w2  = 0. The second part follows from the residual action of the isometry on diagonal algebraic curvature tensors as described in Section 2.5. Regarding the eigenvalues w1, w2, w3 ∈ R as parameters subject to the restriction w1 + w2 + w3 = 0, (3.17) we can solve this equation for t1, t2, t3 ∈ R, depending on the number of equal eigenvalues: 1. If w1, w2, w3 are pairwise different with wα < wβ < wγ then (3.16) describes a one sheeted hyperboloid( tα aα )2 − ( tβ aβ )2 + ( tγ aγ )2 = 1 with semi axes aα = √ (wβ − wα)(wγ − wα), aβ = √ (wβ − wα)(wγ − wβ), aγ = √ (wγ − wα)(wγ − wβ). The Variety of Integrable Killing Tensors on the 3-Sphere 21 2. If wα = wβ 6= wγ , then (3.17) implies that the solutions of (3.16) are those (t1, t2, t3) ∈ R3 with tβ = ±tα. Since (tα,−tβ,−tγ) and (tα, tβ, tγ) are equivalent under the isometry group, we can assume tβ = tα without loss of generality. 3. If w1 = w2 = w3, then (3.17) implies that w1 = w2 = w3 = 0. In this case (3.16) is satisfied for any choice of t1, t2, t3 ∈ R. In other words, an algebraic curvature tensor with zero Weyl component is always integrable. 4 The algebraic geometry of the KS-variety We have seen that under isometries any integrable Killing tensor is equivalent to one whose algebraic curvature tensor is diagonal and that the space of such tensors is described by the following variety. Definition 4.1. Let A ∼= R5 be the vector space of matrices whose symmetric part is diagonal and trace free, i.e. of matrices of the form M = ∆1 −t3 t2 t3 ∆2 −t1 −t2 t1 ∆3  (4.1) with trM = 0. We denote by V ⊂ A the algebraic variety defined by the equation detM = 0 and call its projectivisation PV ⊂ PA ∼= P4 the Killing–Stäckel variety (KS-variety). A matrix in V will be called a Killing–Stäckel matrix (KS-matrix). Remark 4.2. The residual isometry group action on diagonal algebraic curvature tensors defines a natural action of the permutation group S4 on the Killing–Stäckel variety. This action is given by conjugation with matrices in SO(3) under the embedding S4 ⊂ SO(3) defined in Section 2.5. Later it will be useful to consider S4 ⊂ SO(3) as the symmetry group of an octahedron in R3 with vertices ±ei and adjacent faces opposedly oriented. The Killing–Stäckel variety PV is a 3-dimensional projective subvariety in P4 and the quo- tient PV /S4 encodes all information on integrable Killing tensors modulo isometries, the metric and scalar multiples. In this section we will investigate its structure from a purely algebraic geometric point of view. This will finally lead to a complete and explicit algebraic description of Stäckel systems. Recall that Stäckel systems are n-dimensional vector spaces of mutually commuting inte- grable Killing tensors. Note that every Stäckel system contains the metric and that the metric corresponds to the zero KS-matrix. Therefore, if we assume that the algebraic curvature tensors of the Killing tensors in a Stäckel system are mutually diagonalisable, then Stäckel systems on S3 correspond to 2-dimensional planes in V or projective lines in the projectivisation PV. We will see a posteriori that our assumption is justified. Hence Stäckel systems constitute a subvariety of the variety of projective lines on PV, also called the Fano variety of PV and denoted by F1(PV). LetD3 be the generic determinantal variety. This is the variety of 3×3 matrices with vanishing determinant or, equivalently, matrices of rank one or two. By definition PV is a subvariety of PD3. Hence its Fano variety F1(PV) is a subvariety of the Fano variety F1(PD3). But the latter is well understood. It contains different types of projective linear spaces: 22 K. Schöbel 1) projective spaces of matrices whose kernel contains a fixed line, 2) projective spaces of matrices whose image is contained in a fixed plane and 3) the projective plane of antisymmetric matrices. This motivates why we seek for such spaces in PV. Note that the projective plane of antisym- metric matrices is obviously contained in PV. Definition 4.3. We call a projective subspace in the KS-variety consisting of matrices whose kernel contains a fixed line (respectively whose image is contained in a fixed plane) an isokernel space (respectively an isoimage space). A 3×3 matrix M of rank two has a 2-dimensional image and a 1-dimensional kernel. Both are given by its adjugate matrix AdjM , which is the transpose of the cofactor matrix and satisfies (AdjM)M = M(AdjM) = (detM)I. (4.2) The adjugate of a matrix M ∈ V is AdjM = t1t1 + ∆2∆3 t2t1 + ∆3t3 t3t1 −∆2t2 t1t2 −∆3t3 t2t2 + ∆3∆1 t3t2 + ∆1t1 t1t3 + ∆2t2 t2t3 −∆1t1 t3t3 + ∆1∆2  . (4.3) The adjugate matrix of a rank two 3 × 3 matrix has rank one, which means that the columns of AdjM span a line in R3. And because detM = 0, we deduce from (4.2) that the kernel of M is the column space of AdjM . The adjugate matrix of a rank one 3 × 3 matrix is zero. It is known, that the singular lo- cus of PD3 is the subvariety of rank one matrices. The following proposition characterises the singular locus of the KS-variety. For simplicity we will be a bit imprecise and not strictly dis- tinguish between a matrix M ∈ V \{0} and its image RM ∈ PV under the canonical projection V \{0} → PV. Depending on the context, we will refer to an element of PV as “matrix” or “point”. Proposition 4.4 (singularities of the KS-variety). The KS-variety PV has ten singular points, given by the six rank one matrices V±1 := 0 0 0 0 +1 ∓1 0 ±1 −1  , V±2 := −1 0 ±1 0 0 0 ∓1 0 +1  , V±3 := +1 ∓1 0 ±1 −1 0 0 0 0  (4.4a) and the four skew symmetric matrices C0 := V+1 + V+2 + V+3, Cα := V−α + V+β + V+γ , (4.4b) where (α, β, γ) denotes an even permutation of (1, 2, 3). Moreover, the rank one matrices in PV are exactly those six in (4.4a). Definition 4.5. We will call the six singularities V±α the rank one singular points in PV and the four singularities Ci the skew symmetric singular points in PV. Proof. PV is singular at RM if and only if V is singular at M . Since V is the zero locus of the determinant function det : A → R, this is the case if and only if the derivative of the determinant function at M ∈ V, given by (ddet)M : A −→ R, A 7→ tr(AAdjM), (4.5) The Variety of Integrable Killing Tensors on the 3-Sphere 23 is the zero map. The condition that tr(AAdjM) = 0 for all A ∈ A is equivalent to t21 + ∆2∆3 = t22 + ∆3∆1 = t23 + ∆1∆2, ∆1t1 = ∆2t2 = ∆3t3 = 0. We leave it to the reader to verify that the solutions M ∈ V of these equations are exactly the matrices (4.4) and their multiples. The last statement now follows from (4.5) and the equivalence of rankM = 1 and AdjM = 0 for 3× 3-matrices. � Remark 4.6. Under the natural S4-action on PV, all rank one singular points are equivalent. This means that the singularities V±α are all mapped to a single point V in the quotient PV /S4. The same holds for the skew symmetric singular points, which are mapped to a single point C ∈ PV /S4. We now compute the isokernel spaces in PV. To find all matrices M ∈ V annihilating a given vector ~n ∈ R3, we consider the equation M~n = 0 as an equation in M for a fixed ~n = (n1, n2, n3), where M is of the form M = ∆ + T with ∆ = ∆1 0 0 0 ∆2 0 0 0 ∆3  , T =  0 −t3 t2 t3 0 −t1 −t2 t1 0  . We can regard ~t = (t1, t2, t3) ∈ R3 as a parameter and solve ∆~n = −T~n for ∆. For the time being, let us assume n1n2n3 6= 0. We then obtain a linear family of matrices M = n2 n1 t3 − n3 n1 t2 −t3 t2 t3 n3 n2 t1 − n1 n2 t3 −t1 −t2 t1 n1 n3 t2 − n2 n3 t1  . (4.6a) All these matrices satisfy detM = 0 and the condition trM = 0 imposes the restriction ~t ⊥ ( n3 n2 − n2 n3 , n1 n3 − n3 n1 , n2 n1 − n1 n2 ) (4.6b) on the parameter ~t ∈ R3. This shows that the matrices in V annihilating a fixed vector ~n 6= 0 form a linear subspace. If the right hand side of (4.6b) is zero, i.e. if the condition is void, this subspace has dimension three and defines a projective plane in PV. This happens if and only if \|n1\| = \|n2\| = \|n3\|. (4.7) If the right hand side of (4.6b) is not zero, this subspace has dimension two and defines a projective line in PV. One can check that this also holds true for the case n1n2n3 = 0. We have shown the following: Lemma 4.7. If ~n ∈ R3 satisfies (4.7), then the set of matrices in PV annihilating ~n is a pro- jective plane of the form (4.6a). Otherwise it is a projective line. Recall that the kernel of a rank two matrix of the form (4.1) is the row space of the adjugate matrix (4.3). This defines a rational map π : PV 99K P2, whose fibres are the maximal isokernel spaces. This map is well defined except for the six rank one matrices in PV. 24 K. Schöbel We want to give a parametrisation of the isokernel lines that is uniform and more geometric than (4.6). For this purpose we define two embeddings ι, ν : P2 → PV given by ι(n) =  0 −n3 n2 n3 0 −n1 −n2 n1 0  , ν(n) = n22 − n23 −n1n2 n3n1 n1n2 n23 − n21 −n2n3 −n3n1 n2n3 n21 − n22  (4.8) for n = (n1 : n2 : n3) ∈ P2. One readily checks that ι(n) and ν(n) indeed lie in PV. Note that the image of ι is the projective plane of skew symmetric matrices in PV. Under the inclusion PV ⊂ PA ∼= P4 the map ι is a linear embedding of P2 in P4 whilst ν is the composition of the Veronese embedding P2 ↪→ P5, (n1 : n2 : n3) 7→ ( n21 : n22 : n23 : n2n3 : n3n1 : n1n2 ) , and a projection P5 � P4 under which the Veronese surface remains smooth. Proposition 4.8 (isokernel spaces in the KS-variety). Recall from Proposition 4.4 that the singular points in the KS-variety PV are the four skew symmetric singular points C0, C1, C2, C3 as well as six rank one singular points V±α, α = 1, 2, 3. 1. The projective plane through V+1, V+2 and V+3 is an isokernel plane in PV and con- tains C0. In the same way, the three points V−α, V+β and V+γ define an isokernel plane through Cα for α = 1, 2, 3. Let the maps π : PV 99K P2 and ι, ν : P2 → PV be defined as above. 2. For any non-singular point M ∈ PV, the points M , ι(π(M)) and ν(π(M)) are collinear but do not all three coincide. The projective line they define lies in PV and is an isokernel space through M . Each of these lines contains a unique skew symmetric point, namely ι(π(M)). Moreover, any isokernel space in PV of maximal dimension is of either of the above forms and hence contains a unique skew symmetric point. Remark 4.9. A corresponding characterisation of isoimage spaces follows from the fact that matrix transposition defines an involution of PV which interchanges isokernel and isoimage spaces. Proof. 1 follows directly from Lemma 4.7 and the explicit form (4.4) of the singular points. For 2 notice that n := π(M) is the kernel ofM and well defined forM 6= Vα. The definitions (4.8) of ι and ν then show that n is also the kernel of ι(n) and ν(n). These definitions also show that M , ι(n) and ν(n) do not all coincide for M 6= Ci. Their collinearity follows from Lemma 4.7 in the case where n does not satisfy (4.7) and from ι(n) = ν(n) in case it does. The last statement is now a consequence of Lemma 4.7. � We can illustrate the content of Proposition 4.8 geometrically as follows. This is depicted in Fig. 1. The six rank one singular points of PV ⊂ P4 constitute the vertices of a regular octahedron in P4 whose faces (and the planes they define) are entirely contained in PV. The set of eight faces is divided into two sets of non-adjacent faces, corresponding to the four isokernel planes (shaded) respectively the four isoimage planes in PV. Opposite faces intersect in their respective centres. These are the skew symmetric singular points of PV and the intersection points of the octahedron with the projective plane of skew symmetric matrices. In Fig. 1 this ninth projective plane in PV is depicted as the insphere of the octahedron. The Variety of Integrable Killing Tensors on the 3-Sphere 25 Figure 1. Singularities and projective planes in the KS-variety. The permutation group S4 acts on this configuration by symmetries of the octahedron pre- serving the two sets of non-adjacent faces. Matrix transposition corresponds to a point reflection, which exchanges opposite faces and completes the S4-action to the full octahedral symmetry. In anticipation of the results in the next section we introduce the following name: Definition 4.10. We call the isokernel lines of the form 2 in Proposition 4.8 “Stäckel lines”. Every Stäckel line intersects the projective plane of skew symmetric matrices transversely. Conversely, every point on this plane determines a unique Stäckel line unless it is one of the four singular skew symmetric points. The Stäckel lines through these points are exactly those projective lines in each isokernel face which pass through the respective face center. 5 Stäckel systems In this section we express the condition that two Killing tensors commute as an algebraic con- dition on the corresponding algebraic curvature tensors. This will eventually justify the term “Stäckel lines” chosen in Definition 4.10 by proving that these lines indeed correspond to Stäckel systems. Remark 5.1. Most results in this and the following section are valid for an arbitrary non-flat constant curvature (pseudo-)Riemannian manifold M that is embedded into some vector space V as the hypersurface M = {v ∈ V : g(v, v) = 1} ⊂ V of length one vectors with respect to a non-degenerated symmetric bilinear form g on V . Note that any non-flat constant curvature (psudo-)Riemannian manifold is locally isometric to such a model. 26 K. Schöbel 5.1 Young tableaux Throughout this section we use Young tableaux in order to characterise index symmetries of tensors. Young tableaux define elements in the group algebra of the permutation group Sd. That is, a Young tableau stands for a (formal) linear combination of permutations of d objects. In our case, these objects will be certain tensor indices. For the sake of simplicity of notation we will identify a Young tableau with the group algebra element it defines. It will be sufficient for our purposes to consider the examples below. For more examples and a detailed explanation of the notation as well as the techniques used here, we refer the reader to [41]. A Young tableau consisting of a single row denotes the sum of all permutations of the indices in this row. For example, using cycle notation, a2 c1 c2 = e+ (a2c1) + (c1c2) + (c2a2) + (a2c1c2) + (c2c1a2). This is an element in the group algebra of the group of permutations of the indices a2, c1 and c2 (or any superset). In the same way a Young tableau consisting of a single column denotes the signed sum of all permutations of the indices in this column, the sign being the sign of the permutation. For example, a1 b1 d2 = e− (a1b1)− (b1d2)− (d2a1) + (a1b1d2) + (d2b1a1). We call these row symmetrisers respectively column antisymmetrisers. The reason we define them without the usual normalisation factors is that then all numerical constants appear expli- citly in our computations (although irrelevant for our concerns). The group multiplication extends linearly to a natural product in the group algebra. A general Young tableau is then simply the product of all column antisymmetrisers and all row symmetri- sers of the tableau. We will only deal with Young tableaux having a “hook shape”, such as the following: a1 a2 c1 c2 b1 d2 = a1 a2 c1 c2 a1 b1 d2 . (5.1) The inversion of group elements extends linearly to an involution of the group algebra. If we consider elements in the group algebra as linear operators on the group algebra itself, this involution is the adjoint with respect to the natural inner product on the group algebra, given by defining the group elements to be an orthonormal basis. Since this operation leaves sym- metrisers and antisymmetrisers invariant, it simply exchanges the order of symmetrisers and antisymmetrisers in a Young tableau. The adjoint of (5.1) for example is a1 a2 c1 c2 b1 d2 ? = a1 b1 d2 a1 a2 c1 c2 . (5.2) Properly scaled, Young tableaux define projectors onto irreducible Sd-representations. We are now able to state a lemma which expresses the decomposition ⊗ ∼= ⊕ , given by the Littlewood–Richardson rule, in terms of Young projectors. Its proof can be found in [41]. The Variety of Integrable Killing Tensors on the 3-Sphere 27 Lemma 5.2. In the group algebra of the permutation group of the indices a1, a2, c1, c2, b1 and d2 the following identity holds: 1 2! b1 d2 · 1 4! a1 a2 c1 c2 = 1 10368 a1 a2 c1 c2 b1 d2 a1 a2 c1 c2 b1 d2 ? + 1 34560 b1 a1 a2 c1 c2 d2 ? b1 a1 a2 c1 c2 d2 . (5.3) The action of the permutation group on tensors extends linearly to an action of the group algebra. In particular, any Young tableau acts on tensors with corresponding indices. For example, b1 a2 c2 Tb1a2c2 = Tb1a2c2 − Ta2b1c2 − Tb1c2a2 − Tc1a2b2 + Ta1c2b2 + Tc1b2a2 . To give another example, the operator (5.1) acts on a tensor Tb1b2d1d2a2c2 by an antisymmetri- sation in the indices b1, a2, c2 and a subsequent symmetrisation in the indices b1, b2, d1, d2. In the same way its adjoint (5.2) acts by first symmetrising and then antisymmetrising. 5.2 Commuting Killing tensors In the following we express the fact that two Killing tensors commute (as endomorphisms) as a purely algebraic condition on their corresponding algebraic curvature tensors. Notice that the commuting of Killing tensors as endomorphisms on the tangent space is not equivalent to the commuting of their respective algebraic curvature tensors as endomorphisms on Λ2V , as one may guess näıvely. Proposition 5.3 (algebraic counterpart of the commutator). Let K and K̃ be two Killing tensors on a non-flat constant curvature manifold with algebraic curvature tensors R and R̃. Then the following statements are equivalent [K, K̃] = 0, (5.4a) b1 b2 d1 d2 a2 c2 ? gijR i b1a2b2R̃ j d1c2d2 = 0, (5.4b) a2 c2 b1 b2 d1 d2 gijR i b1a2b2R̃ j d1c2d2 = 0. (5.4c) Proof. Using ∇vxb = vb we write (1.6) in local coordinates for coordinate vectors v = ∂α and w = ∂β: Kαβ = Ra1b1a2b2x a1xa2∇αxb1∇βxb2 . The product of K and K̃, regarded as endomorphisms, is then given by Kα γ K̃ γ β = Ra1b1a2b2R̃c1d1c2d2x a1xa2xc1xc2∇αxb1∇γxb2∇γxd1∇βxd2 . In [41] we proved the following identity: ∇γxb2∇γxd1 = gb2d1 − xb2xd1 . As a consequence of the antisymmetry of algebraic curvature tensors in the last index pair, the term xb2xd1 does not contribute when substituting this identity into the previous expression: Kα γ K̃ γ β = gb2d1Ra1b1a2b2R̃c1d1c2d2x a1xa2xc1xc2∇αxb1∇βxd2 . 28 K. Schöbel The commutator [K, K̃] is therefore given by [K, K̃]αβ = gb2d1Ra1b1a2b2R̃c1d1c2d2x a1xa2xc1xc2∇[αx b1∇β]xd2 and vanishes if and only if gb2d1Ra1b1a2b2R̃c1d1c2d2x a1xa2xc1xc2v[b1wd2] = 0 (5.5) for all x ∈M and v, w ∈ TxM . That is, for all x, v, w ∈ V with g(x, x) = 1, g(x, v) = g(x,w) = 0. (5.6) We can drop the restriction g(x, x) = 1 since RM ⊆ V is open. We can also drop the restrictions g(x, v) = 0 and g(x,w) = 0 by decomposing arbitrary vectors v, w ∈ V under the decomposition V = TxM ⊕ Rx. To see this, notice that (5.5) is trivially satisfied for v = x or for w = x. Indeed, in this case the tensor Ra1b1a2b2R̃c1d1c2d2 (5.7) is implicitly symmetrised over five indices and Dirichlet’s drawer principle tells us that this comprises a symmetrisation in one of the four antisymmetric index pairs. This means we can omit the restrictions (5.6) completely. In other words, [K, K̃] = 0 is equivalent to (5.5) being satisfied for all x, v, w ∈ V . Now notice that the tensor xa1xa2xc1xc2v[b1wd2] appearing in (5.5) is completely symmetric in the indices a1, a2, c1 and c2 and completely antisymmetric in the indices b1 and d2. Applying (5.3) to this tensor therefore yields xa1xa2xc1xc2v[b1wd2] = 1 2! b1 d2 · 1 4! a1 a2 c1 c2 x a1xa2xc1xc2v[b1wd2] = 1 10368 · a1 a2 c1 c2 b1 d2 a1 a2 c1 c2 b1 d2 ? xa1xa2xc1xc2v[b1wd2] + 1 34560 · b1 a1 a2 c1 c2 d2 ? b1 a1 a2 c1 c2 d2 xa1xa2xc1xc2v[b1wd2]. (5.8) From the definition of the Young tableaux and their adjoints, a1 a2 c1 c2 b1 d2 = a1 a2 c1 c2 a1 b1 d2 , a1 a2 c1 c2 b1 d2 ? = a1 b1 d2 a1 a2 c1 c2 , b1 a1 a2 c1 c2 d2 ? = b1 d2 b1 a1 a2 c1 c2 , b1 a1 a2 c1 c2 d2 = b1 a1 a2 c1 c2 b1 d2 , together with the properties a1 a2 c1 c2 x a1xa2xc1xc2 = 4!xa1xa2xc1xc2 , b1 d2 v[b1wd2] = 2 b1 d2 vb1wd2 and a1 b1 d2 2 = 3! a1 b1 d2 , b1 a1 a2 c1 c2 2 = 5! b1 a1 a2 c1 c2 The Variety of Integrable Killing Tensors on the 3-Sphere 29 we see that (5.8) simplifies to xa1xa2xc1xc2v[b1wd2] = ( 4! · 3! 10368 a1 a2 c1 c2 b1 d2 + 2 · 5! 34560 b1 a1 a2 c1 c2 d2 ? ) xa1xa2xc1xc2vb1wd2 . When substituted into (5.5), this yields gb2d1Ra1b1a2b2R̃c1d1c2d2 ( 1 72 a1 a2 c1 c2 b1 d2 + 1 144 b1 a1 a2 c1 c2 d2 ? ) xa1xa2xc1xc2vb1wd2 = 0. (5.9) As elements of the group algebra of the permutation group, the Young tableaux are linear combinations of permutations π of the indices a1, a2, c1, c2, b1, d2. For any such permutation π we have gb2d1Ra1b1a2b2R̃c1d1c2d2x π(a1)xπ(a2)xπ(c1)xπ(c2)vπ(b1)wπ(d2) = gb2d1Rπ?(a1)π?(b1)π?(a2)b2R̃π?(c1)d1π?(c2)π?(d2)x a1xa2xc1xc2vb1wd2 , where π? = π−1 is the adjoint of π. We can thus replace the Young tableaux in (5.9) acting on upper indices by its adjoint acting on lower indices:(( 1 72 a1 a2 c1 c2 b1 d2 ? + 1 144 b1 a1 a2 c1 c2 d2 ) gb2d1Ra1b1a2b2R̃c1d1c2d2 ) xa1xa2xc1xc2vb1wd2 = 0.(5.10) Now notice that the second Young tableau involves a symmetrisation over the five indices b1, a1, a2, c1, c2 and that, as above, the symmetrisation of the tensor (5.7) in any five indices is zero. Hence the second term in (5.10) vanishes and we obtain( a1 a2 c1 c2 b1 d2 ? gb2d1Ra1b1a2b2R̃c1d1c2d2 ) xa1xa2xc1xc2vb1wd2 = 0. Recall that [K, K̃] = 0 is equivalent to this condition being satisfied for all x, v, w ∈ V . By polarising in x we get( a1 a2 c1 c2 b1 d2 ? gb2d1Ra1b1a2b2R̃c1d1c2d2 ) xa1ya2zc1tc2vb1wd2 = 0 for all x, y, z, t, v, w ∈ V and hence a1 a2 c1 c2 b1 d2 ? gb2d1Ra1b1a2b2R̃c1d1c2d2 = 0. This is the same as (5.4b) after appropriately renaming, lowering and rising indices. We have proven the equivalence (5.4a) ⇔ (5.4b). We now prove the equivalence (5.4b) ⇔ (5.4c). Start from (5.4b) by expanding the Young tableau: b1 a2 c2 b1 b2 d1 d2 gijR i b1a2b2R̃ j d1c2d2 = 0. In order to sum over all 4! permutations when carrying out the symmetrisation in the indices b1, b2, d1, d2, one can first take the sum over the 4 cyclic permutations of b1, b2, d1, d2, then fix the index b1 and finally sum over all 3! permutations of the remaining 3 indices b2, d1, d2: b1 a2 c2 b2 d1 d2 gij ( Rib1a2b2R̃ j d1c2d2 +Rib2a2d1R̃ j d2c2b1 +Rid1a2d2R̃ j b1c2b2 +Rid2a2b1R̃ j b2c2d1 ) = 0. 30 K. Schöbel For a better readability we underlined each antisymmetrised index. Permuting the indices of the terms in the parenthesis under symmetrisation in b2, d1, d2 and antisymmetrisation in b1, a2, c2 we can gather the first and last as well as the second and third term: b1 a2 c2 b2 d1 d2 gij ( (Rib1a2b2 +Rib2a2b1)R̃jd1c2d2 +Rid1a2d2(R̃jb1c2b2 + R̃jb2c2b1 ) ) = 0. Using the symmetries of Rb1a2b2 , the terms in the inner parentheses can be rewritten as Rib1a2b2 = −Rib1b2a2 , Rib2a2b1 = −Ria2b1b2 −R i b1b2a2 = Ria2b2b1 −R i b1b2a2 resulting in b1 a2 c2 b2 d1 d2 gij ( Ria2b2b1R̃ j d1c2d2 +Rid1a2d2R̃ j c2b2b1 ) = 0. As above, when carrying out the antisymmetrisation over b1, a2, c2, we can first sum over the three cyclic permutations of b1, a2, c2, then fix b1 and finally sum over the two permutations of a2, c2. This results in a2 c2 b2 d1 d2 gij ( Ria2b2b1R̃ j d1c2d2 +Rid1a2d2R̃ j c2b2b1 +Ric2b2a2R̃ j d1b1d2 +Rid1c2d2R̃ j b1b2a2 +Rib1b2c2R̃ j d1a2d2 +Rid1b1d2R̃ j a2b2c2 ) = 0. The symmetrisation of this in b1, b2, d1, d2 yields a2 c2 b1 b2 d1 d2 gij ( Ria2b2b1R̃ j d1c2d2 +Rid1a2d2R̃ j c2b2b1 +Ric2b2a2R̃ j d1b1d2 +Rid1c2d2R̃ j b1b2a2 +Rib1b2c2R̃ j d1a2d2 +Rid1b1d2R̃ j a2b2c2 ) = 0, where we again underlined antisymmetrised indices. By the antisymmetry of algebraic curvature tensors in the second index pair, all but the fourth and fifth term in the parenthesis vanish and we get a2 c2 b1 b2 d1 d2 gij ( Rid1c2d2R̃ j b1a2b2 +Rib1c2b2R̃ j d1a2d2 ) = 0. Due to the symmetrisation and antisymmetrisation we can permute the indices of the terms inside the parenthesis to find that a2 c2 b1 b2 d1 d2 gijR i b1a2b2 R̃jd1c2d2 = 0. Recall that this equation has been obtained from (5.4b) by a symmetrisation. This proves (5.4b) ⇒ (5.4c). The converse follows easily by antisymmetrising (5.4c) in b1, a2, c2. This achieves the proof of the theorem. � Remark 5.4. Obviously (5.4b) must be true for R̃ = R. This fact is not evident, but has been proven in [41]. For diagonal algebraic curvature tensors R and R̃ on a four-dimensional vector space the condition (5.4c) can be written in the form det 1 Rijij R̃ijij 1 Rjkjk R̃jkjk 1 Rkiki R̃kiki  = 0 (5.11) for all distinct i, j, k ∈ {0, 1, 2, 3}. The Variety of Integrable Killing Tensors on the 3-Sphere 31 Corollary 5.5. Stäckel lines5 in the KS-variety correspond to Stäckel systems. More precisely, the preimage of a Stäckel line under the map from integrable Killing tensors with diagonal algebraic curvature tensor to the KS-variety is a Stäckel system. In particular, the Killing tensors in a Stäckel system have simultaneously diagonalisable algebraic curvature tensors. Proof. A Stäckel line is the projective line through the (collinear but not coinciding) points M , ι(π(M)) and ν(π(M)) in the KS-variety for any non-singular KS-matrix M . We have to prove that corresponding Killing tensors K0, K1 and K2 mutually commute. We first show that K1 and K2 commute. Let R and R̃ be their diagonal algebraic curvature tensors. From their KS-matrices (4.8) and ∆α = wβ − wγ we can read off the values wα = 0, tα = nα, w̃α = n2α, t̃α = nβnγ , where we have neglected the scalar curvature, which can be chosen appropriately. The diagonals of R and R̃ are then given by the columns of the matrix 1 R0101 R̃0101 1 R0202 R̃0202 1 R0303 R̃0303 1 R2323 R̃2323 1 R3131 R̃3131 1 R1212 R̃1212  =  1 w1 + t1 w̃1 + t̃1 1 w2 + t2 w̃2 + t̃2 1 w3 + t3 w̃3 + t̃3 1 w1 − t1 w̃1 − t̃1 1 w2 − t2 w̃2 − t̃2 1 w3 − t3 w̃3 − t̃3  =  1 +n1 n21 + n2n3 1 +n2 n22 + n3n1 1 +n3 n23 + n1n2 1 −n1 n21 − n2n3 1 −n2 n22 − n3n1 1 −n3 n23 − n1n2  . The four square matrices in (5.11) can be obtained from this matrix by discarding the rows (1, 2, 3), (1, 5, 6), (2, 4, 6) respectively (3, 4, 5) and it is not difficult to see that their determinants vanish. This shows that K1 and K2 commute. If ι(π(M)) and ν(π(M)) are linearly independent, then M is a linear combination of both and hence K0 is a linear combination of g, K1 and K2. In other words, K0, K1 and K2 mutually commute. If ι(π(M)) and ν(π(M)) are linearly dependent, then n = π(M) satisfies \|n1\| = \|n2\| = \|n3\|. For simplicity, suppose n1 = n2 = n3 = 1. The other choices of the signs are analogous. By Lemma 4.7, M is of the form (4.6a) and thus has wα = −tα. Consequently, the diagonal algebraic curvature tensors R and R̃ of K0 respectively K2 are given by 1 R0101 R̃0101 1 R0202 R̃0202 1 R0303 R̃0303 1 R2323 R̃2323 1 R3131 R̃3131 1 R1212 R̃1212  =  1 w1 + t1 w̃1 + t̃1 1 w2 + t2 w̃2 + t̃2 1 w3 + t3 w̃3 + t̃3 1 w1 − t1 w̃1 − t̃1 1 w2 − t2 w̃2 − t̃2 1 w3 − t3 w̃3 − t̃3  =  1 0 2 1 0 2 1 0 2 1 2t1 0 1 2t2 0 1 2t3 0  . As above, they satisfy the commutation condition (5.11) and we conclude that K0, K1 and K2 commute. It remains to prove that every Stäckel system consists of Killing tensors with mutually di- agonalisable algebraic curvature tensors. This follows from the fact that every Stäckel system contains a Killing tensor with simple eigenvalues and that such a Killing tensor K uniquely determines the Stäckel system [3]. By the very definition of a Stäckel system, K is integrable and by Theorem 1.13 we can assume it to have a diagonal algebraic curvature tensor. The cor- responding point on the KS variety lies on a Stäckel line. By the above, K then lies in a Stäckel system with diagonal algebraic curvature tensors. This proves the statement, since the Stäckel system determined by K is unique. � 5Recall Definition 4.10. 32 K. Schöbel Remark 5.6. Our result also shows that every integrable Killing tensor on S3 is contained in a Stäckel system, not only those with simple eigenvalues. This follows from the above corollary in conjunction with Proposition 4.8. We can now prove Theorem 1.19. Recall from Proposition 4.8 that each Stäckel line contains a unique skew symmetric point ι(n), where n ∈ P2, and that each non-singular skew symmetric point ι(n) determines a unique Stäckel line, given by the two points ι(n) 6= ν(n). Moreover, the singular skew symmetric points are the four points n = (±1:±1:±1) ∈ P2 for which ι(n) = ν(n). Hence it suffices to show that the subspace spanned by ι(n(t)) and ν(n(t)) has a well defined limit for t → 0 if n(0) is one of these and that this limit depends on ṅ(0) ∈ P2. We leave it to the reader to verify that the limit space is spanned by ι(n(0)) and ι(ṅ(0)). 6 Geometric constructions of integrable Killing tensors In this section we present several geometric constructions of integrable Killing tensors and in- terpret each of them within our algebraic picture from the last section. 6.1 Special Killing tensors There are two ways in which integrable Killing tensors are related to geodesically equivalent metrics. The first is via special conformal Killing tensors. Definition 6.1. A special conformal Killing tensor on a Riemannian manifold is a symmetric tensor Lαβ satisfying the Sinjukov equation ∇γLαβ = λαgβγ + λβgαγ , (6.1a) where λ = 1 2∇ trL, (6.1b) as can be seen from contracting α and β. Special conformal Killing tensors parametrise geodesically equivalent metrics in the following way [44]. Theorem 6.2. A metric g̃ is geodesically equivalent to g if and only if the tensor L := ( det g̃ det g ) 1 n+1 g̃−1 (6.2) is a special conformal Killing tensor. From the definition we see that special conformal Killing tensors on a manifold M form a vector space which is invariant under the isometry group of M . In other words, they define a representation of this group. The following lemma shows that this representation is isomorphic to a subrepresentation of the representation of the isomorphism group on Killing tensors. Lemma 6.3. If L is a special conformal Killing tensor, then K := L − (trL)g is a Killing tensor. This defines an injective map from the space of special conformal Killing tensors to the space of Killing tensors, which is equivariant with respect to the action of the isometry group. The Variety of Integrable Killing Tensors on the 3-Sphere 33 Proof. It is straightforward to check the first statement. Taking the trace on both sides of K = L− (trL)g yields trK = (1− n) trL and hence L = K − trK n− 1 g. (6.3) This shows injectivity. Equivariance is obvious. � Definition 6.4. We define a special Killing tensor to be a Killing tensor of the form K = L− (trL)g, where L is a special conformal Killing tensor. Proposition 6.5. A special conformal Killing tensor as well as the corresponding special Killing tensor have vanishing Nijenhuis torsion. In particular, special Killing tensors are integrable. Proof. Substituting the Sinjukov equation (6.1) into the expression (1.3) for K = L shows that the Nijenhuis torsion of L is zero. Substituting K = L− (trL)g into (1.3) shows that this implies that the Nijenhuis torsion of K is also zero. � For a non-flat constant curvature manifold M ⊂ V we can identify the subrepresentation of special Killing tensors inside the space of Killing tensors. Recall that the space of Killing tensors is isomorphic to the space of algebraic curvature tensors on V and an irreducible GL(V )- representation. Under the subgroup SO(V ) ⊂ GL(V ) this space decomposes into a Weyl com- ponent and a Ricci component. Definition 6.6. The Kulkarni–Nomizu product of two symmetric tensors h and k on a vector space V is the algebraic curvature tensor h7 k, given by (h7 k)a1b1a2b2 := ha1a2kb1b2 − ha1b2kb1a2 − hb1a2ka1b2 + hb1b2ka1a2 . (6.4) The Ricci component consists of all algebraic curvature tensors of the form h 7 g, where h is a symmetric tensor. Therefore the dimension of the Ricci part is equal to N(N + 1)/2 where N = dimV . This is exactly the dimension of the space of special conformal Killing tensors on M [44]. But there is no other SO(V )-subrepresentation of the same dimension inside the space of algebraic curvature tensors. This shows the following. Lemma 6.7. A special Killing tensor on a non-flat constant curvature manifold is a Killing tensor whose algebraic curvature tensor R has a vanishing Weyl tensor. That is, R is of the form R = h7 g for some symmetric tensor h on V , where “ 7” denotes the Kulkarni–Nomizu product (6.4). Alternatively, this lemma can be checked directly by verifying that (6.3) satisfies (6.1) if K has an algebraic curvature tensor of the form R = h7 g. The following consequence is a special case of the so called “cone construction” [32]: Any special conformal Killing tensor on a Riemannian manifold M can be extended to a covariantly constant symmetric tensor on the metric cone over M . In our case, where M ⊂ V is a constant curvature manifold, this cone is nothing but the embedding space V . That is why a special conformal Killing tensor is the restriction of a constant symmetric tensor on the ambient space. 34 K. Schöbel Proposition 6.8. Let R = h 7 g be the algebraic curvature tensor of a special Killing tensor K = L − (trL)g on a non-flat constant curvature manifold M ⊂ V . Then the corresponding special conformal Killing tensor L = K − trK n− 1 g (6.5) is the restriction of the (constant) symmetric tensor L̂ = h− Trh n− 1 g (6.6) from V to M , where “ Tr” denotes the trace on V and “ tr” the trace on the tangent space. In particular, if h is tracefree, then L is simply the restriction of h from V to M . Proof. The special Killing tensor K is given by substituting R = h 7 g into (1.6) using (6.4). This yields Kx(v, w) = h(x, x)g(v, w)− h(x, v)g(x,w)− h(x,w)g(x, v) + h(v, w)g(x, x) for tangent vectors v, w ∈ TxM at a point x ∈ M , i.e. vectors x, v, w ∈ V with g(x, x) = 1 and g(x, v) = g(x,w) = 0. This simplifies to Kx(v, w) = h(v, w) + h(x, x)g(v, w). (6.7) Let e1, . . . , en be an orthonormal basis of TxM and complete it with e0 := x to an orthonormal basis of V . Then trK = n∑ α=1 Kx(eα, eα) = n∑ α=1 ( h(eα, eα) + h(e0, e0)g(eα, eα) ) = n∑ i=0 h(ei, ei) + (n− 1)h(e0, e0) = Trh+ (n− 1)h(x, x). (6.8) Recall that “tr” denotes the trace on TxM , whereas “Tr” denotes the trace on V . Substitu- ting (6.7) and (6.8) into the right hand side of (6.5) now yields (6.6). � For S3 we can verify the integrability of a special Killing tensor on S3 algebraically, using the Theorems 1.13 and 1.14: Lemma 6.9. 1. R = h7g is diagonalisable. More precisely, R is diagonal in a basis where h is diagonal. In this basis the diagonal elements of R are given in terms of the diagonal elements hi := hii by Rijij = hi + hj , i 6= j. 2. If dimV = 4, the KS-matrix (1.9) of R = h7 g is antisymmetric with ∆α = 0, tα = h0 + hα. In particular it has determinant zero. Corollary 6.10. The space of special Killing tensors on S3 corresponds to the projective space of antisymmetric matrices inside the KS-variety. Proposition 4.8 and Corollary 5.5 now give a unique representative in each Stäckel system. Corollary 6.11. Every Stäckel system on S3 contains a special Killing tensor which is unique up to multiplication with constants and addition of multiples of the metric. The Variety of Integrable Killing Tensors on the 3-Sphere 35 6.2 Killing tensors of Benenti type The second way in which integrable Killing tensors arise from geodesically equivalent metrics is described by the following theorem [2, 30, 33, 39]. Theorem 6.12. If a metric g̃ on a (pseudo-)Riemannian manifold is geodesically equivalent to g, then K := ( det g det g̃ ) 2 n+1 g̃ (6.9) is an integrable Killing tensor for g. Definition 6.13. We will call a Killing tensor of the form (6.9) a Benenti–Killing tensor. Recall that the standard metric on the unit sphere Sn ⊂ V in a Euclidean vector space (V, g) is the restriction of the scalar product g from V to Sn, which we denoted by g as well. Its geodesics are the great circles on Sn. Consider the map f : Sn → Sn, x 7→ f(x) := Ax ‖Ax‖ , for A ∈ GL(V ). Since A takes hyperplanes to hyperplanes, f takes great circles to great circles. Hence the pullback g̃ of g under f is geodesically equivalent to g by definition. The Killing tensor obtained from applying (6.9) to g̃ is given by [41] Kx(v, w) = g(Ax,Ax)g(Av,Aw)− g(Ax,Av)g(Ax,Aw) (detA) 4 n+1 , v, w ∈ TxM. Comparing with (1.6), we see that the algebraic curvature tensor of this Killing tensor is pro- portional to h7 h, where “7” denotes the Kulkarni–Nomizu product (6.4) and h(v, w) := g(Av,Aw) = g ( v,AtAw ) . Moreover, we can replace the scalar product h by any – not necessarily positive definite – symmetric tensor. This motivates the following definition. Definition 6.14. We will say a Killing tensor on a constant curvature manifold of Benenti type if its algebraic curvature tensor is proportional to R = h7 h for some symmetric tensor h on V . A Killing tensor of Benenti type with positive definite h is a Benenti–Killing tensor and thus integrable. With the aid of the algebraic integrability conditions one can verify that any Killing tensor of Benenti type is integrable [41]. On S3 we can check this directly, using the Theorems 1.13 and 1.14: Lemma 6.15. 1. For a symmetric tensor h on a vector space V , the algebraic curvature tensor R = h7h is diagonalisable. More precisely, R is diagonal in a basis where h is diagonal. In this basis the diagonal elements of R are given in terms of the diagonal elements hi := hii by Rijij = 2hihj , i 6= j. (6.10) 2. If dimV = 4, the KS-matrix (1.9) of R = h7 h is given by ∆α = (h0 − hα)(hβ − hγ), tα = h0hα − hβhγ and has zero determinant and trace. 36 K. Schöbel 6.3 Benenti systems Special Killing tensors and Benenti–Killing tensors only constitute particular cases of a more general construction of integrable Killing tensors out of special conformal Killing tensors. This construction appears in the literature in different guises and under a number of different names: as “Newtonian systems of quasi-Lagrangian type” [40], as “systems admitting special conformal Killing tensors” [8], as “cofactor systems” [31], as “bi-Hamiltonian structures” [6, 16], as “bi- quasi-Hamiltonian systems” [9, 10], as “L-systems” [4] or as “Benenti systems” [7]. Here we follow Bolsinov and Matveev, who call them “Benenti systems” [7], because their formulation is directly reflected in our algebraic description for S3. Recall that the adjugate matrix AdjM of a matrix M is the transpose of its cofactor matrix, which is equal to (detM)M−1 if M is invertible. Given a metric, we can identify endomor- phisms and bilinear forms. This allows us to extend this definition to bilinear forms. As for endomorphisms, we denote the adjugate of a bilinear form L by AdjL, which is again a bilinear form. Theorem 6.16 ([7]). For a special conformal Killing tensor L the family K(λ) = Adj(L− λg), λ ∈ R is a family of mutually commuting integrable Killing tensors. In particular, for λ = 0 we recover the Benenti–Killing tensor (6.9) from (6.2). Definition 6.17. The family K(λ) = Adj(L − λg) of mutually commuting Killing tensors is called the Benenti system of the special conformal Killing tensor L. In order to state the algebraic counterpart of the above theorem for constant curvature manifolds, we first need the following lemma. Lemma 6.18. Let h be a symmetric tensor on a vector space V of dimension n+ 1. Then the algebraic curvature tensor (Adjh) 7 (Adjh) deth (6.11) is well defined and homogeneous in h of degree n− 1. Proof. Chose a basis in which h is diagonal with diagonal elements hi := hii. Then Adjh is also diagonal and has diagonal elements h0 · · ·hi−1hi+1 · · ·hn. By (6.10) the algebraic curvature tensor (6.11) is diagonal and given by Rijij = 2 h0 · · ·hi−1hi+1 · · ·hn · h0 · · ·hj−1hj+1 · · ·hn h0 · · ·hn = 2h0 · · ·hi−1hi+1 · · ·hj−1hj+1 · · ·hn for i 6= j. This is well defined and homogeneous in h of degree n− 1. � Proposition 6.19 (algebraic representation of Benenti systems). Let L be a special confor- mal Killing tensor on a non-flat constant curvature manifold M ⊂ V , i.e. the restriction of a (constant) symmetric tensor L̂ from V to M . To avoid confusion, we denote the metric on V by ĝ and its restriction to M ⊂ V by g. Then the Killing tensors of the Benenti system K = Adj(L− λg), λ ∈ R (6.12) The Variety of Integrable Killing Tensors on the 3-Sphere 37 correspond to the algebraic curvature tensors R = Adj(L̂− λĝ) 7 Adj(L̂− λĝ) det(L̂− λĝ) . (6.13) In particular, a member of this Benenti system is of Benenti type unless λ is an eigenvalue of L̂ and it comes from a geodesically equivalent metric if −λ is sufficiently large. Proof. For simplicity of notation, we absorb the trace terms “λg” and “λĝ” into L respectively L̂ by formally setting λ to 0. At a fixed point x ∈M we consider the block decomposition of the endomorphism L̂ under the splitting V = TxM ⊕ Rx. Using the blockwise LU decomposition of L̂, L̂ = ( L l lt n ) = ( L 0 lt 1 )( I L−1l 0 n− ltL−1l ) , we can compute the block decomposition of the adjugate of L̂: Adj L̂ = ( 1 det L ( det L̂ Adj L + (Adj L)llt(Adj L)t ) −(Adj L)l −lt(Adj L)t det L ) . Note that we have assumed without loss of generality that L is invertible. Now take a basis e1, . . . , en of TxM and complete it with e0 := x to a basis of V . With respect to this basis the components of the Killing tensor K corresponding to the algebraic curvature tensor (6.13) read Kij = R0i0j = (Adj L̂)00(Adj L̂)ij − (Adj L̂)0i(Adj L̂)0j det L̂ = (AdjL)ij . That is, K is given by (6.12), as was to be shown. � The adjugate of an (n× n)-matrix M is a polynomial in M of degree n− 1, given by AdjM = χM (λ)− χM (0) λ ∣∣∣∣ λ=M , where χM is the characteristic polynomial of M . Therefore a Benenti system is a polynomial in λ of degree n− 1: K(λ) = Adj(L− λg) = n−1∑ j=0 Kjλ j . Up to signs, the leading coefficient of this polynomial is the metric, the next coefficient is the special Killing tensor (6.2) and the constant term is the Benenti–Killing tensor (6.9): Kn−1 = lim λ→∞ 1 λn−1 K(λ) = lim λ→∞ Adj ( 1 λL− g ) = (−1)n−1g, Kn−2 = d dλ ∣∣∣∣ λ=0 λn−1K ( 1 λ ) = d dλ ∣∣∣∣ λ=0 Adj (λL− g) = (−1)n−1 ( L− (trL)g ) , K0 = K(0) = AdjL. Note that since all the K(λ) commute, the n coefficients Kj span a vector space of mutually commuting integrable Killing tensors. If the n coefficients are linearly independent, then this space is a Stäckel system and the Kj form a basis therein. However, it is in general an open question under which conditions this will be the case. For S3 we can derive the answer directly from our algebraic description. 38 K. Schöbel Lemma 6.20. For a symmetric tensor h on a four dimensional vector space we have (Adjh) 7 (Adjh) deth = ∗(h7 h)∗, (6.14) where ∗ is the Hodge star operator. Remark 6.21. Regarding (2.2), the conjugation of an algebraic curvature tensor with the Hodge star is simply a sign change in the trace free Ricci part: R = W + T + S, ∗R∗ = W −T + S. Proof. Choose a basis in which h is diagonal with diagonal elements hi := hii. For simplicity of notation we assume that deth 6= 0. Then hi 6= 0 and Adjh is diagonal with diagonal elements deth/hi. Substituting Adjh for h in (6.10) and comparing to (6.10) shows that the diagonal algebraic curvature tensors R̃ = (Adjh) 7 (Adjh), R = h7 h are related by R̃ijij = 2 deth hi deth hj = 2(deth) h0h1h2h3 hihj = (deth)2hkhl = (deth)Rklkl, for {i, j, k, l} = {0, 1, 2, 3}. Now the identity (6.14) follows from (2.13) and the fact that R and ∗R∗ differ in the sign of the trace free Ricci part. � Proposition 6.22. Let L be a special conformal Killing tensor on S3 ⊂ V , i.e. the restriction of a (constant) symmetric tensor L̂ from V to S3, and consider the corresponding Benenti system K(λ) = Adj(L− λg) = K2λ 2 +K1λ+K0. Then the coefficients K0, K1 and K2 span a Stäckel system if and only if L̂ has no triple eigenvalue. Proof. By Proposition 6.19 and the identity (6.14) the algebraic curvature tensor of K(λ) is the polynomial R(λ) = ∗ ( (L̂− λg) 7 (L̂− λg) ) ∗ = λ2g 7 g + 2λL̂7 g + ∗ ( L̂7 L̂ ) ∗ =: λ2R2 + λR1 +R0. The coefficients R0, R1 and R2 are the algebraic curvature tensors ofK0, K1 andK2 respectively. We can assume them to be diagonal. Since K2 is minus the metric, K0, K1 and K2 are linearly independent if and only if R0 and R1 define two distinct points in the KS-variety. Choose a basis in which L̂ is diagonal with the eigenvalues Λi on the diagonal. Without loss of generality we can assume that L̂ is trace free. Motivated by (2.12), we can thus parametrise L̂ by a vector n ∈ R3 as Λ0 = nα + nβ + nγ 2 , Λα = nα − nβ − nγ 2 , (6.15a) where nα = Λ0 + Λα = −(Λβ + Λγ). (6.15b) Then L̂ has a triple eigenvalue if and only if \|n1\| = \|n2\| = \|n3\|. This condition is equivalent to the condition that the KS-matrices ν(n) and ι(n) in (4.8) are proportional, i.e. define the same The Variety of Integrable Killing Tensors on the 3-Sphere 39 point in the KS-variety. Our proof will therefore be finished if we show that R0 and R1 define the points ν(n) respectively ι(n) in the KS-variety. That the algebraic curvature tensor R1 = 2L̂ 7 g defines the point ι(n) is the statement of Lemma 6.9. And since R0 = ∗(L̂ 7 L̂)∗ differs from L̂ 7 L̂ in the sign of the trace free Ricci part, Lemma 6.15 implies that for R0 the KS-matrix (1.9) is given by ∆α = (Λ0 − Λα)(Λβ − Λγ) = n2β − n2γ , tα = −(Λ0Λα − ΛβΛγ) = nβnγ . Now compare this to ν(n) in (4.8) to finish the proof. � The Benenti system of a special conformal Killing tensors L defines a map R→ K(S3) given by λ 7→ K(λ) = Adj(L − λg). In the proof above we have actually shown that this defines a map from P1 = R ∪ {∞} to the KS-variety which maps 0 7→ ν(n) and ∞ 7→ ι(n), where n ∈ P2 is determined by L. If the points ι(n) and ν(n) coincide, the image of this map is a single point. If they are distinct, the image is – by definition – the Stäckel line through both points. In this case the image of the Benenti system is this Stäckel line without the point ι(n). We can now prove Theorem 1.17. It suffices to show that the image of diagonal algebraic curvature tensors of the form λ2h 7 h is dense in the KS-variety. By Proposition 6.19 the algebraic curvature tensors of this form are dense in a Benenti system, so it suffices to show that the image of Benenti systems in the KS-variety is dense. But this follows from the preceding paragraph and part 2 of Proposition 4.8. 6.4 Extension of integrable Killing tensors If we fix an orthogonal decomposition Rn1+n2+2 = Rn1+1⊕Rn2+1, then two algebraic curvature tensors R1 on Rn1+1 and R2 on Rn2+1 define an algebraic curvature tensor R1⊕R2 on Rn1+n2+2. Correspondingly, two Killing tensors K1 on Sn1 and K2 on Sn2 define a Killing tensor K1⊕K2 on Sn1+n2+1. In particular, we can extend any Killing tensor K on Sn−1 by zero to a Killing tensor K ⊕ 0 on Sn. From the algebraic integrability conditions we see that K1 ⊕K2 is integrable if K1 and K2 are both integrable. This direct sum operation defines an embedding K(Sn1)×K(Sn2) ↪→ K ( Sn1+n2+1 ) , where K(Sn) denotes the space of integrable Killing tensors on Sn. In particular we have an embedding K ( Sn−1 ) ↪→ K ( Sn ) . Definition 6.23. We will call (integrable) Killing tensors arising in the way described above extensions of Killing tensors from (Sn1 , Sn2) respectively Sn−1. As an example, let us determine which integrable Killing tensors on S3 are extensions from S2 respectively (S1, S1). Note that in dimension one and two all Killing tensors are integrable. This follows directly from the integrability conditions. An algebraic curvature tensor R on R4 is an extension by zero of an algebraic curvature tensor on R3 under the inclusion (x1, x2, x3) 7→ (0, x1, x2, x3) if and only if Ra1b1a2b2 is zero whenever one of the indices is zero. In particular, the diagonal algebraic curvature tensor (1.8) on S3 is an extension from S2 if and only if wα + s 12 = −tα. In this case the KS-matrix (1.9) is equal to (4.6a) with n1 = n2 = n3. Therefore the extensions of Killing tensors from S2 correspond to the isokernel planes in the KS-variety. In particular, the face centers of the octahedron in the KS-variety correspond to the extension of the metric on S2 under the four embeddings S2 ⊂ S3 given by intersecting S3 ⊂ R4 with each of the four coordinate hyperplanes. 40 K. Schöbel An algebraic curvature tensor R on R4 is a sum under the decomposition of (x0, x1, x2, x3) into (x0, x1) and (x2, x3) if and only if the components Ra1b1a2b2 are zero unless all four indices are either in {0, 1} or in {2, 3}. In particular, a diagonal algebraic curvature tensor (1.8) is an extension from (S1, S1) if and only if w2 + s 12 = t2 = w3 + s 12 = t3 = 0. In this case the KS-matrix (1.9) is equal to0 0 0 0 −w1 −t1 0 +t1 +w1  = −w1 − t1 2 V−1 − w1 + t1 2 V+1, where V+1 and V−1 are rank one singular points in the KS-variety, cf. (4.4). Therefore the exten- sions of Killing tensors from (S1, S1) correspond to the projective lines joining opposite vertices of the octahedron in the KS-variety. In particular, each vertex of the octahedron corresponds to an extension of the metrics on two orthogonal copies of S1 in S3, given by intersecting S3 ⊂ R4 with a pair of orthogonal coordinate planes. 7 Separation coordinates We eventually demonstrate how separation coordinates on S3 and their classification arise nat- urally and in a purely algebraic way from our description of integrable Killing tensors. Kalnins and Miller proved that separation coordinates on Sn are always orthogonal [22]. From the work of Eisenhart we know that orthogonal separation coordinates are in bijective cor- respondence with Stäckel systems [13]. And in the preceding sections we have proven that on S3 every Stäckel system contains a special Killing tensor which is essentially unique. By definition, special Killing tensors are in bijective correspondence with special conformal Killing tensors. But Crampin has shown that the eigenvalues of a special conformal Killing tensor are constant on the coordinate hypersurfaces of the corresponding separation coordinates [8]. Hence, if these eigenvalues are simple and non-constant, they can be used as separation coordinates. This re- duces the classification of separation coordinates on S3 to a computation of the eigenvalues of special conformal Killing tensors. The computation is considerably simplified by the fact that a special conformal Killing tensor on Sn ⊂ V is the restriction of a (constant) symmetric tensor on V . 7.1 Eigenvalues of special conformal Killing tensors Let L be a special conformal Killing tensor on a non-flat constant curvature manifold M ⊂ V , i.e. the restriction from V to M of a (constant) symmetric tensor L̂. From now on we consider L and L̂ as endomorphisms. We want to compute the eigenvalues λ1(x) 6 · · · 6 λn(x) of L : TxM → TxM for a fixed point x ∈ M and relate them to the (constant) eigenvalues Λ0 6 · · · 6 Λn of L̂ : V → V . Let P : V → V be the orthogonal projection from V ∼= TxM ⊕ Rx to TxM . Then the restriction of PL̂P to TxM is L and the restriction of PL̂P to x is zero. Therefore x is an eigenvector of PL̂P with eigenvalue 0 and the remaining eigenvectors and eigenvalues are those of L. This means the eigenvalue equation Lv = λv is equivalent to PL̂Pv = λv for v ⊥ x. By the definition of P we have Pv = v and PL̂v = L̂v − g(L̂v, x)x. Hence we seek common solutions to the two equations (L̂− λ)v = g(L̂v, x)x, g(x, v) = 0. If λ is not an eigenvalue of L̂, then (L̂ − λ) is invertible and v is proportional to (L̂ − λ)−1x. The condition v ⊥ x then yields the equation g (( L̂− λ )−1 x, x ) = 0 The Variety of Integrable Killing Tensors on the 3-Sphere 41 for λ. In an eigenbasis of L̂ this equation reads q(λ) := n∑ k=0 x2k Λk − λ = 0. (7.1) That is, the zeroes of the function q(λ) are eigenvalues of L. As depicted in Fig. 2, q(λ) goes to 0 for λ → ±∞, has poles at the eigenvalues Λk of L̂ unless xk = 0 and is monotonely increasing in between. λ f(λ) Λ0 Λ1 λ1 Λ2 λ2 Λn−1 Λn λn Figure 2. Eigenvalues of a special conformal Killing tensor on Sn. Let us first suppose that L̂ has only simple eigenvalues. In this case (7.1) is the defining equation for elliptic coordinates λ1(x), . . . , λn(x) on Sn with parameters Λ0 < Λ1 < · · · < Λn [36]. If xk 6= 0 for all k = 0, . . . , n, then q has exactly n+ 1 poles at Λ0, . . . ,Λn and n zeroes in between. If xk = 0, then q has no pole at Λk. But xk = 0 means that the k-th eigenvector of L̂ is orthogonal to x and hence also an eigenvector of L with eigenvalue Λk. We conclude that the eigenvectors vk of L̂ which are orthogonal to x are eigenvectors of L with the same eigenvalue Λk, and that the remaining eigenvalues of L are the zeroes λk of the function q(λ) with corresponding eigenvector vk = (L̂ − λk)−1x. In particular, the eigenvalues of L̂ and L intertwine: Λ0 6 λ1(x) 6 Λ1 6 λ2(x) 6 · · · 6 λn(x) 6 Λn. (7.2) Note that the common denominator of the fractions in the sum (7.1) is the characteristic poly- nomial χL̂(λ) of L̂. Therefore the function q is the quotient of the characteristic polynomials of L and L̂. If L̂ has a simple eigenvalue, then the corresponding eigenspace is one dimensional and has only a trivial intersection with a generic tangent space to M , because M ⊂ V is of codimen- sion one. If L̂ has a double eigenvalue, then the corresponding eigenspace is a plane and has a one-dimensional intersection with a generic tangent space to M . This defines a field of com- mon eigenvectors for L and L̂ on M . The corresponding eigenvalue is the same for L and L̂ and hence constant on M . This also follows from (7.2). In the same way an eigenvalue of L̂ with multiplicity m+ 1 defines an m-dimensional eigenspace distribution of L with constant eigenvalue. Proposition 7.1. Let L be a special conformal Killing tensor on a constant curvature manifold M ⊂ V , i.e. the restriction of a constant symmetric trace free tensor L̂ on V . Then: 1. If L̂ has only simple eigenvalues, then the eigenvalues of L define elliptic coordinates on M whose parameters are the eigenvalues of L̂. 2. Any multiple eigenvalue of L̂ is a constant eigenvalue for L. 3. L has simple eigenvalues if and only if L̂ has at most double eigenvalues. 42 K. Schöbel 7.2 Killing tensors and separation coordinates on S2 Before we turn to S3, let us briefly discuss integrable Killing tensors and separation coordinates on S2. The reason is that their extensions will appear on S3. First note that in dimension two every Killing tensor is integrable. So we do not have to deal with the integrability conditions. Note also that algebraic curvature tensors on R3 have a zero Weyl part and are thus determined by their Ricci tensor T alone. This means that Killing tensors on S2 ⊂ R3 correspond to symmetric tensors T on R3, where the metric corresponds to the identity. Stäckel systems on S2 are of dimension two and contain the metric. That implies that each Stäckel system on S2 corresponds to a plane of Ricci tensors spanned by the identity and some trace free symmetric tensor, the latter being unique up to multiples. Moreover, the isometry group acts by conjugation on the Ricci tensor. Accordingly, the classification of separation coordinates on S2 modulo isometries is equivalent to the classification of trace free symmetric tensors T on R3 under the orthogonal group O(3). In particular, every Killing tensor on S2 is special and we can make use of the results in the previous section. Depending on the multiplicities of the eigenvalues t1 6 t2 6 t3 of the trace free Ricci tensor T of a Killing tensor on S2 we consider the following two cases. Elliptic coordinates. If t1 < t2 < t3, then the restriction L of T from R3 to S2 ⊂ R3 has two different eigenvalues, given by the zeros of its characteristic polynomial χL(λ) = (t2 − λ)(t3 − λ)x21 + (t3 − λ)(t1 − λ)x22 + (t1 − λ)(t2 − λ)x23 = 0. Using that x21 + x22 + x23 = 1 and t1 + t2 + t3 = 0 and writing the endomorphisms T and AdjT as forms, this quadratic equation reads λ2 + T (x, x)λ+ (AdjT )(x, x) = 0. The two solutions λ1(x) and λ2(x) are real and satisfy t1 < λ1(x) < t2 < λ2(x) < t3. As mentioned above, they define elliptic coordinates on S2 with parameters t1 < t2 < t3. Spherical coordinates. If t1 < t2 = t3, the characteristic polynomial of L reduces to χL(λ) = (t2 − λ) [ (t2 − λ)x21 + (t1 − λ) ( x22 + x23 )] = 0. One of the eigenvalues only depends on x1 and the other is constant: λ1(x) = t2x 2 1 + t1 ( 1− x21 ) , λ2(x) = t2. The curves of constant λ1(x) are circles of constant latitude and the meridians complete them to an orthogonal coordinate system – the spherical coordinates. The case t1 = t2 < t3 is analogue and yields spherical coordinates as well. Note that even so, both coordinate systems are not isometric. This is because they differ not only by an isometry but also by an exchange of the two coordinates. We can use the action of the isometry group O(3) to diagonalise the Ricci tensor and thereby the algebraic curvature tensor. Hence the projective variety of integrable Killing tensors on S2 with diagonal algebraic curvature tensor is isomorphic to the projective plane P2, where the point (1 : 1 : 1) ∈ P2 corresponds to the metric on S2. Moreover, Stäckel systems on S2 correspond to projective lines through this point, since they always contain the metric. The residual isometry group action on Killing tensors with diagonal algebraic curvature tensor is the natural action of S3 on P2 by permutations. 7.3 The classif ication We now recover the well known classification of separation coordinates on S3 [13, 22]. In particular, this will explain the graphical procedure given in [22] in terms of the classification The Variety of Integrable Killing Tensors on the 3-Sphere 43 Table 1. Classification of separation coordinates on S3. type coordinates induced from [13] [22] (0123) elliptic – V (1) (01(23)) oblate Lamé rotational – I (2a) (0(12)3) prolate Lamé rotational – I (2b) ((01)(23)) cylindrical (S1, S1) III (5) (0(123)) Lamé subgroup reduction S2 elliptic IV (3) (0(1(23))) spherical S2 spherical II (4) of trace free symmetric tensors under the orthogonal group. The results are summarised in Table 1. As before, we will denote by Λ0 6 Λ1 6 Λ2 6 Λ3 the ordered eigenvalues of the symmetric ten- sor L̂ associated to a Stäckel system and parametrised by a vector (n1, n2, n3) ∈ R3 via (6.15a). Recall that we have the freedom to add a multiple of the metric to a Killing tensor or to mul- tiply it with a constant. Both operations only affect the eigenvalues, not the eigenspaces, and therefore result in different parametrisations of the same coordinate system. Whereas Kalnins and Miller use this freedom to set Λ0 = 0 and Λ1 = 1, here we use it to assume that L̂ is trace free and then to consider (Λ0 : Λ1 : Λ2 : Λ3) respectively (n1 : n2 : n3) as a point in P2. In particular, changing the sign of L̂ reverses the order of the eigenvalues. This leaves us with the following five different alternatives for the multiplicities of the ordered set of eigenvalues of L̂. For simplicity we write “i” for “Λi” and denote multiple eigenvalues by parenthesising them: (0123), (01(23)), (0(12)3), ((01)(23)), (0(123)). (0123) Elliptic coordinates. We have already seen that if the eigenvalues of L̂ are simple, then the eigenvalues of L are also simple and define elliptic coordinates on Sn given implicitly by (7.1). (01(23)) Oblate Lamé rotational coordinates. If L̂ has a double eigenvalue Λk−1 = Λk, then the corresponding eigenvalue λk of L is constant and can not be used as a coordinate. Nevertheless, we can derive the missing coordinate from the fact that the coordinate hypersur- faces are orthogonal to the eigenspaces of L. Let E be an eigenplane of L̂ and denote by E⊥ its orthogonal complement in V . Recall that the corresponding eigenspace of L at a point x ∈ Sn is given by the intersection of E with the tangent space TxS n. Its orthogonal complement in TxS n is the intersection of TxS n with the hyperplane that contains both x and E⊥. The coordinate hypersurfaces are therefore the intersections of hyperplanes containing E⊥ with Sn and can be parametrised by the polar angle in E. Hence this angle can be taken as a coordinate. By (7.2), L has single eigenvalues given by (7.1) if and only if L̂ has at most double eigen- values and every double eigenvalue of L̂ determines a constant eigenvalue for L. If we replace every constant eigenvalue by the polar angle in the corresponding eigenplane of L̂, we obtain a complete coordinate system. Concretely, in the case Λ0 < Λ1 < Λ2 = Λ3 the separation coordinates are defined by the solutions λ1(x) and λ2(x) of (7.1) together with the polar angle in the (x2, x3)-plane. For Λ0 < Λ1 < Λ2 = Λ3 we have n1 < n2 = n3. The corresponding Stäckel line thus goes through the two points in the KS-variety given by the KS-matrices (4.8): 0 −n2 n2 n2 0 −n1 −n2 n1 0  ,  0 −n1n2 n1n2 n1n2 n22 − n21 −n22 −n1n2 n22 n21 − n22  . 44 K. Schöbel Subtracting n1 times the first matrix from the second, we see that the same Stäckel line is also determined by the two points 0 −n2 n2 n2 0 −n1 −n2 n1 0  , V−1 = 0 0 0 0 +1 −1 0 +1 −1  . (0(12)3) Prolate Lamé rotational coordinates. The case Λ0 < Λ1 = Λ2 < Λ3 is analogue and yields Lamé rotational coordinates which differ from the above by an isometry and an exchange of two of the coordinates. ((01)(23)) Cylindrical coordinates. In the case Λ0 = Λ1 < Λ2 = Λ3 equation (7.1) has only one non-constant solution, namely λ2(x): λ1 = Λ1 6 λ2(x) = Λ1 ( x22 + x23 ) + Λ3 ( x20 + x21 ) 6 λ3 = Λ3. The separation coordinates are obtained by replacing the constant eigenvalues λ1 and λ3 by the polar angle in the (x0, x1) respectively (x2, x3)-plane. For Λ0 = Λ1 < Λ2 = Λ3 we have n1 < n2 = n3 = 0 and the corresponding Stäckel line goes through the two points in the KS-variety given by the KS-matrices (4.8):0 0 0 0 0 −n1 0 +n1 0  , 0 0 0 0 −n21 0 0 0 n21  . Equivalently, this Stäckel line goes through the two points V−1 = 0 0 0 0 +1 −1 0 +1 −1  , V+1 = 0 0 0 0 +1 +1 0 −1 −1  , i.e. opposed vertices of the octahedron in the KS-variety. We have seen that this describes extensions from (S1, S1) to S3. (0(123)) Coordinates extended from S2 ⊂ S3. In the remaining case Λ0<Λ1=Λ2=Λ3, equation (7.1) has the solutions λ1(x) = Λ1x 2 0 + Λ0 ( x21 + x22 + x23 ) , λ2 = λ3 = Λ1. We can take the non-constant eigenvalue λ1(x) as one separation coordinate on S3. However, since the other two eigenvalues of L are equal, we can not recover all separation coordinates from the special conformal Killing tensor L in this case. We will have to consider another Killing tensor in the Stäckel system instead – one with simple eigenvalues. Before we do so, let us collect some facts from previous sections for the case Λ0<Λ1=Λ2=Λ3. Proposition 7.2. Consider the following objects associated to a system of separation coordinates on S3: • the Stäckel system, • a special conformal Killing tensor L whose eigenvalues are constant on the coordinate hypersurfaces, • the Benenti system K(λ) = Adj(L− λg) of L, • the special Killing tensor K = L− (trL)g in the Stäckel system, • the Stäckel line in the KS-variety, The Variety of Integrable Killing Tensors on the 3-Sphere 45 • the trace free symmetric tensor L̂ on V which restricts to L, • the vector ~n = (n1, n2, n3) ∈ R3 parametrising L̂ via (6.15a). Then the following statements are equivalent: 1. The Stäckel system consists of extensions of Killing tensors from S2 ⊂ S3 (up to the addition of a multiple of the metric). 2. L has a multiple eigenvalue (which is constant). 3. K(λ) does not span the Stäckel system. 4. K corresponds to a singular skew symmetric point in the KS-variety. 5. The Stäckel line lies on an isokernel plane in the KS-variety. 6. L̂ has a triple eigenvalue. 7. \|n1\| = \|n2\| = \|n3\|. Proof. The equivalence of 6 and 7 is a direct consequence of (6.15a). The equivalence to the statements 1–5 follows from Section 6.4, Proposition 7.1, Proposition 6.22, Proposition 4.4 and Lemma 4.7, respectively. � To make the extension of Killing tensors from S2 ⊂ S3 more explicit, consider a Killing tensor on S2 with algebraic curvature tensor R and Ricci tensor T and choose a basis in which T is diagonal with diagonal elements t1, t2 and t3. Then R is also diagonal and, if T is trace free, given by R1212 = t1 + t2 = −t3, R2323 = t2 + t3 = −t1, R3131 = t3 + t1 = −t2. In Section 6.4 we saw that this Killing tensor extends to a Killing tensor on S3 corrsponding to the point in the KS-variety that is defined by the KS-matrix Vt := t3 − t2 −t3 t2 t3 t1 − t3 −t1 −t2 t1 t2 − t1  = t1V+1 + t2V+2 + t3V+3. If T is not trace free, we obtain the same result. This shows that the eigenvalues of the Ricci tensor of a Killing tensor on S2 induce barycentric coordinates on the isokernel planes in the KS-variety. In Section 7.2 we have already computed the eigenvalues λ2(x) and λ3(x) of a Killing tensor on S2, parametrised by the eigenvalues of its Ricci tensor T . We see that if one perturbs the special Killing tensor on S3 within a Stäckel system, the constant double eigenvalue λ2 = λ3 splits up into two simple eigenvalues λ2(x) and λ3(x), which come from separation coordinates on S2 under an embedding S2 ⊂ S3. On the other hand, the eigenvalue λ1(x) remains unchanged. Together, they define separation coordinates on S3. We denote this by writing the multiplicities of L̂ as (0(123)) respectively (0(1(23))), depending on whether the multiplicities of the Ricci tensor T are (123) or (1(23)). This yields the remaining two systems of separation coordinates: (0(1(23))) Spherical coordinates. These are extensions of spherical coordinates on S2 and correspond to Stäckel lines through a face center and a vertex of the octahedron in the KS-variety. (0(123)) Lamé subgroup reduction. These are extensions of elliptic coordinates on S2 and correspond to Stäckel lines through a face center, but not a vertex of the octahedron in the KS-variety. 46 K. Schöbel 7.4 The classif ication space The non-generic separation coordinates are characterised by n1 = ±n2, n2 = ±n3 and n3 = ±n1. This defines six projective lines that divide the projective plane of skew symmetric matrices in the KS-variety into twelve congruent triangles. The four singular skew symmetric points are given by \|n1\| = \|n2\| = \|n3\| and each of the triangles has two of them as vertices. Upon a blow-up in these points, the triangles become pentagons. Any of these pentagons constitutes a fundamental domain for the S4-action. Regarding our notation for the multiplicities of L̂, we realise that the vertices are labeled by the different possibilities to parenthesise the ordered set of eigenvalues of L̂ with pairs of correctly matching parentheses and that the edges correspond to a single application of the associativity rule. This naturally identifies the classification space for separation coordinates on S3 with the associahedron K4, also known as Stasheff polytope [46]. (0123) elliptic cylindrical spherical spherical spherical spherical ((01)(23)) (0(1(23))) (0((12)3)) ((0(12))3) (((01)2)3) oblate Lamé Lamé subgroup prolate Lamé Lamé subgroup oblate Lamé (0 1( 23 )) (0(123)) (0(12)3) (( 01 2) 3) ((01)23) Figure 3. Identification of the moduli space of separation coordinates on S3 with the associahedron K4. Acknowledgements I would like to express my gratitude to Robert Milson for his motivation and the inspiring discussions about my findings. I would also like to thank Alexander P. Veselov for pointing out the link between my solution and moduli spaces of stable curves. Finally, I would like to thank the anonymous referees, who helped to improve the paper considerably with their comments and additional references. References [1] Ballesteros Á., Enciso A., Herranz F.J., Ragnisco O., A maximally superintegrable system on an n- dimensional space of nonconstant curvature, Phys. D 237 (2008), 505–509, math-ph/0612080. [2] Benenti S., Inertia tensors and Stäckel systems in the Euclidean spaces, Rend. Sem. Mat. Univ. Politec. Torino 50 (1992), 315–341. [3] Benenti S., Orthogonal separable dynamical systems, in Differential Geometry and its Applications (Opava, 1992), Math. Publ., Vol. 1, Silesian Univ. Opava, Opava, 1993, 163–184. [4] Benenti S., Special symmetric two-tensors, equivalent dynamical systems, cofactor and bi-cofactor systems, Acta Appl. Math. 87 (2005), 33–91. [5] Benenti S., Chanu C., Rastelli G., Remarks on the connection between the additive separation of the Hamilton–Jacobi equation and the multiplicative separation of the Schrödinger equation. II. First integrals and symmetry operators, J. Math. Phys. 43 (2002), 5223–5253. http://dx.doi.org/10.1016/j.physd.2007.09.021 http://arxiv.org/abs/math-ph/0612080 http://dx.doi.org/10.1007/s10440-005-1138-9 http://dx.doi.org/10.1063/1.1506181 The Variety of Integrable Killing Tensors on the 3-Sphere 47 [6] B laszak M., Bi-Hamiltonian representation of Stäckel systems, Phys. Rev. E 79 (2009), 056607, 9 pages, arXiv:0904.2070. [7] Bolsinov A.V., Matveev V.S., Geometrical interpretation of Benenti systems, J. Geom. Phys. 44 (2003), 489–506. [8] Crampin M., Conformal Killing tensors with vanishing torsion and the separation of variables in the Hamilton–Jacobi equation, Differential Geom. Appl. 18 (2003), 87–102. [9] Crampin M., Projectively equivalent Riemannian spaces as quasi-bi-Hamiltonian systems, Acta Appl. Math. 77 (2003), 237–248. [10] Crampin M., Sarlet W., Bi-quasi-Hamiltonian systems, J. Math. Phys. 43 (2002), 2505–2517. [11] Deeley R.J., Horwood J.T., McLenaghan R.G., Smirnov R.G., Theory of algebraic invariants of vector spaces of Killing tensors: methods for computing the fundamental invariants, in Symmetry in Nonlinear Mathematical Physics, Proceedings of Institute of Mathematics, Vol. 50, Editors A.G. Nikitin, V.M. Boyko, R.O. Popovych, I.A. Yehorchenko, Institute of Mathematics, Kyiv, 2004, Part 3, 1079–1086. [12] Deligne P., Mumford D., The irreducibility of the space of curves of given genus, Inst. Hautes Études Sci. Publ. Math. (1969), 75–109. [13] Eisenhart L.P., Separable systems of Stäckel, Ann. of Math. 35 (1934), 284–305. [14] Frolov V.P., Zelnikov A., Introduction to black hole physics, Oxford University Press, Oxford, 2011. [15] Horwood J.T., McLenaghan R.G., Smirnov R.G., Invariant classification of orthogonally separable Hamil- tonian systems in Euclidean space, Comm. Math. Phys. 259 (2005), 670–709, math-ph/0605023. [16] Ibort A., Magri F., Marmo G., Bihamiltonian structures and Stäckel separability, J. Geom. Phys. 33 (2000), 210–228. [17] Kalnins E.G., Separation of variables for Riemannian spaces of constant curvature, Pitman Monographs and Surveys in Pure and Applied Mathematics, Vol. 28, Longman Scientific & Technical, Harlow, 1986. [18] Kalnins E.G., Kress J.M., Miller Jr. W., Superintegrability in a non-conformally-flat space, J. Phys. A: Math. Theor. 46 (2013), 022002, 12 pages, arXiv:1211.1452. [19] Kalnins E.G., Kress J.M., Miller Jr. W., Winternitz P., Superintegrable systems in Darboux spaces, J. Math. Phys. 44 (2003), 5811–5848, math-ph/0307039. [20] Kalnins E.G., Kress J.M., Winternitz P., Superintegrability in a two-dimensional space of nonconstant curvature, J. Math. Phys. 43 (2002), 970–983, math-ph/0108015. [21] Kalnins E.G., Kuznetsov V.B., Miller Jr. W., Quadrics on complex Riemannian spaces of constant curvature, separation of variables, and the Gaudin magnet, J. Math. Phys. 35 (1994), 1710–1731, hep-th/9308109. [22] Kalnins E.G., Miller Jr. W., Separation of variables on n-dimensional Riemannian manifolds. I. The n-sphere Sn and Euclidean n-space Rn, J. Math. Phys. 27 (1986), 1721–1736. [23] Kalnins E.G., Miller Jr. W., Winternitz P., The group O(4), separation of variables and the hydrogen atom, SIAM J. Appl. Math. 30 (1976), 630–664. [24] Knudsen F.F., The projectivity of the moduli space of stable curves. II. The stacks Mg,n, Math. Scand. 52 (1983), 161–199. [25] Knudsen F.F., The projectivity of the moduli space of stable curves. III. The line bundles on Mg,n, and a proof of the projectivity of M̄g,n in characteristic 0, Math. Scand. 52 (1983), 200–212. [26] Koenigs G.X.P., Sur les géodésiques a integrales quadratiques, in Le cons sur la théorie générale des surfaces, Vol. 4, Editor J.G. Darboux, Chelsea Publishing, 1972, 368–404. [27] Kuznetsov V.B., Equivalence of two graphical calculi, J. Phys. A: Math. Gen. 25 (1992), 6005–6026. [28] Kuznetsov V.B., Quadrics on real Riemannian spaces of constant curvature: separation of variables and connection with Gaudin magnet, J. Math. Phys. 33 (1992), 3240–3254. [29] Lamé G., Sur les surfaces isothermes dans les corps homogènes en équilibre de température, J. Math. Pures Appl. 2 (1837), 147–188. [30] Levi-Civita T., Sulla integrazione della equazione di Hamilton–Jacobi per separazione di variabili, Math. Ann. 59 (1904), 383–397. [31] Lundmark H., Newton systems of cofactor type in Euclidean and Riemannian spaces, Dissertations, Vol. 719, Linköping Studies in Science and Technology, Linköping, 2001. [32] Matveev V.S., Mounoud P., Gallot–Tanno theorem for closed incomplete pseudo-Riemannian manifolds and applications, Ann. Global Anal. Geom. 38 (2010), 259–271, arXiv:0909.5344. http://dx.doi.org/10.1103/PhysRevE.79.056607 http://arxiv.org/abs/0904.2070 http://dx.doi.org/10.1016/S0393-0440(02)00054-2 http://dx.doi.org/10.1016/S0926-2245(02)00140-7 http://dx.doi.org/10.1023/A:1024907509042 http://dx.doi.org/10.1063/1.1462856 http://dx.doi.org/10.1007/BF02684599 http://dx.doi.org/10.1007/BF02684599 http://dx.doi.org/10.2307/1968433 http://dx.doi.org/10.1007/s00220-005-1331-8 http://arxiv.org/abs/math-ph/0605023 http://dx.doi.org/10.1016/S0393-0440(99)00051-0 http://dx.doi.org/10.1088/1751-8113/46/2/022002 http://dx.doi.org/10.1088/1751-8113/46/2/022002 http://arxiv.org/abs/1211.1452 http://dx.doi.org/10.1063/1.1619580 http://dx.doi.org/10.1063/1.1619580 http://arxiv.org/abs/math-ph/0307039 http://dx.doi.org/10.1063/1.1429322 http://arxiv.org/abs/math-ph/0108015 http://dx.doi.org/10.1063/1.530566 http://arxiv.org/abs/hep-th/9308109 http://dx.doi.org/10.1063/1.527088 http://dx.doi.org/10.1137/0130058 http://dx.doi.org/10.1088/0305-4470/25/22/025 http://dx.doi.org/10.1063/1.529542 http://dx.doi.org/10.1007/BF01445149 http://dx.doi.org/10.1007/BF01445149 http://dx.doi.org/10.1007/s10455-010-9211-7 http://arxiv.org/abs/0909.5344 48 K. Schöbel [33] Matveev V.S., Topalov P.̆I., Trajectory equivalence and corresponding integrals, Regul. Chaotic Dyn. 3 (1998), 30–45. [34] McLenaghan R.G., Milson R., Smirnov R.G., Killing tensors as irreducible representations of the general linear group, C. R. Math. Acad. Sci. Paris 339 (2004), 621–624. [35] Milson R., Schöbel K., Integrable Killing tensors on 3-dimensional constant curvature spaces, in preparation. [36] Neumann C., De problemate quodam mechanico, quod ad primam integralium ultraellipticorum classem revocatur, J. Reine Angew. Math. 56 (1859), 46–63. [37] Nijenhuis A., Xn−1-forming sets of eigenvectors, Nederl. Akad. Wetensch. Proc. Ser. A. 54 (1951), 200–212. [38] Olevskĭı M.N., Triorthogonal systems in spaces of constant curvature in which the equation ∆2u + λu = 0 allows a complete separation of variables, Mat. Sb. 27 (1950), 379–426. [39] Painlevé P., Sur les intégrales quadratiques des équations de la dynamique, Compt. Rend. 124 (1897), 221–224. [40] Rauch-Wojciechowski S., Marciniak K., Lundmark H., Quasi-Lagrangian systems of Newton equations, J. Math. Phys. 40 (1999), 6366–6398, solv-int/9909025. [41] Schöbel K.P., Algebraic integrability conditions for Killing tensors on constant sectional curvature manifolds, J. Geom. Phys. 62 (2012), 1013–1037, arXiv:1004.2872. [42] Schöbel K.P., Veselov A.P., Separation coordinates, moduli spaces and Stasheff polytopes, arXiv:1307.6132. [43] Singer I.M., Thorpe J.A., The curvature of 4-dimensional Einstein spaces, in Global Analysis (Papers in Honor of K. Kodaira), Univ. Tokyo Press, Tokyo, 1969, 355–365. [44] Sinjukov N.S., On the theory of a geodesic mapping of Riemannian spaces, Dokl. Akad. Nauk SSSR 169 (1966), 770–772. [45] Stäckel P., Die Integration der Hamilton–Jacobischen Differentialgleichung mittelst Separation der Vari- ablen, habilitationsschrift, Universität Halle, Halle, 1891. [46] Stasheff J.D., Homotopy associativity of H-spaces. I, Trans. Amer. Math. Soc. 108 (1963), 275–292. http://dx.doi.org/10.1070/rd1998v003n02ABEH000069 http://dx.doi.org/10.1016/j.crma.2004.07.017 http://dx.doi.org/10.1515/crll.1859.56.46 http://dx.doi.org/10.1063/1.533098 http://arxiv.org/abs/solv-int/9909025 http://dx.doi.org/10.1016/j.geomphys.2012.01.006 http://arxiv.org/abs/1004.2872 http://arxiv.org/abs/1307.6132 http://dx.doi.org/10.1090/S0002-9947-1963-99939-9 1 Introduction 1.1 The problem 1.2 Prior results 1.3 Method and results 1.4 Generalisations 1.5 Structure of the article 2 Properties of algebraic curvature tensors 2.1 Decomposition 2.2 The action of the isometry group 2.3 Aligned algebraic curvature tensors 2.4 Diagonal algebraic curvature tensors 2.5 The residual action of the isometry group 3 Solution of the algebraic integrability conditions 3.1 Reformulation of the first integrability condition 3.2 Integrability implies diagonalisability 3.3 Solution of the second integrability condition 4 The algebraic geometry of the KS-variety 5 Stäckel systems 5.1 Young tableaux 5.2 Commuting Killing tensors 6 Geometric constructions of integrable Killing tensors 6.1 Special Killing tensors 6.2 Killing tensors of Benenti type 6.3 Benenti systems 6.4 Extension of integrable Killing tensors 7 Separation coordinates 7.1 Eigenvalues of special conformal Killing tensors 7.2 Killing tensors and separation coordinates on S2 7.3 The classification 7.4 The classification space References

The Variety of Integrable Killing Tensors on the 3-Sphere

Репозитарії

Схожі ресурси