Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables

Olver and Rosenau studied group-invariant solutions of (generally nonlinear) partial differential equations through the imposition of a side condition. We apply a similar idea to the special case of finite-dimensional Hamiltonian systems, namely Hamilton-Jacobi, Helmholtz and time-independent Schröd...

Повний опис

Збережено в:

Бібліографічні деталі
Опубліковано в: :	Symmetry, Integrability and Geometry: Methods and Applications
Дата:	2012
Автори:	Broadbridge, P., Chanu, C.M., Miller Jr., Willard
Формат:	Стаття
Мова:	Англійська
Опубліковано:	Інститут математики НАН України 2012
Онлайн доступ:	https://nasplib.isofts.kiev.ua/handle/123456789/148652
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:	Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables / P. Broadbridge, C.M. Chanu, Willard Miller Jr. // Symmetry, Integrability and Geometry: Methods and Applications. — 2012. — Т. 8. — Бібліогр.: 36 назв. — англ.

Репозитарії

Digital Library of Periodicals of National Academy of Sciences of Ukraine

_version_	1859906568684830720
author	Broadbridge, P. Chanu, C.M. Miller Jr., Willard
author_facet	Broadbridge, P. Chanu, C.M. Miller Jr., Willard
citation_txt	Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables / P. Broadbridge, C.M. Chanu, Willard Miller Jr. // Symmetry, Integrability and Geometry: Methods and Applications. — 2012. — Т. 8. — Бібліогр.: 36 назв. — англ.
collection	DSpace DC
container_title	Symmetry, Integrability and Geometry: Methods and Applications
description	Olver and Rosenau studied group-invariant solutions of (generally nonlinear) partial differential equations through the imposition of a side condition. We apply a similar idea to the special case of finite-dimensional Hamiltonian systems, namely Hamilton-Jacobi, Helmholtz and time-independent Schrödinger equations with potential on N-dimensional Riemannian and pseudo-Riemannian manifolds, but with a linear side condition, where more structure is available. We show that the requirement of N−1 commuting second-order symmetry operators, modulo a second-order linear side condition corresponds to nonregular separation of variables in an orthogonal coordinate system, characterized by a generalized Stäckel matrix. The coordinates and solutions obtainable through true nonregular separation are distinct from those arising through regular separation of variables. We develop the theory for these systems and provide examples.
first_indexed	2025-12-07T16:00:09Z
format	Article
fulltext	Symmetry, Integrability and Geometry: Methods and Applications SIGMA 8 (2012), 089, 31 pages Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables? Philip BROADBRIDGE †, Claudia M. CHANU ‡ and Willard MILLER Jr. § † School of Engineering and Mathematical Sciences, La Trobe University, Melbourne, Australia E-mail: P.Broadbridge@latrobe.edu.au ‡ Dipartimento di Matematica G. Peano, Università di Torino, Torino, Italy E-mail: claudiamaria.chanu@unito.it § School of Mathematics, University of Minnesota, Minneapolis, Minnesota, 55455, USA E-mail: miller@ima.umn.edu URL: http://www.ima.umn.edu/~miller/ Received September 21, 2012, in final form November 19, 2012; Published online November 26, 2012 http://dx.doi.org/10.3842/SIGMA.2012.089 Abstract. Olver and Rosenau studied group-invariant solutions of (generally nonlinear) partial differential equations through the imposition of a side condition. We apply a similar idea to the special case of finite-dimensional Hamiltonian systems, namely Hamilton–Jacobi, Helmholtz and time-independent Schrödinger equations with potential on N -dimensional Riemannian and pseudo-Riemannian manifolds, but with a linear side condition, where more structure is available. We show that the requirement of N−1 commuting second-order symmetry operators, modulo a second-order linear side condition corresponds to nonregular separation of variables in an orthogonal coordinate system, characterized by a generalized Stäckel matrix. The coordinates and solutions obtainable through true nonregular separation are distinct from those arising through regular separation of variables. We develop the theory for these systems and provide examples. Key words: nonregular separation of variables; Helmholtz equation; Schrödinger equation 2010 Mathematics Subject Classification: 35Q40; 35J05 1 Introduction The primary motivation for this paper was the construction by Olver and Rosenau of group- invariant solutions of (generally nonlinear) partial differential equations through the imposition of a side condition [27, 28]. Various forms of conditional symmetry methods, including nonclas- sical symmetry reduction, have been applied by several authors, e.g. [1, 5, 6, 10, 21, 26, 29]. We are interested in using linear second-order side conditions and separation of variables to find explicit solutions of Hamilton–Jacobi, Helmholtz, Laplace, wave and heat equations. We exploit the special properties of finite order classical Hamiltonian systems and their quantum analogues represented by Schrödinger equations to obtain new results on separation of variables. The present paper is devoted to Hamilton–Jacobi and Helmholtz or time independent Schrödinger equations with potential but the ideas are clearly applicable to more general Hamiltonian sys- tems and to non-Hamiltonian systems such as diffusion equations that share some features with time-dependent Schrödinger equations. ?This paper is a contribution to the Special Issue “Symmetries of Differential Equations: Frames, Invariants and Applications”. The full collection is available at http://www.emis.de/journals/SIGMA/SDE2012.html mailto:P.Broadbridge@latrobe.edu.au mailto:claudiamaria.chanu@unito.it mailto:miller@ima.umn.edu http://www.ima.umn.edu/~miller/ http://dx.doi.org/10.3842/SIGMA.2012.089 http://www.emis.de/journals/SIGMA/SDE2012.html 2 P. Broadbridge, C.M. Chanu and W. Miller Jr. An example is the Schrödinger equation HΨ = EΨ where H = ∆ + V (x) and ∆ is the Laplace–Beltrami operator on some Riemannian or pseudo-Riemannian manifold. We look for solutions of this equation that also satisfy a side condition SΨ = 0 where S is some given linear partial differential operator in the variables x. A consistency condition for the existence of nontrivial solutions Ψ is [H,S] = AS for some linear partial differential operator A. Moreover a linear differential operator L will be a symmetry operator for H, (modulo SΨ = 0) if [H,L] = BS for some linear partial differential operator B. To make contact with solutions that are separable in some system of coordinates, we restrict to the case where the symmetries and the side condition are second-order partial differential operators. A Hamilton–Jacobi analog is the equation H = E where, in orthogonal coordinates, H = N∑ j=1 gjj(x)p2j + V (x), which is double the classical Hamiltonian. We look for solutions u(x) of the Hamilton–Jacobi equation, where pj = ∂xju, subject to the side condition S(x,p) = 0. The consistency require- ment is the Poisson bracket relation {H,S} = AS for some function A on phase space. Moreover a phase space function L will be a constant of the motion for H, (modulo S = 0) if {H,L} = BS for some phase space function B. Finding such systems directly from their definition leads to great computational complexity. We explore a new method, based on a generalization of Stäckel form, i.e., a generalization of separable systems corresponding tho a Stäckel matrix, that allows us to generate such systems efficiently. The second motivation for this paper is the general theory of separation of variables for both linear and nonlinear partial differential equations [17, 20, 23, 24, 25]. In these works the authors point out that there are two types of variable separation: regular and nonregular. Regular separation is the most familiar and was exploited by pioneers such as Stäckel [33] and Eisenhart [12, 14], [13, Appendix 13]. For regular orthogonal separation of a Helmholtz or Schrödinger equation on anN -dimensional manifold there are always N separation constants and the associated separable solutions form a basis for the solution space. There is a well developed theory for regular orthogonal separation of these equations, including classification of possible separable coordinate systems in various constant curvature spaces and intrinsic characterizations of the separable systems, see for example [3, 15, 16, 18, 19, 22, 31, 32] in addition to earlier cited references. For nonregular separation, on the other hand, the number of separation constants is strictly less than N and the separable solutions do not form a basis. Symmetry adapted solutions of partial differential equations are prominent examples of this class, but except for these special solutions there is virtually no structure or classification theory. First attempts of geometric interpretation of nonregular separation are given in [2, 8, 9, 11] where nonregular separation is considered as separation in which the separated solution must satisfy additional constraints that can be seen as side conditions for the equation. In this paper, however, we will show that solutions of Helmholtz and Schrödinger equations with second-order side conditions provide a class of nonregular orthogonal separation of variables that can be characterized intrinsically. Further, each of these systems is associated with a generalized Stäckel matrix and this association enables us to generate nonregular separable systems very easily. In Sections 2 and 3 we review the Stäckel construction for regular additive separation of Hamilton–Jacobi equations and regular multiplicative separation andR-separation for Helmholtz equations, and make some comments on their geometric characterizations. We also discuss the effect of adding vector and scalar potentials. In Section 4 we introduce a generalized Stäckel matrix with one arbitrary column and show that its use leads to additive separable solutions Nonregular Separation of Variables 3 of the Hamilton–Jacobi equation, with a side condition. We express the results in a Hamil- tonian formalism. Then in Section 5 we carry out the analogous construction for Helmholtz and Schrödinger eigenvalue equations. Section 7 is the main theoretical contribution of our paper. We show that the requirement of maximal nonregular separation for Hamilton–Jacobi equations is equivalent to separation with a generalized Stäckel matrix and can be characterized geometrically. With some modifications, the same is true of R-separation for Helmholtz and Schrödinger equations (though there is still a “generalized Rodrigues form” gap in the geometri- cal characterization). Section 8 is devoted to examples of nonregular separation and discussion of their various types and significance. We prove a “no go” theorem to the effect that nonregular R-separation does not occur for the Helmholtz (or Schrödinger) equation with no potential or scalar potential on a 2D Riemannian or pseudo-Riemannian manifold. However 2D nonregular separation can occur for equations with vector or magnetic potentials. Section 8.3 develops the theory for two-dimensional systems with vector potential. The self adjoint Schrödinger equation for a charged particle in two spatial dimensions, interacting with a classical electromagnetic field, again has no new separable coordinate systems that are obtainable from a generalised Stäckel matrix. However, nonself-adjoint equations such as the analogous solute transport equa- tion with first-order convective terms replacing magnetic potential terms, do indeed have new nonregular separable coordinate systems. We provide examples of nonregular R-separation for various zero potential, scalar potential and vector potential 3D systems: Euclidean space, Minkowski space and nonzero constant cur- vature space, including a Euclidean space example due to Sym [34] of nonregular separation with two side conditions. The final Section 9 sums up our conclusions and points the way for future research on nonregular separation. The coordinate systems and solutions for true nonregular separation, i.e., nonregular separation for which regular separation doesn’t occur, are distinct from those for regular separation and further study of their scope and significance is in order. 2 Review of regular orthogonal separation for the Hamilton–Jacobi equation H = E Write the Hamilton–Jacobi equation in terms of ui = ∂iu(x), as H ≡ N∑ i=1 H−2i u2i + V (x) = E. (2.1) Here, the metric in the orthogonal coordinates xi is ds2 = N∑ i=1 H2 i (dxi)2. We want to obtain additive separation, so that ∂jui ≡ ∂j∂iu = 0 for i 6= j. Requiring that the solution u depends on n parameters (λ1, . . . , λN ) implies the existence of separation equations in the form u2i + vi ( xi ) + N∑ j=1 sij ( xi ) λj = 0, i = 1, . . . , N, λ1 = −E. (2.2) Here ∂ksij(x i) = 0 for k 6= i and det(sij) 6= 0. We say that S = (sij) is a Stäckel matrix. Set T = S−1. Then (2.1) can be recovered from (2.2) provided H−2j = T 1j and V = ∑ j vjT 1j . The quadratic forms L` = N∑ j=1 T `j(u2j + vj) satisfy L` = −λ` for a separable solution. Furthermore, setting ui = pi, we claim{ L`,Lj } = 0, ` 6= j (2.3) 4 P. Broadbridge, C.M. Chanu and W. Miller Jr. where {H,K} = N∑ i=1 (∂xiK∂piH−∂xiH∂piK) is the Poisson Bracket. Thus the L`, 2 ≤ ` ≤ N , are constants of the motion for the Hamiltonian H = L(1). For the proof of (2.3) one notes that N∑ j=1 T `jsjk ( xj ) = δ`k, Differentiating this identity with respect to xi, we find N∑ j=1 ∂iT `jsjk ( xj ) + T `is′ik ( xi ) = 0, so ∂iT `j = −T `i N∑ k=1 s′ikT kj . We substitute this expression into the left hand side of (2.3) and obtain the desired result after a routine computation. 3 Review of the Stäckel procedure for the Helmholtz or Schrödinger equation We can perform an analogous construction of eigenfunctions for a Helmholtz operator, using the Stäckel matrix S. We demand eigenfunctions of H in the separated form Ψ = N∏ j=1 Ψ(j)(xj) and depending on the maximal number of parameters. Then, the separation equations are of the form ∂2`Ψ + f`(x `)∂`Ψ + v`(x`)− N∑ j=1 s`j ( x` ) λj Ψ = 0, ` = 1, . . . , N (3.1) for suitable functions f`, v` to be determined. Thus we have the eigenvalue equations LkΨ ≡ N∑ `=1 T k` ( ∂2` + f`∂` + v` ) Ψ = λkΨ, k = 1, . . . , N, where L1 = H. Based on our calculations of the preceding section, we can establish the com- mutation relations [Ls, Lt] = 0. More generally we can consider R-separation for a general Helmholtz equation. In local coor- dinates zj on an N -dimensional pseudo-Riemannian manifold this equation takes the invariant form HΘ ≡ ∆N + N∑ j=1 F j∂j + V Θ = EΘ, (3.2) Nonregular Separation of Variables 5 where ∆N ≡ 1 √ g N∑ j,k=1 ∂j ( gjk √ g∂k ) is the Laplace–Beltrami operator. We say that this equation is R-separable in local orthogonal coordinates xj if there is a fixed nonzero function R(x) such that (3.2) admits solutions Θ = exp(R)Ψ = exp(R) N∏ j=1 Ψ(j) ( xj ) , where Ψ is a regular separated solution, i.e., it satisfies the separation equations (3.1). In this case the symmetry operators are L̃k = exp(R)Lk exp(−R) and equations (3.1) become ∂2`Θ + ( f` ( x` ) − 2∂`R ) ∂`Θ + v`(x`)− ∂``R+ (∂`R)2 − N∑ j=1 s`j ( x` ) λj Θ = 0, ` = 1, . . . , N . Then we have [L̃s, L̃t] = 0, where L̃1 = H. Now consider the case H̃Θ ≡ (∆N + V )Θ = EΘ, (3.3) i.e., the case where there is no magnetic field, and V is real. We can define an inner product on the space of C∞ real valued functions f (1)(z), f (2)(z) with compact support in RN , with respect to which H̃ is formally self-adjoint 〈f (1), f (2)〉 = ∫ RN f (1)(z)f (2)(z) √ g(z) dz. If L = N∑ j,k=1 ajk(z)∂2jk + N∑̀ =1 h`(z)∂` +W (z) is a real symmetry operator then it can be uniquely decomposed as L = L(1) + L(2) where L(1) is formally self-adjoint and L(2) is formally skew- adjoint L(1) = 1 √ g N∑ j,k=1 ∂j ( ajk √ g∂k ) + W̃ , L(2) = 1 √ g N∑ `=1 ( h̃`∂` + 1 2 ∂`h̃ ) . Moreover, both L(1), L(2) are symmetry operators. Note that we have L(2) = 0 unless H̃ admits a first-order symmetry operator. Suppose the system admits N algebraically independent commuting symmetry operators L̃s such that the coefficients (ajks ) of the second-order terms in the symmetries admit a basis of common eigenforms. (Without loss of generality we can restrict to the self-adjoint case L̃s = L̃ (1) s .) Then it is well established [19], from the examination of the third-order terms in the relations [L̃s, L̃t] = 0 that there is an orthogonal coordinate system xj and corresponding Stäckel matrix ( sjk(x j) ) such that H̃ = N∑ j=1 1 h ∂j ( hH−2j ∂j ) + V, h = H1H2 · · ·HN . 6 P. Broadbridge, C.M. Chanu and W. Miller Jr. Since the metric is in Stäckel form, it is straightforward to verify that the symmetries can be rewritten as L̃s = N∑ j=1 T sj ( ∂2j + ( ∂j h S ) ∂j +vj ) , making it evident that the first derivative terms are a gradient. (Here, S is the determinant of the Stäckel matrix.) Thus via an R-transform we can express our system in the form H = N∑ j=1 H−2j ∂2j + V̂ . Then the third derivative terms in the commutation relations [Ls, Lt] = 0 are unchanged and the cancellation of second derivative terms is satisfied identically. The first derivative terms just tell us that the transformed potential V̂ is a Stäckel multiplier, so that it permits separation in the coordinates xj . The zero-th order relation is satisfied identically. Thus the integrable system (3.3) is R-separable. Under the same assumptions, but with a magnetic term added, this is no longer necessarily true. It is easy to see that if there is a function G such that F j = ∂jG, i.e., if the magnetic potential is a gradient, then the system is again R-separable. However, if the magnetic potential is not a gradient then it is no longer necessarily true that integrability implies R-separability, even though the second-order terms in the Laplacian admit a common basis of eigenforms. See [4] for some examples. 4 A generalization of Stäckel form We define a N ×N generalized Stäckel matrix by S =  s11 ( x1 ) s12 ( x1 ) · · · s1,N−1 ( x1 ) a1(x) s21 ( x2 ) s22 ( x2 ) · · · s2,N−1 ( x2 ) a2(x) · · · · · · · · · · · · · · · sN1 ( xN ) sN2 ( xN ) · · · sN,N−1 ( xN ) aN (x)  . (4.1) where the ai are arbitrary analytic functions of the variables x1, . . . , xN . We require that S is a nonsingular matrix. Set T = S−1. Now we assume existence of separation equations in the form u2i + vi ( xi ) + N−1∑ ξ=1 siξ ( xi ) λξ = 0, i = 1, . . . , N, λ1 = −E. (4.2) (Here Latin indices take values 1, . . . , N and Greek indices take values 1, . . . , N − 1.) Note that the term with λN is missing. Thus the general separated solution u will depend on N parameters (rather than N + 1), an example of (maximal) nonregular separation. Note that equations (4.2) can be considered as the restriction to the case λN = 0 of u2i + vi ( xi ) + N−1∑ ξ=1 siξ ( xi ) λξ + λNai(x) = 0, which are not separated for λN = 0. However, any solution of them is a solution of the N equations L` ≡ N∑ j=1 T `j ( u2j + vj ) = λ`, 1 ≤ ` ≤ N, LN = 0, (4.3) Nonregular Separation of Variables 7 where H = L1 is the Hamiltonian. Hence, by solving (4.2) we get a separated solution of H = E satisfying also LN = 0 as a side condition. This construction shows that separation with a generalized Stäckel matrix means (nonregular) separation with a side condition. In this case, functions (4.3) become our “restricted constants of the motion” Lα = λα (modulo LN = 0). Since N∑ j=1 T `jsjk = δ`k, differentiating this identity with respect to xi, gives N∑ j=1 ∂iT `j sjξ(xj) + T `i s′iξ ( xi ) = 0, and N∑ j=1 ( ∂iT `j sjN + T `j∂iaj ) = 0, so ∂iT `j + T `i N−1∑ ξ=1 s′iξT ξj + TNj N∑ h=1 T `h∂iah = 0. (4.4) Using this result it is a straightforward computation to verify the Poisson bracket relations: { Li,Lj } =  N∑ k,h=1 ( T ikT jh − T jkT ih )∂ak ∂xh ph LN . (4.5) Relations (4.5) can be considered as the consistency conditions that guarantee the Li are con- stants of the motion for the Hamiltonian H, modulo the side condition LN = 0. We have verified the relations{ Li,Lj } LN=0 = 0, i, j = 1, . . . , N for L1 = H and linearly independent quadratic forms Li. Thus our construction has shown that separation with a generalized Stäckel matrix implies the existence of N independent constants of motion in involution (modulo the side condition) diagonalised in the separable coordinates. We can generalize Eisenhart’s treatment of Stäckel form in which he represented the quadratic forms T `i in terms of their eigenvalues with respect to the metric T 1j = H−2j : T `j = ρ (`) j H−2j . Here, ρ (1) j = 1. Then (4.4) can be rewritten as a system of partial differential equations for the ρ (`) j : ∂iρ (`) j + ρ (`) j ∂iH −2 j H−2j + ρ (`) i H−2i N−1∑ ξ=1 s′iξρ (ξ) j + ρ (N) j N∑ h=1 ρ (`) h H−2h ∂iah = 0. (4.6) In the special case ` = 1 these equations reduce to ∂iH −2 j H−2j +H−2i N−1∑ ξ=1 s′iξρ (ξ) j + ρ (N) j N∑ h=1 H−2h ∂iah = 0. 8 P. Broadbridge, C.M. Chanu and W. Miller Jr. Substituting this result in (4.6) we obtain ∂iρ (`) j + ( ρ (`) j − ρ (`) i )∂iH−2j H−2j + ρ (N) j N∑ h=1 ( ρ (`) h − ρ (`) i ) H−2h ∂iah = 0. (4.7) Note that if ∂iah = 0 for i 6= h then we recover Stäckel form and (4.7) simplifies to Eisenhart’s equation [12, 14] ∂iρ (`) j + ( ρ (`) j − ρ (`) i )∂iH−2j H−2j = 0. A way of expressing the identity (4.7) that does not require the introduction of the terms ah is to note that at least one of the ρ (N) j , must be nonzero, say for j = 1. Setting j = 1 in (4.7) we obtain N∑ h=1 ( ρ (`) h − ρ (`) i ) H−2h ∂iah = − 1 ρ (N) 1 ( ∂iρ (`) 1 + ( ρ (`) 1 − ρ (`) i )∂iH−21 H−21 ) . Substituting this result back into (4.7) we conclude that ∂iρ (`) j + ( ρ (`) j − ρ (`) i )∂iH−2j H−2j = ρ (N) j ρ (N) 1 ( ∂iρ (`) 1 + ( ρ (`) 1 − ρ (`) i )∂iH−21 H−21 ) . (4.8) For future use we remark that if we restrict to the case ` = N then, for Bj = ρ (N) j /ρ (N) 1 , (4.8) becomes ∂iBj = (Bi −Bj) ∂iH −2 j H−2j +Bj(1−Bi) ∂iH −2 1 H−21 , i, j = 1, . . . , N. (4.9) Remark 1. There is an equivalence relation obeyed by generalized Stäckel matrices. If S is the matrix (4.1), then for any nonzero function f(x), the generalized Stäckel matrix S′ =  s11 ( x1 ) s12 ( x1 ) · · · s1,N−1 ( x1 ) a1(x)f(x) s21 ( x2 ) s22 ( x2 ) · · · s2,N−1 ( x2 ) a2(x)f(x) · · · · · · · · · · · · · · · sN1 ( xN ) sN2 ( xN ) · · · sN,N−1 ( xN ) aN (x)f(x)  , (4.10) defines exactly the same Hamilton–Jacobi equation, separation equations and side condition as does S. Remark 2. This construction of Hamilton–Jacobi systems with a side condition can easily be extended to construct systems with two or more side conditions. For example, with two side conditions LN−1 = 0, LN = 0, the last two columns of the N × N generalized Stäckel matrix would be arbitrary and the symmetries would be modulo the side conditions:{ Li,Lj } = Ai,jLN +Bi,jLN−1, i, j = 1, . . . , N. In a similar fashion nonregular separability of Helmholtz equations with multiple linear side conditions can be defined. Nonregular Separation of Variables 9 5 Generalized Stäckel form for the Helmholtz equation Now we perform an analogous construction of eigenfunctions for a Helmholtz-like operator, using the same generalized Stäckel matrix S. We want eigenfunctions of the separated form Ψ = N∏ j=1 Ψ(j)(xj). We take separation equations in the form ∂2`Ψ + f` ( x` ) ∂`Ψ + ( v` ( x` ) − N−1∑ α=1 s`α ( x` ) λα ) Ψ = 0, ` = 1, . . . , N. Then we have the eigenvalue equations LβΨ ≡ N∑ `=1 T β` ( ∂2` + f` ∂` + v` ) Ψ = λβΨ, β = 1, . . . , N − 1, and the side condition LNΨ ≡ N∑ `=1 TN` ( ∂2` + f`∂` + v` ) Ψ = 0. We take L1 = H, λ1 = E, so −1 2H is the standard Hamiltonian operator. Let X` = ∂2` + f` ∂` + v`, Y` = ∂2` + f` ∂`. We need to compute the commutator [Lα, Lj ] for α = 1, . . . , N − 1, j = 1, . . . , N . Now LαLj = (∑ i ρ (α) i H−2i Xi )(∑ k ρ (j) k H−2k Xk ) = ∑ i,k ρ (α) i ρ (j) k H−2i H−2k XiXk + ∑ i,k ρ (α) i H−2i Yi ( ρ (j) k H−2k ) Xk + 2 ∑ i,k ρ (α) i H−2i ∂i ( ρ (j) k H−2k ) ∂iXk, LjLα = (∑ k ρ (j) k H−2k Xk )(∑ i ρ (α) i H−2i Xi ) = ∑ i,k ρ (j) k ρ (α) i H−2k H−2i XkXi + ∑ i,k ρ (j) k H−2k Yk ( ρ (α) i H−2i ) Xi + 2 ∑ i,k ρ (j) k H−2k ∂k ( ρ (α) i H−2i ) ∂kXi, so [ Lα, Lj ] = ∑ i,k ( ρ (α) i Yi ( ρ (j) k H−2k ) − ρ(j)i Yi ( ρ (α) k H−2k )) H−2i Xk + 2 ∑ i,k ( ρ (α) i ∂i ( ρ (j) k H−2k ) − ρ(j)i ∂i ( ρ (α) k H−2k )) H−2i ∂iXk. Using (4.7) we can establish the identities( ρ (α) i ∂i ( ρ (j) k H−2k ) − ρ(j)i ∂i ( ρ (α) k H−2k )) H−2i ∂i = ρ (N) k H−2k H−2i N∑ h=1 ( ρ (α) h ρ (j) i − ρ (α) i ρ (j) h ) H−2h (∂iah)∂i, 10 P. Broadbridge, C.M. Chanu and W. Miller Jr. ρ (N) k H−2k ∂iF = − ( ρ (N) i ∂iH −2 k + ρ (N) k H−2k N∑ h=1 ( ρ (N) i − ρ(N) h ) H−2h ∂iah ) F + ∂i ( ρ (N) k H−2k F ) , for any function F , and( ρ (α) i Yi ( ρ (j) k H−2k ) − ρ(j)i Yi ( ρ (α) k H−2k )) H−2i = 2ρ (N) i H−2i ∂iH −2 k N∑ h=1 ( ρ (α) h ρ (j) i − ρ (j) h ρ (α) i ) H−2h ∂iah + ρ (N) k H−2k H−2i ( ρ (j) i ∂i ( N∑ h=1 ρ (α) h H−2h ∂iah ) − ρ(α)i ∂i ( N∑ h=1 ρ (j) h H−2h ∂iah )) + ρ (N) k H−2k H−2i ( N∑ h=1 ( ρ (N) i − ρ(N) h ) H−2h ∂iah )( N∑ h=1 ( ρ (j) i ρ (α) h − ρ (α) i ρ (j) h ) H−2h ∂iah ) . Thus, [ Lα, Lj ] = N∑ i=1 H−2i (( 2 N∑ h=1 ( ρ (j) i ρ (α) h − ρ (α) i ρ (j) h ) H−2h ∂iah ) × ( −1 2 N∑ h=1 ( ρ (N) i − ρ(N) h ) H−2h ∂iah + ∂i ) + ( N∑ h=1 ( ρ (N) i − ρ(N) h ) H−2h ∂iah ) × ( N∑ h=1 ( ρ (j) i ρ (α) h − ρ (α) i ρ (j) h ) H−2h ∂iah )) LN = FαjL N , where Fαj is a first-order partial differential operator. We see that there is no obstruction to lifting our classical nonregular separation to the operator case. A difficulty occurs, however, when we try to write the pure operator part of H as a Laplace–Beltrami operator on a Riemannian manifold. Then there is an obstruction, a gene- ralized Robertson condition, to be worked out. Also we need to examine the effect of permitting R-separation. 6 Maximal nonregular separation as regular separation with a side condition Another way to approach the classical Hamilton–Jacobi problem is to use the Kalnins–Miller method for variable separation [17, 25] and consider maximal nonregular separation as regular separation with a side condition. (Here the nonregular separation is maximal in the sense that with a single side condition the number of separation constants is the maximum possible for nonregular separation, i.e., just 1 less than that for regular separation.) We look for additively separable solutions of the equation N∑ i=1 H−2i u2i + V = E (6.1) with the side condition N∑ i=1 L−2i u2i +W = 0, (6.2) Nonregular Separation of Variables 11 i.e., solutions u such that uij = 0 for i 6= j. From (6.1) we find ujj = − Vj + ∑ i ∂jH −2 i u2i 2H−2j uj , and from (6.2) ujj = − Wj + ∑ i ∂jL −2 i u2i 2L−2j uj . These expressions must be equal modulo the side condition (6.2), so Vj + ∑ i ∂jH −2 i u2i 2H−2j uj = Wj + ∑ i ∂jL −2 i u2i 2L−2j uj + νj (∑ i L−2i u2i +W ) , (6.3) for some functions νj . Similarly, equations derived from ujjk = 0 for j 6= k must hold modulo the side condition. Requiring that all of the above equations hold identically, i.e., requiring that we have regular separation modulo the side condition we eventually obtain the conditions that 1. There are functions ωj` = ω`j for all j 6= ` such that −∂j`L−2i + ∂jL −2 ` L−2` ∂`L −2 i + ∂`L −2 j L−2j ∂jL −2 i + ωj`L −2 i = 0 (6.4) and −Wj` +W` ∂jL −2 ` L−2` +Wj ∂`L −2 j L−2j + ωj`W = 0 (6.5) for all i = 1, . . . , N . (If all the L` are nonzero, this means that the L−2i are in confor- mal Stäckel form, i.e., an arbitrary function times a Stäckel form matrix, and that W is a conformal Stäckel form potential [23, 24].) 2. There are functions τj , j = 1, . . . , N such that ∂jH −2 i H−2j = ∂jL −2 i L−2j + τjL −2 i , (6.6) for all i = 1, . . . , N , and Vj H−2j = Wj L−2j + τjW. (6.7) 3. Let Cij be the second-order differential operator acting on functions f by Cij(f) = ∂ijf − ∂jH −2 i H−2i ∂if − ∂i lnH−2j H−2j ∂jf. (6.8) There are functions µj` = µ`j for all j 6= ` such that Cj` ( H−2i ) = µj`L −2 i , (6.9) and Cj`(V ) = µj`W (6.10) for all j, ` = 1, . . . , N , with j 6= `. 12 P. Broadbridge, C.M. Chanu and W. Miller Jr. 7 Maximal nonregular separation ⇒ generalized Stäckel form We have shown that the first row of the inverse of a generalized Stäckel matrix with arbitrary N -th column is an orthogonal metric H−2i whose associated Hamilton–Jacobi equation admits nonregular separation on the hypersurfaces given by the level set LN = 0 of a function LN , quadratic in momentum variables, which is a constrained first integral on the same level set and whose components are the N -th row of the inverse of the generalized Stäckel matrix. Now we prove the converse, i.e., that if an orthogonal geodesic Hamiltonian is separable in orthogonal coordinates on the level set of a quadratic first integral, then it is a row of the inverse of a gene- ralized metric and the quadratic coefficients of LN are the N -th row of the generalized Stäckel matrix. Our starting point here is the geometrical framework of regular separation of variables: a complete separated solution of the Hamilton–Jacobi equation is a foliation parametrized by N parameters for the integral manifold of the distribution generated by the N vector fields Di = ∂xi +Ri∂yi where the Ri are determined by the condition that the Di are tangent to H = const. The classical Levi-Civita conditions [22] represent the integrability conditions of the distribution. In our case we need this distribution to be integrable only on the submanifold S defined by LN = 0. We also need the vector fields to be tangent to the submanifold S (closely related to the compatibility of the side condition). Our first step will be to write the differential conditions that in this case play the role of the Levi-Civita condition for regular separation. In this case they mix H and LN . Moreover, they include also the condition that Di are tangent to LN = 0. We will show that these equations are exactly the equations for nonregular separation derived in Section 6. Then we will relate these conditions to the existence of some more intrinsic geometrical object for the eigenvalues of Killing tensors (associated with quadratic in the momenta constants of motion), whose integrability conditions are equivalent to the Levi-Civita conditions. Finally, we will construct the family of quadratic “first integrals” Lh = N∑ j=1 T hju2j commuting and constant for the motion on LN = 0 diagonalized in the coordinates we are considering, and show that the inverse of the matrix of the components T hj is a generalized Stäckel matrix. Then we will extend our analysis to prove the corresponding results for multiplicative separation or R-separation of the Helmholtz or time independent Schrödinger equations. 7.1 Differential conditions for nonregular separation on a quadratic first integral leaf As in Section 6 we consider a natural Hamiltonian in orthogonal coordinates x = (xi) on the cotangent bundle of a N -dimensional Riemannian manifold Q with Hamiltonian H = L1 = N∑ i=1 H−2i p2i + V (x), (7.1) and a function LN which is also quadratic in the momenta (pi) and diagonalized in the same coordinates LN = N∑ i=1 ρ (N) i H−2i p2i +W (x), where ρ (N) i are the eigenvalues with respect to the metric H−2i , i.e., Li = ρ (N) i H−2i . We want to study the existence of separated solutions u of the Hamilton–Jacobi equation N∑ i=1 H−2i u2i + V (x) = E, ui = ∂iu, (7.2) Nonregular Separation of Variables 13 with the side condition, or constraint, N∑ i=1 ρ (N) i H−2i u2i +W (x) = 0. Recall that the conditions for nonregular separation are (6.4), (6.5), (6.6), (6.7), (6.9), (6.10) for Li = ρ (N) i H−2i and some functions ω`j , τj , µj`. However, in this case, we can avoid the introduction of additional unknown functions, since we can solve the equation LN = 0 with respect to a momentum variable ui (as for instance u21 = u21(x j , uα) with α = 2, . . . , N) so an expression vanishes on LN = 0 if and only if it vanishes for all uα for α = 2, . . . , N after the substitution of u1 by u1(x j , uα). This simplifies very much the task of finding equivalent conditions such as the link with generalized Stäckel matrices and it is possible only because we are assuming orthogonal coordinates. The conditions (6.4), (6.5), (6.6), (6.7), (6.9), (6.10) are equivalent (supposing without loss of generality that ρ (N) 1 6= 0) to imposing that for u21 = − W ρ (N) 1 H−21 − N∑ α=2 ρ (N) α H−2α ρ (N) 1 H−21 u2α (7.3) the expressions (6.4)–(6.10) vanish for all values of u2α. Even easier, inserting (7.3) in (6.3) we find that the coefficient of νj vanishes and equating coefficients of u2α we get ∂jρ (N) α H−2α ρ (N) α H−2α − ρ (N) j ρ (N) α ∂jH −2 α H−2α = ∂jρ (N) 1 H−21 ρ (N) 1 H−21 − ρ (N) j ρ (N) 1 ∂jH −2 1 H−21 = ∂jW W − ρ (N) j W ∂jV. (7.4) Similarly we find Cij ( H−2α ) ρ (N) α H−2α = Cij ( H−21 ) ρ (N) 1 H−21 = Cij(V ) W . (7.5) In all these equations we follow the convention that the vanishing of a factor ρ (N) α or W in a denominator in one of expressions (7.4), (7.5), implies that the numerator vanishes. Note that (7.5) and (7.4) give conditions for separation of the geodesic Hamiltonian with V = W = 0, with additional conditions that V and W must satisfy. Supposing for simplicity for the moment V = W = 0, we can rewrite (7.5) and (7.4) as Cij ( H−2α ) H−2α = ρ (N) α ρ (N) 1 Cij ( H−21 ) H−21 , (7.6) ∂i ρ (N) α ρ (N) 1 = ( ρ (N) i ρ (N) 1 − ρ (N) α ρ (N) 1 ) ∂iH −2 α H−2α + ρ (N) α ρ (N) 1 ( 1− ρ (N) i ρ (N) 1 ) ∂iH −2 1 H−21 , (7.7) which are the necessary and sufficient conditions. Equations (7.7) can be interpreted as a first-order system in the N − 1 unknowns Bα = ρ (N) α ρ (N) 1 . Proposition 1. A geodesic Hamiltonian H admits nonregular separation on the submanifold LN = 0 in a given orthogonal coordinate system if and only if the functions Bk = ρ (N) k ρ (N) 1 (k = 1, . . . , N) satisfy ∂iBk = (Bi −Bk) ∂iH −2 k H−2k +Bk(1−Bi) ∂iH −2 1 H−21 . (7.8) 14 P. Broadbridge, C.M. Chanu and W. Miller Jr. Proof. By rewriting the necessary and sufficient conditions (7.6) and (7.7) in terms of the N functions Bk = B1, Bα with B1 = 1 we get Cij ( H−2k ) H−2k = Bk Cij ( H−21 ) H−21 (7.9) and (7.8). However, it is a straightforward calculation that (7.9) is a differential consequence of (7.8), indeed we have ∂i∂jBk − ∂j∂iBk = (Bi −Bj) ( Cij ( H−2k ) H−2k −Bk Cij ( H−21 ) H−21 ) . � (7.10) Note that the system of equations (7.8) coincides with (4.9), verifying again that separation with a generalized Stäckel matrix is nonregular. Proposition 2. A natural Hamiltonian H admits nonregular separation on the submanifold LN = 0 in a given orthogonal coordinate system only if the ratios Cij(H −2 α )H−21 Cij(H −2 1 )H−2α are independent of i and j and the eigenvalues of the quadratic function LN are proportional to them Bα = ρ (N) α ρ (N) 1 = Cij ( H−2α ) H−21 Cij ( H−21 ) H−2α . (7.11) Remark 3. The function LN is naturally defined up to a multiplicative factor f on Q: if an expression is zero on LN = 0 then it is also zero on fLN = 0 for all functions on T ∗Q, but, since we are interested on quadratic in the momenta functions, we can normalize f on Q. (From the Stäckel matrix point of view this corresponds to the multiplication of the N -th column by f . Hence, equations (7.9) determine the unique (up to a factor) quadratic hypersurface where separation could occur. Indeed, equations (7.10) are the complete integrability conditions for the first-order PDE system (7.8). These conditions are identically satisfied for all (Bk, x i) on an open subset of C2n only if Cij(H −2 k ) = 0, that is only if regular separation occurs. However, a single solution Bk = Bk(x i) could exist, provided it satisfies the original conditions (7.8), that is if it takes the form (7.11), or with all Bk equal, that is for LN = H. In this case we get separation of the null equation H = 0 (in which case a column of the Stäckel matrix was arbitrary, so the result is consistent with the known one). Let us examine the relation of this kind of separation with the generalized Stäckel matrix. Theorem 1. Suppose the natural Hamiltonian (7.1) admits N−2 other functions L2, . . . ,LN−1, quadratic in the momenta such that 1) L1 = H,L2, . . . ,LN−1,LN are pointwise independent, 2) L1 = H,L2, . . . ,LN−1,LN are constants of the motion, modulo LN , that is{ H,Lk } \|LN=0 = 0, 3) the quadratic terms of L1 = H,L2, . . . ,LN−1,LN are diagonal in the coordinates (xi), that is L` = N∑ i=1 ρ (`) i H−2i ( u2i + vi(x) ) , ρ (1) i = 1 ∀ i. Nonregular Separation of Variables 15 Then, 1) ∂jvi = 0 for j 6= i and the eigenvalues ρ (`) i of L` satisfy the following generalization of the Eisenhart conditions ∂iρ (`) α + ( ρ(`)α − ρ (`) i )∂iH−2α H−2α = ρ (N) α ρ (N) 1 ( ∂iρ (`) 1 + ( ρ (`) 1 − ρ (`) i )∂iH−21 H−21 ) ; (7.12) 2) the inverse matrix of T `i = ( ρ (`) i H−2i ) is a generalized Stäckel matrix with the last column made of arbitrary functions of N variables a1(x), . . . , an(x); 3) the Hamilton Jacobi equation (7.2) admits a maximal nonregular separated solution on LN = 0 depending on N − 1 parameters; 4) the N functions L1 = H,L2, . . . ,LN−1,LN are in involution on S, that is{ L`,Lk } \|LN=0. Proof. 1) By inserting (7.3) in the Poisson brackets { L`,H } = ∑ i 2H−2i ui ( N∑ k=1 ( ∂i ( ρ (`) k H−2k ) − ρ(`)i ∂iH −2 k ) u2k + N∑ k=1 ( ∂i ( ρ (`) k H−2k vk ) − ρ(`)i ∂i ( H−2k vk ))) , we get that ∂ivk = 0 for i 6= k and that ρ (`) k satisfies the generalized Eisenhart conditions (7.12). 2) Since the quadratic diagonal functions are pointwise independent, the matrix T `i has a nonzero determinant and therefore it admits an inverse skh such that T `kskh = δ`h (7.13) (we partially follow the proof of equation (4.7) and a proof of the Eisenhart theorem given in [11]). We want to show that ∂ishξ = 0 for all i 6= h and ξ 6= N if and only if ρ (`) j = T `j/T 1j satisfy the generalized Eisenhart conditions (7.12). Differentiating (7.13) with respect to xi we get ∑ k ( ∂iT `kskh + T `k∂iskh ) = 0, that is∑ k (( H−2k ∂iρ (`) k + ρ (`) k ∂iH −2 k ) skh + ρ (`) k H−2k ∂iskh ) = 0, ` 6= 1, (7.14)∑ k ( ∂iH −2 k skh +H−2k ∂iskh ) = 0, ` = 1. (7.15) Adding (7.15) multiplied by −ρ(`)i to equation (7.14) we get∑ k (( H−2k ∂iρ (`) k + ( ρ (`) k − ρ ` i ) ∂iH −2 k ) skh + ( ρ (`) k − ρ (`) i ) H−2k ∂iskh ) = 0, ` 6= 1. (7.16) 16 P. Broadbridge, C.M. Chanu and W. Miller Jr. Let us suppose that ∂iskh = 0 for all i 6= k and h 6= N and ∂iskN = ∂iah, by multiplying (7.16) by T hj , we get∑ k ( H−2k ∂iρ (`) k + ( ρ (`) k − ρ (`) i ) ∂iH −2 k ) δkj + ∑ h ∑ k (( ρ (`) k − ρ (`) i ) H−2k ∂iskh ) T hj = 0, that is H−2j ∂iρ (`) j + ( ρ (`) j − ρ (`) i ) ∂iH −2 j + TNj ∑ k (( ρ (`) k − ρ (`) i ) H−2k ∂iak ) = 0. (7.17) Writing (7.17) for j = 1 and j = α and by combining them in order to eliminate the common factor ∑ k (( ρ (`) k −ρ (`) i ) H−2k ∂iak ) , we get (7.12). Conversely, let us suppose that (7.12) hold; this means that the quantities H−2k ∂iρ (`) k + ( ρ (`) k − ρ (`) k ) ∂iH −2 k TNk are independent of k, that is H−2k ∂iρ (`) k + ( ρ (`) k −ρ (`) k ) ∂iH −2 k = TNkQ`i . Therefore (7.16) becomes TNkQ`iskh + ∑ k (( ρ (`) k − ρ (`) i ) H−2k ∂iskh ) = δNh Q ` i + ∑ k (( ρ (`) k − ρ (`) i ) H−2k ∂iskh ) = 0. If we fix the values of h and i, then the above equations are a linear system of N − 1 equations (` = 2, . . . , N) in N − 1 unknown ∂iskh k = 1, . . . , N , k 6= h). For h 6= N the system is homogeneous and the matrix of coefficient M `k = ( ρ (`) k − ρ (`) i ) H−2k has nonzero determinant (see [11]). Hence, the only solution is ∂iskh = 0 for h 6= N . 3) If equations (7.12) hold, they hold in particular for ` = N and we get (7.7) which are proven to be equivalent to conditions (7.8) that are necessary and sufficient for the nonregular separation with the side condition LN = 0. Further, the conditions (7.4) on V , W are exactly the consistency conditions that must be satisfied by V , W in order that {H,LN} = 0 hold on LN = 0, and conditions (7.5) are exactly the integrability conditions for V , W that must be satisfied if {H,LN} = 0 is to hold. 4) This is essentially the computation in Section 4. � Theorem 2. If the N pointwise independent functions L1 = H,L2, . . . ,LN−1,LN are in involution on LN = 0 and admit common eigenvectors, i.e., if the quadratic forms in the momenta associated with L`−ρ(`)H have N common eigenvectors, then the eigenvectors are normal: there exist orthogonal coordinates (xi) such that the N functions are simultaneously diagonalized. Remark 4. This is essentially the converse of the results of Section 4. It shows that separation with a side condition implies the existence of a generalized Stäckel matrix that defines the separation. Proof. Apply Theorem 7.5 of [2] with Mab = KN , where KN is the symmetric 2-tensor asso- ciated with LN . � Remark 5. The functions L` are defined up to multiples of LN and the conformal Killing tensors are defined up to multiple of the metric tensor. Indeed, L` + f `LN have the same eigenvalues of L` (provided that the L` have common eigenvectors) and they are in involution when restricted to LN = 0:{ L` + f `LN ,Lk + fkLN } = QLN . Nonregular Separation of Variables 17 Now we close the logical loop. Suppose the conditions (7.8)–(7.10) of maximal nonregular separation are satisfied and consider the linear system of equations ∂iρj + (ρj − ρi) ∂iH −2 j H−2j = Bj ( ∂iρ1 + (ρ1 − ρi) ∂iH −2 1 H−21 ) , i, j = 1, . . . , N, (7.18) for the unknowns (ρ1, . . . , ρN ). Remark 5 suggests that this system has solutions for any choice of ρ1. Therefore, we choose the function ρ1 arbitrarily and consider (7.18) as a system of N(N−1) independent differential equations for the unknowns (ρ2, . . . , ρN ). It is then a straight- forward exercise to verify that the integrability conditions for (7.18) are satisfied identically, due to the nonregular separation conditions. We already know the two solutions (1, 1, . . . , 1) and (B2, . . . , BN ). Indeed about any regular point x0 we can find a unique solution such that ρj(x0) = βj , 2 ≤ j ≤ N for any choice of constants βj . For example we could take ρ1 ≡ 1 and define N − 1 linearly independent solutions by choosing the βj accordingly. Then we could take ρ1 ≡ 0 and find another solution such that the full N solutions form a basis. This basis determines a generalized Stäckel matrix. It remains to consider the potential terms, the given functions V and W . Using the func- tions ρ (`) i of the generalized Stäckel matrix computed above we define functions L` = N∑ i=1 ρ (`) i H−2i u2i +W (`), ` = 1, . . . , N − 1, where V = W (1) but the remaining W (`) will be determined by imposing the requirement that there exist functions Fj linear in the momenta and such that{ H,LN } = FNLN , { H,L` } = F`LN . By working out the Poisson brackets, it is straightforward to verify that equations (7.4) are exactly the consistency conditions for the existence of FN and (7.5) are exactly the consistency conditions for the F` and the integrability conditions for the existence of the potentials W (`). Now note that since the generalized Stäckel matrix is invertible we can uniquely determine functions vi(x) such that W (j) = N∑ i=1 viρ (j) i H−2i , j = 1, . . . , N. With this observation we have verified all of the assumptions of Theorem 1. This proves the following result. Theorem 3. If a natural Hamiltonian H admits maximal nonregular separation on the sub- manifold LN = 0 in a given orthogonal coordinate system, then the system is separable with a side condition and there exists a generalized Stäckel matrix for the separation. All of these results for additive nonregular separation of Hamilton–Jacobi equations extend to multiplicative nonregular separation for the Helmholtz (or Schrödinger) equations on the same manifold, but with an obstruction. If a system is nonregular R-separable for the Helmholtz equation then it is nonregular separable for the Hamilton–Jacobi equation. However, if the Hamilton–Jacobi equation admits nonregular separation in some coordinate system then there is a “generalized Robertson condition” to be solved to determine for which potentials the Helmholtz equation admits R-separation in these coordinates. On the other hand, we can always find some family of vector potentials for which the Helmholtz equation does admit nonregular R-separation in the coordinates. That is, it is easy to show that given a nonregular separable system for the Hamilton–Jacobi equation, we can construct a family of vector potentials for which the corresponding Helmholtz equation is nonregular R-separable. We give examples in Section 8. 18 P. Broadbridge, C.M. Chanu and W. Miller Jr. 8 Examples of nonregular separability for Hamilton–Jacobi and Helmholtz equations 8.1 Examples of restricted regular separation The simplest examples of Theorems 2 and 3, and their extensions to the Helmholtz equation, are those for which regular orthogonal separation already occurs. Thus the “generalized” Stäckel matrix (4.1) is a true Stäckel matrix. Consider first the separable Hamilton–Jacobi equation in the form H ≡ N∑ i=1 H−2i u2i + V (x) = E. Here, the metric in the orthogonal separable coordinates xi is ds2 = N∑ i=1 H2 i ( dxi )2 . There is an associated Stäckel matrix and constants of the motion L`, 2 ≤ ` ≤ N , where H = L1. We apply the side condition LN ′ ≡ LN − λN = 0 where λN is a constant scalar. Note that LN ′ is also a constant of the motion. Now the separation equations become u2i + ( vi ( xi ) + λNsiN ( xi )) + N−1∑ j=1 sij ( xi ) λj = 0, i = 1, . . . , N, λ1 = −E. Our “restricted constants of the motion” become L`′ ≡ N∑ j=1 T `j ( u2j + vj + λNsjN ) , so we have modified the potential. The side conditions are satisfied automatically. Effectively, we have restricted our Hamiltonian system to the N−1 dimensional hypersurface LN = λN and seen that the result is again a Hamiltonian system. Under some circumstances this has a simple interpretation as a separable system on a manifold of one less dimension. Example 1. Consider the Kepler Hamiltonian H = L1 = p2x + p2y + p2z + α r , r = √ x2 + y2 + z2 in real Euclidean space. The system is separable in spherical coordinates r, θ, φ where x = r sin θ cosφ, y = r sin θ sinφ, z = r cos θ, r ≥ 0, 0 ≤ θ ≤ π, 0 ≤ φ < 2π. Here H = p2r + L2 r2 + α r , L2 = p2θ + L3 sin2 θ , L3 = p2φ. We choose the side condition L3 = λ3. Then the reduced Hamiltonian becomes H′ = p2r + p2θ r2 + λ3 r2 sin2 θ + α r . Nonregular Separation of Variables 19 In terms of the new variables X = r cos θ, Y = r sin θ this becomes H′ = p2X + p2Y + α√ X2 + Y 2 + λ3 Y 2 , a regular separable system in the first quadrant of the Euclidean plane. In the special case λ3 = 0 the singularity on the X-axis disappears and we can extend the system to the full punctured plane. Note that this construction is conceptually distinct from simply restricting a trajectory confined to a plane. A similar construction works for the Helmholtz equation. We want eigenfunctions of the separated form Ψ = N∏ j=1 Ψ(j)(xj). We take separation equations in the form ∂2`Ψ + f` ( x` ) ∂`Ψ + ( v` ( x` ) − s`N ( x` ) λN − N−1∑ α=1 s`α ( x` ) λα ) Ψ = 0, ` = 1, . . . , N. Then we have the restricted eigenvalue equations Lβ ′ Ψ ≡ N∑ `=1 T β` ( ∂2` + f`∂` + v` − s`NλN ) Ψ = λβΨ, β = 1, . . . , N − 1, and the side condition LN ′ Ψ ≡ N∑ `=1 TN` ( ∂2` + f`∂` + v` − s`NλN ) Ψ = 0. We take L1′ = H ′. Effectively, we have restricted our separable quantum system to an eigenspace {Ψ : LNΨ = λNΨ} of a symmetry operator. The restriction is again a Hamiltonian system. Again, under some circumstances this has a simple interpretation as a separable quantum system on a manifold of one less dimension. Example 2. Consider the hydrogen atom Hamiltonian, a constant multiple of H = L1 = ∂2x + ∂2y + ∂2z + α r , r = √ x2 + y2 + z2 in real Euclidean space. The system is separable in spherical coordinates r, θ, φ: H = ∂2r + 2 r ∂r + 1 r2 L2 + α r , L2 = ∂2θ + cot θ∂θ + 1 sin2 θ L3, L3 = ∂2φ. We choose the side condition L3Ψ = λ3Ψ. Then the reduced Hamiltonian becomes H ′ = ∂2r + 2 r ∂r + 1 r2 ∂2θ + cot θ r2 ∂θ + λ3 r2 sin2 θ + α r . Now we make an R-transformation Ψ = RΘ where R = 1 r √ sin θ . Then the eigenvalue equation H ′Ψ = EΨ becomes( ∂2r + 1 r2 ∂2θ + λ3 + 1 4 r2 sin2 θ + 1 4 r2 + α r ) Θ = EΘ, ∂2φΘ = λ3Θ. 20 P. Broadbridge, C.M. Chanu and W. Miller Jr. Now write Θ = Φ(φ)Ξ(r, θ). In terms of the new variables X = r cos θ, Y = r sin θ the energy equation is( ∂2X + ∂2Y + α√ X2 + Y 2 + 1 4 X2 + Y 2 + λ3 + 1 4 Y 2 ) Ξ = EΞ, a regular separable system in the upper half Euclidean plane. If λ3 = −1/4 this equation can be extended to the punctured plane and regarded as a modification of the hydrogen atom in the plane. Note, however, that due to the R factor, the usual L2 normalization for eigenstates in Euclidean space doesn’t restrict to the usual L2 normalization for eigenstates in the plane. Thus, in analogy with the virial theorem for the mapping between the Coulomb and pseudo- Coulomb problems [36], one needs to check that the bound state spectra are preserved under the restriction. 8.2 Nonregular separation in 2D and a “no go” theorem Now we take up the issue of true nonregular separation in a coordinate system in which regular separation is impossible. Let us first look at 2D examples, a very special case. By using the facts that we can always replace a separable coordinate by an invertible function of itself, and we can perform linear transformations on the separation constants without changing the system, we can always assume that the generalized Stäckel matrix looks like S̃ = ( 1 A(u, v) 1 B(u, v) ) . We must require that A − B 6= 0 so that the matrix is invertible and that AB 6= 0 so that the metric is nondegenerate. Further, by making use of the equivalence relation (4.10) we can put the matrix in canonical form S = ( 1 1 1 f(u, v) ) , with inverse T = 1 f − 1 ( f −1 −1 1 ) , where f = B/A. This form will lead to true nonregular separation unless f can be factored as f(u, v) = U(u)V (v), because in that case we can take B = V (v), A = 1/U(u) and find an equivalent true Stäckel matrix. It is easy to see that this form leads to true nonregular separation for the Hamilton–Jacobi equation. However, in the 2D case the construction fails for Helmholtz equations. Theorem 4. For a 2D manifold the Helmholtz equation ∆Ψ = λΨ never admits true nonregular separation. Proof. In the 2D case, the Laplacian takes the form ∆ = f f − 1 ( ∂uu + 1 2 fu f ∂u ) − 1 f − 1 ( ∂vv − 1 2 fv f ∂v ) . For nonregular separation it is necessary that fu/f is a function of u alone and fv/f is a function of v alone. Thus f = U(u)V (v) and the system admits regular separation. Similarly if Ψ = eRΘ for some fixed R, then R-separation implies that there exist functions U(u), V (v) such that Ru = −1 4 fu f + U(u), Rv = 1 4 fv f + V (v). Since ∂vRu = ∂uRv it follows easily that ∂uv ln f = 0. Thus f = Û(u)V̂ (v) and the system must admit regular R-separation. � Nonregular Separation of Variables 21 However, true nonregular separation may occur when vector potentials are included, even in the 2D case. The self-adjoint Schrödinger equation for a non-relativistic charged particle influenced by a classical electromagnetic 4-potential, still does not admit true nonregular sepa- ration. Nonetheless true nonregular separation does occur when the vector potential gives rise to skew-adjoint first-order terms, as in a diffusion-convection equation for transport of solute. 8.3 Quantum particle in R2 with a magnetic field In the Schrödinger picture with Cartesian coordinates, the eigenvalue equation for the quantized Hamiltonian of a non-relativistic charged particle interacting with a classical electromagnetic field, is (e.g. [4]) HΨ = −(~)2 2 ∇2Ψ− i~ 2 [A.∇+∇.A(x)]Ψ + Φ(x)Ψ = λΨ. (A1, A2, 0) is the vector potential for the magnetic field and Φ is the scalar potential for the electric field. We consider the equivalent eigenvalue problem HΨ = ∂21Ψ + ∂22Ψ + i 2 [ a1(x)∂1 + a2(x)∂2 +∇.a(x) ] Ψ + Φ(x)Ψ = EΨ. (8.1) We extend this equation to a general curvilinear orthogonal coordinate system. As before, ∆ = [ H−21 ∂21 +H−22 ∂22 +H−21 ∂1 log H2 H1 ∂1 −H−22 ∂2 log H2 H1 ∂2 ] . In a general orthogonal coordinate system, the scalar divergence, involving Christoffel symbol Γ, of vector a, is div a = ∂am ∂xm + amΓiim = 1 √ g ∂ ∂xi ( ai √ g ) = ∂ia i + a1[∂1 logH1 + ∂1 logH2] + a2[∂2 logH1 + ∂2 logH2]. The eigenvalue equation for the Hamiltonian is HΨ = H−21 ∂21Ψ +H−22 ∂22Ψ + [ H−21 ∂1 log H2 H1 ] ∂1Ψ− [ H−22 ∂2 log H2 H1 ] ∂2Ψ + i 2 [ a1(x)∂1 + ∂1a 1(x) + a2(x)∂2 + ∂2a 2(x) ] Ψ + i 2 [ a1∂1 log(H1) + a1∂1 log(H2) + a2∂2 log(H1) + a2∂2 log(H2) ] Ψ + Φ(x)Ψ = EΨ. Consider R-separation Θ = Ψe−R(x) = Θ(1) ( x1 ) Θ(2) ( x2 ) . Then [ H−21 ∂21 +H−22 ∂22 +H−21 { ∂1 log H2 H1 + 2∂1R } ∂1 +H−22 { −∂2 log H2 H1 + 2∂2R } ∂2 ] Θ +H−21 [ ∂21R+ (∂1R)2 + (∂1R)∂1 log H2 H1 ] Θ +H−22 [ ∂22R+ (∂2R)2 − (∂2R)∂1 log H2 H1 ] Θ + Φ(x)Θ 22 P. Broadbridge, C.M. Chanu and W. Miller Jr. + i 2 [ a1∂1Θ + a2∂2Θ ] + i 2 [ a1∂1R+ a2∂2R+ ∂1a 1 + ∂2a 2 ] Θ + i 2 a1[∂1 logH1 + ∂1 logH2]Θ + i 2 a2[∂2 logH1 + ∂2 logH2]Θ = EΘ. Assume the separation equations ∂21Θ + [f1(x 1) + ig1(x 1)]∂1Θ + [ [v1(x 1) + iw1(x 1)]− S11(x1)E ] Θ = 0, ∂22Θ + [f2(x 2) + ig2(x 2)]∂2Θ + [ [v2(x 2) + iw2(x 2)]− S21(x2)E ] Θ = 0, ⇒ T11∂ 2 1Θ + T12∂ 2 2Θ + T11[f1 + ig1]∂1Θ + T12[f2 + ig2]∂2Θ + [T11[v1 + iw1] + T12[v2 + iw2]] Θ = EΘ, T21∂ 2 1Θ + T22∂ 2 2Θ + T21[f1 + ig1]∂1Θ + T22[f2 + ig2]∂2Θ + [T21[v1 + iw1] + T22[v2 + iw2]] Θ = 0, where matrix T is the inverse of generalized Stäckel matrix S. In order for the above eigenvalue equation to be identified with the Schrödinger equation, we again require T11 = H−21 and T12 = H−22 , and in addition, f1 ( x1 ) = ∂1 log H2 H1 + 2∂1R, f2 ( x2 ) = −∂2 log H2 H1 + 2∂2R, (8.2) T11v1 + T12v2 = Φ +H−21 ∂21R+H−22 ∂22R+H−21 (∂1R)2 +H−22 (∂2R)2 +H−21 (∂1R)∂1 log H2 H1 −H−22 (∂2R)∂2 log H2 H1 , g1 ( x1 ) = 1 2 H2 1a 1, g2 ( x2 ) = 1 2 H2 2a 2, T11w1 + T12w2 = 1 2 [ a1∂1R+ a2∂2R+ ∂1a 1 + ∂2a 2 ] + 1 2 a1[∂1 logH1 + ∂1 logH2] + 1 2 a2[∂2 logH1 + ∂2 logH2]. (8.3) From (8.2) we can easily deduce that ∂1∂2R = 0 and ∂1∂2 log(H2/H1) = 0. Hence H2/H1 = Π1 ( x1 ) Π2 ( x2 ) and R = R1 ( x1 ) +R2 ( x2 ) (8.4) for some functions Π1, Π2, R1, R2. Now assuming the canonical form with H1 = 1, we have a1 = 2g1 ( x1 ) and a2 = 2g2 ( x2 ) Π2 ( x2 )−2 Π1 ( x1 )−2 , whence it follows by separation of variables in (8.3), that Π2 1[w1 − g1∂1R1 − ∂1g1 − g1∂1 log Π1] = γ4, Π−22 [w2 − g2∂2R2 − ∂2g2 + g2∂2 log Π2] = −γ4, where γ4 is constant. By integration, a1 = 2Π−11 e−R1 [ γ5 + ∫ eR1 [ w1Π1 − γ4Π−11 ] dx1 ] , a2 = 2Π−21 Π−12 e−R2 [ γ6 + ∫ eR2 [ w2Π −1 2 + γ4Π2 ] dx2 ] , with γ4, γ5 and γ6 constant. As a standard test, we first recover the known possibility of separation in polar coordinates in E2, with x1 = r, x2 = θ, Π1 = r, Π2 = 1. Nonregular Separation of Variables 23 In (8.2) we may choose f1 = f2 = 0 without any loss of generality to the class of allowable scalar and vector potentials. Consequently, from (8.2), R1 = −1 2 log r and R2 = 0. This yields the potential fields that admit separation of the Schrödinger equation a1 = α1(r) = γ5r −1/2 + 2γ4r −1 + r−1/2 ∫ r1/2w1(r)dr, a2 = α2(θ)r −2 = 2γ6r −2 + 2r−2 ∫ γ4 + w2(θ)dθ, Φ = v1(r) + r−2v2(θ)− 1 4 r−2. The above form for the scalar potential agrees with that known to allow separation of polar coordinates in the Helmholtz operator [19]. When the vector potential a in (8.1) takes values in R2, the separation condition (8.4) again leads to the conclusion that in this case, generalized Stäckel matrices do not lead to new separable systems. 8.4 Solute transport The possibilities of new separable equations are broadened if we replace the pure imaginary coefficients of first-order terms in (8.1) by first-order terms with real coefficients. Consider a solute diffusion-convection equation in Cartesian coordinates, ∂tΨ = HΨ = ∂21Ψ + ∂22Ψ− [ q1(x)∂1 + q2(x)∂2 ] Ψ− µ(x)Ψ. In this application, Ψ represents the solute concentration, q(x) represents a steady velocity field of the solvent, and µ(x) is an adsorption coefficient for removal of solute by the solid substrate of a porous medium or by another (usually solid) component of the mixture. If Ψ(x) satisfies HΨ = EΨ, then eEtΨ(x) is a solution of the time dependent solute equation. In standard applications, E ≤ 0, unless the solid component of the mixture is releasing solute (µ(x) < 0) rather than adsorbing solute. After making the replacements q = (−i/2)a and µ(x) = −Φ(x)− (i/2)∇.a, the solute equation is directly analogous to the Schrödinger equation with magnetic field, except that now qj must be real, corresponding to aj being pure imaginary, aj = −iAj , Aj = −2qj , ∂tΨ = HΨ = ∂21Ψ + ∂22Ψ + 1 2 [ A1(x)∂1 +A2(x)∂2 +∇.A(x) ] Ψ + Φ(x)Ψ. In a general orthogonal coordinate system, HΨ = H−21 ∂21Ψ +H−22 ∂22Ψ + [ H−21 ∂1 log H2 H1 ] ∂1Ψ− [ H−22 ∂2 log H2 H1 ] ∂2Ψ + 1 2 [ A1(x)∂1 + ∂1A 1(x) +A2(x)∂2 + ∂2A 2(x) ] Ψ + 1 2 [ A1∂1 log(H1) +A1∂1 log(H2) +A2∂2 log(H1) +A2∂2 log(H2) ] Ψ + Φ(x)Ψ = EΨ, consider R-separation Θ = Ψe−R(x) = Ψ(1) ( x1 ) Ψ(2) ( x2 ) . Then [ H−21 ∂21 +H−22 ∂22 +H−21 { ∂1 log H2 H1 + 2∂1R } ∂1 +H−22 { −∂2 log H2 H1 + 2∂2R } ∂2 ] Θ +H−21 [ ∂21R+ (∂1R)2 + (∂1R)∂1 log H2 H1 ] Θ 24 P. Broadbridge, C.M. Chanu and W. Miller Jr. +H−22 [ ∂22R+ (∂2R)2 − (∂2R)∂1 log H2 H1 ] Θ + Φ(x)Θ + 1 2 [ A1∂1Θ +A2∂2Θ ] + 1 2 [ A1∂1R+A2∂2R+ ∂1A 1 + ∂2A 2 ] Θ + 1 2 A1[∂1 logH1 + ∂1 logH2]Θ + 1 2 A2[∂2 logH1 + ∂2 logH2]Θ = EΘ. Assume the separation equations ∂21Θ + f1 ( x1 ) ∂1Θ + [ v1 ( x1 ) − S11 ( x1 ) E ] Θ = 0, ∂22Θ + f2 ( x2 ) ∂2Θ + [ v2 ( x2 ) − S21 ( x2 ) E ] Θ = 0, ⇒ T11∂ 2 1Θ + T12∂ 2 2Θ + T11f1∂1Θ + T12f2∂2Θ + [T11v1 + T12v2]Θ = EΘ, T21∂ 2 1Θ + T22∂ 2 2Θ + T21f1∂1Θ + T22f2∂2Θ + [T21v1 + T22v2]Θ = 0, where matrix T is the inverse of generalized Stäckel matrix S. In order for the above eigenvalue equation to be identified with the solute equation, we again require T11 = H−21 , T12 = H−22 . Since the first-order terms of the target solute equation no longer have imaginary coefficients, the identification of first-order terms now leads to more general possibilities H2 1q 1 = −f1 ( x1 ) + ∂1 log H2 H1 + 2∂1R, H2 2q 2 = −f2 ( x2 ) − ∂2 log H2 H1 + 2∂2R. (8.5) By differentiating throughout, this implies ∂1∂2 log H2 H1 + 2∂1∂2R− ∂2 ( H2 1q 1 ) = 0, −∂1∂2 log H2 H1 + 2∂1∂2R− ∂1 ( H2 2q 2 ) = 0. This system is equivalent to 4∂1∂2R− ∂2 ( H2 1q 1 ) − ∂1 ( H2 2q 2 ) = 0, 2∂1∂2 log H2 H1 − ∂2 ( H2 1q 1 ) + ∂1 ( H2 2q 2 ) = 0. Similarly, identification of the zero-th order terms also leads to a more general condition T11v1 + T12v2 = −µ+H−21 ∂21R+H−22 ∂22R+H−21 (∂1R)2 +H−22 (∂2R)2 +H−21 (∂1R)∂1 log H2 H1 −H−22 (∂2R)∂2 log H2 H1 − q1∂1R− q2∂2R. (8.6) In principle, the compatible velocity field q is recovered from (8.5) after substituting these expressions in (8.6) which determines the function R. The condition H2/H1 = Π1(x 1)Π2(x 2), which is the key condition for the possibility of replacement by a regular Stäckel matrix construction, is no longer true in general but occurs only if ∂2 ( H2 1q 1 ) − ∂1 ( H2 2q 2 ) = 0. In a description with Cartesian coordinates for Euclidean space, this would be exactly the restriction that the solvent velocity field is irrotational; ∇×q = 0. Consider the canonical form S−1 = T = 1 f(x)− 1 [ f −1 −1 1 ] . Then (8.6) may be written as, −fv1 + v2 = −f∂21R+ ∂22R+ f(∂1R)2 − (∂2R)2 − f∂1R∂1 log(−f) + 2f∂1R∂1 log(1− f) − f1∂1R+ ∂2R∂2 log(−f)− 2∂2R∂2 log(1− f) + f2∂2R+ (f − 1)µ. In looking for examples of genuine nonregular separation, we must consider metrics for which H2/H1 does not separate in variables x1 and x2. Nonregular Separation of Variables 25 Example 3. We consider an example in 2D Euclidean space. The metric is ds2 = dx2 + dy2 = f(u, v)− 1 f(u, v) du2 + (1− f(u, v))dv2, f(u, v) = −1 4 ( u+ v + √ (u+ v)2 − 4 )2 . Here x = (u+ v) cos(φ− u), y = (u+ v) sin(φ− u), φ = 1 2 ( u+ v + √ (u+ v)2 − 4 ) − 2 arctan ( u+ v + √ (u+ v)2 − 4 2 ) , and \|u+ v\| ≥ 2. The Helmholtz equation with vector and scalar potential takes the form( ∆2 − f f − 1 ( 1 2 fu f ∂u − U(u) ) + 1 f − 1 ( −1 2 fv f ∂v − V (v) )) Ψ = EΨ. The separation equations are (∂uu + U(u)− E)Ψ(1)(u) = 0, (∂vv + V (v)− E)Ψ(2)(v) = 0, with Ψ = Ψ(1)(u)Ψ(2)(v). Now r2 = x2 + y2 = (u+ v)2, from which the polar coordinate may be taken to be r = u+ v. Then x = r cos(φ− u), y = r sin(φ− u), implying that the polar angle coordinate is φ− u = θ (mod 2π). Also φ = 1 2 ( r + √ r2 − 4 ) − 2 arctan ( r + √ r2 − 4 2 ) . Note that ±θ is an additive component of each of the variables u = φ(r)−θ and v = r−φ(r)+θ. Therefore physically relevant solutions must be periodic with period 2π in each of the variables u and v. Taking the simplest case v1(= U) = 0 and f1(= V ) = 0, direct separation is possible with R = 0, and from (8.5) (H1) 2q1 = −(H2) 2q2 = 1/ √ (u+ v)2 − 4 = [ r2 − 4 ]−0.5 . The velocity field is defined on the exterior to the circle of radius 2 centred at the origin. The squared magnitude of velocity is( q1 )2 e1.e1 + ( q2 )2 e2.e2 = H2 1 ( q1 )2 +H2 2 ( q2 )2 = 1 r2 − 4 After taking E = −ω2, separated solutions for solute concentration may be combined in a Fourier integral Ψ = ∫ ∞ 0 A(ω)e−ω 2t cos(ω[u+ δ(ω)])cos(ω[v + ε(ω)])dω, with amplitude function A(ω) and phase functions δ(ω), ε(ω). 26 P. Broadbridge, C.M. Chanu and W. Miller Jr. 8.5 Nonregular separation in more than 2 dimensions In dimensions greater than two, both true nonregular separation and R-separation occur even without an added vector potential. Example 4. This is a 3D Minkowski space example. The metric is ds2 = dt2 − dx2 − dy2 = − 4w2 (u2 − v2)2 ( du2 + dv2 ) + dw2 = H2 udu 2 +H2 vdv 2 +H2 wdw 2. Here, t = w u2 − v2 ( 1 4 + ( u2 + v2 )2) , x = w u2 − v2 ( 1 4 − ( u2 + v2 )2) , y = 2wuv u2 − v2 , and we require u, v real and w > 0, \|u\| > \|v\|. The Helmholtz equation in 3D Minkowski space is ∆3Θ = EΘ, where ∆3 = ( u2 − v2 )2 4w2 ( −∂uu − ∂vv + 4w2( u2 − v2 )2∂ww + 8w( u2 − v2 )2∂w ) . We look for R-separable solutions Θ = eRΨ = eRΨ(1)(u)Ψ(2)(v)Ψ(3)(w), eR = 1 w . By direct calculation we can establish the operator identity e−R∆2e R − E = H−2u (∂uu + λ2) +H−2v (∂vv − λ2) +H−2w ( ∂ww − 1 w2 − E ) . Thus the generalized Stäckel matrix and its inverse are S =  0 1 1 + ( u2 − v2 )2 4w2 0 −1 − ( u2 − v2 )2 4w2 1 0 ( u2 − v2 )2 4w2  , T =  − ( u2 − v2 )2 4w2 − ( u2 − v2 )2 4w2 1 − ( u2 − v2 )2 4w2 −1− ( u2 − v2 )2 4w2 0 1 1 0  . In terms of the Ψ functions we have L1Ψ = ( − ( u2 − v2 )2 4w2 (∂uu + ∂vv) + ( ∂ww − 1 w2 )) Ψ = EΨ, L2Ψ = ( − ( u2 − v2 )2 4w2 (∂uu + ∂vv)− ∂vv ) Ψ = −λ2Ψ, L3Ψ = (∂uu + ∂vv) Ψ = 0. The separation equations are (∂uu + λ2)Ψ (1) = 0, (∂vv − λ2)Ψ(2) = 0, ( ∂ww − 1 w2 − E ) Ψ(3) = 0. In terms of the Θ functions the final separated solution of ∆3Θ = EΘ is Θ = Ψ(1)(u)Ψ(2)(v)Ψ(3)(w) w . Nonregular Separation of Variables 27 Example 5. Consider the Helmholtz equation in 3D Euclidean space ∆3Θ = EΘ. It is well known that this equation is regular separable in exactly eleven coordinate systems. Moreover true regular R-separation does not occur for any constant curvature space [15]. This example, due originally to Sym [8, 30, 34, 35] and presented from our point of view, shows that nonregular R-separation with two side conditions occurs in orthogonal coordinates distinct from the usual eleven. Consider Dupin-cyclidic coordinates u, v, w such that x = b2 cosu cosh v + (c cosh v − a cosu)w a cosh v − c cosu , y = b sinu(a cosh v − w) a cosh v − c cosu , z = b sinh v(w − c cosu) a cosh v − c cosu . Here a, b, c are positive parameters, with b2 = a2− c2, c < a, and u ∈ [0, 2π). For any real v we require c cosu < w < a cosh v, so that x, y, z are real. Since ds2 = dx2 + dy2 + dz2 = ( b(a cosh v − w) a cosh v − c cosu )2 du2 + ( b(w − c cosu) a cosh v − c cosu )2 dv2 + dw2 = H2 1du 2 +H2 2dv 2 +H2 3dw 2, these coordinates are orthogonal. Thus ∆3 = ∂xx + ∂yy + ∂zz = 1 H1H2H3 [ ∂u(H−11 H2H3∂u) + ∂v(H 1 1H −1 2 H3∂v) + ∂w(H1H2H −1 3 ∂w) ] . We look for R-separable solutions Θ = eRΨ = eRΨ(1)(u)Ψ(2)(v)Ψ(3)(w), eR = 1√ (w − c cosu)(a cosh v − w) . By direct calculation we can establish the operator identity e−R∆3e R − E = H−21 ( ∂uu + 1 4 ) +H−22 ( ∂vv − 1 4 ) +H−23 (∂ww − E). Thus the generalized Stäckel matrix and its inverse can be taken as S = 0 1 0 0 0 1 1 −H−21 −H−22  , T = H−21 H−22 1 1 0 0 0 1 0  , In terms of the Ψ functions we have L1Ψ = ( H−21 ( ∂uu + 1 4 ) +H−22 ( ∂vv − 1 4 ) + ∂ww ) Ψ = EΨ, L2Ψ = ( ∂uu + 1 4 ) Ψ = 0, L3Ψ = ( ∂vv − 1 4 ) Ψ = 0. The separation equations are( ∂uu + 1 4 ) Ψ(1) = 0, ( ∂vv − 1 4 ) Ψ(2) = 0, (∂ww − E)Ψ(3) = 0, Θ = Ψ(1)(u)Ψ(2)(v)Ψ(3)(w)√ (w − c cosu)(a cosh v − w) . Here we have[ L2, L1 ] = F22L 2 + F23L 3, [ L3, L1 ] = F32L 2 + F33L 3, [ L2, L3 ] = 0, for first-order differential operators Fij , so we have R-separation with two side conditions. 28 P. Broadbridge, C.M. Chanu and W. Miller Jr. Example 6. Again we consider the Helmholtz equation in 3D Euclidean space but now we choose coordinates u, v, w such that x = 1√ 2 (v + w) cosu, y = 1√ 2 (v + w) sinu, z = 1√ 2 (v − w). We take a scalar potential Ṽ (x, y, z) = U(u) (v + w)2 + V (v) +W (w). The metric is ds2 = dx2 + dy2 + dz2 = 1 2 [ (v + w)2du2 + dv2 + dw2 ] , H−2u = 2 (v + w)2 , H−2v = H−2w = 2. Thus ∆3 + Ṽ = ∂xx + ∂yy + ∂zz + Ṽ = 1 H1H2H3 [ ∂u(H−11 H2H3∂u) + ∂v ( H1 1H −1 2 H3∂v ) + ∂w ( H1H2H −1 3 ∂w )] + U(u) (v + w)2 + V (v) +W (w). We look for R-separable solutions Θ = eRΨ = eRΨ(1)(u)Ψ(2)(v)Ψ(3)(w), eR = ( 2 [ x2 + y2 ])− 1 4 . By direct calculation we can establish the operator identity e−R∆3e R − E = 2 (v + w)2 (∂uu + U(u) + 1) + 2 ( ∂vv + V (v) 2 − λ2 ) + 2 ( ∂ww + W (w) 2 − E 2 + λ2 ) . Thus the generalized Stäckel matrix and its inverse can be taken as S =  1 2 1 2− 1 (v + w)2 0 −1 −2 0 0 1  , T = 2 2 2 (v + w)2 0 −1 −2 0 0 1  , The separation equations are (∂uu + U(u) + 1)Ψ(1) = 0, ( ∂vv + V (v) 2 − λ2 ) Ψ(2) = 0,( ∂ww + W (w) 2 − E 2 + λ2 ) Ψ(3) = 0, Θ = Ψ(1)(u)Ψ(2)(v)Ψ(3) √ v + w . Example 7. As a generalization of the previous example, the Euclidean space metrics ds2 = (U1(u)w + U2(u)v + U3(u))2du2 + dv2 + dw2 for arbitrary functions Uj(u) all lead to nonregular separation for Hamilton–Jacobi equations, and in the quantum case, to nonregular R-separation for velocity dependent potentials. The Nonregular Separation of Variables 29 construction of separation equations and of the generalized Stäckel matrix is standard. As an instance, if x = F1(u)v + F2(u)w, y = F3(u)v + F4(u)w, z = F5(u)v + F6(u)w, for F1 = sin2 u, F2 = − cos3 u√ 1 + sin2 u , F3 = sinu cosu, F4 = sinu ( 1 + cos2 u )√ 1 + sin2 u , F5 = cosu, F6 = − sin2 u√ 1 + sin2 u , we have ds2 = dx2 + dy2 + dz2 = (√ 1 + sin2 uv + cosu ( 2 + sin2 u ) 1 + sin2 u w )2 du2 + dv2 + dw2. Example 8. The negative constant curvature metric ds2 = dx2 + dy2 + dz2 z2 = (u+ v)2du2 + dv2 + dw2 w2 for x = (u+ v) sinu+ cosu, y = −(u+ v) cosu+ sinu, z = w, leads to nonregular separation for Hamilton–Jacobi equations, and in the quantum case, to nonregular R-separation for velocity dependent potentials [7]. 9 Conclusions We have demonstrated that the characterization of symmetry related solutions modulo a side condition for Hamilton–Jacobi and Helmholtz or Schrödinger equations coincides with maximal nonregular separation of variables for which there is a generalized Stäckel matrix with one arbitrary column. We have also shown that these systems can be characterized geometrically, i.e., in a coordinate-free manner. We have demonstrated that there is a structure theory for this type of separation; it is not just a collection of examples. This allows us to obtain new separable solutions for these equations that cannot be obtained by the standard (regular) methods of separation of variables. Our work leads to additional questions and possible extensions. • Eisenhart showed that the Robertson condition for Helmholtz separabilty of a Hamilton– Jacobi regular separable coordinate system was the vanishing of the off-diagonal elements of the Ricci tensor in these coordinates [12, 14]. Is there a corresponding geometrical interpretation of the obstruction problem for nonregular separation? • Is there a physical interpretation of the side conditions? • For regular separation of Hamilton–Jacobi and Helmholtz equations all separable systems on N -dimensional Euclidean spaces and N -spheres are known [15]. Can a similar clas- sification of separation with a side condition be carried out? • Much remains to be done in finding explicit, physically interesting, new solutions via this method. • It is clear that separation with a side condition is also appropriate for heat, Laplace and wave equations, where the new solutions are likely to be more interesting physically. This should be done and a structure theory worked out. 30 P. Broadbridge, C.M. Chanu and W. Miller Jr. Acknowledgements This work was partially supported by a grant from the Simons Foundation (#208754 to Willard Miller, Jr.) and by the Australian Research Council (grant DP1095044 to G.E. Prince and P. Broadbridge). References [1] Arrigo D.J., Broadbridge P., Hill J.M., Nonclassical symmetry solutions and the methods of Bluman–Cole and Clarkson–Kruskal, J. Math. Phys. 34 (1993), 4692–4703. [2] Benenti S., Chanu C., Rastelli G., Variable-separation theory for the null Hamilton-Jacobi equation, J. Math. Phys. 46 (2005), 042901, 29 pages. [3] Benenti S., Francaviglia M., The theory of separability of the Hamilton-Jacobi equation and its applications to general relativity, in General Relativity and Gravitation, Vol. 1, Plenum, New York, 1980, 393–439. [4] Bérubé J., Winternitz P., Integrable and superintegrable quantum systems in a magnetic field, J. Math. Phys. 45 (2004), 1959–1973, math-ph/0311051. [5] Bluman G.W., Cole J.D., Similarity methods for differential equations, Applied Mathematical Sciences, Vol. 13, Springer-Verlag, New York, 1974. [6] Bluman G.W., Cole J.D., The general similarity solution of the heat equation, J. Math. Mech. 18 (1969), 1025–1042. [7] Cannon J.W., Floyd W.J., Kenyon R., Parry W.R., Hyperbolic geometry, in Flavors of Geometry, Math. Sci. Res. Inst. Publ., Vol. 31, Cambridge University Press, Cambridge, 1997, 59–115. [8] Chanu C., Geometry of non-regular separation, in Symmetries and Overdetermined Systems of Partial Differential Equations, IMA Vol. Math. Appl., Vol. 144, Editors M. Eastwood, W. Miller Jr., Springer, New York, 2008, 305–317. [9] Chanu C., Rastelli G., Fixed energy R-separation for Schrödinger equation, Int. J. Geom. Methods Mod. Phys. 3 (2006), 489–508, nlin.SI/0512033. [10] Clarkson P.A., Mansfield E.L., Algorithms for the nonclassical method of symmetry reductions, SIAM J. Appl. Math. 54 (1994), 1693–1719. [11] Degiovanni L., Rastelli G., Complex variables for separation of the Hamilton–Jacobi equation on real pseudo- Riemannian manifolds, J. Math. Phys. 48 (2007), 073519, 23 pages, nlin.SI/0610012. [12] Eisenhart L.P., Enumeration of potentials for which one-particle Schrödinger equations are separable, Phys. Rev. 74 (1948), 87–89. [13] Eisenhart L.P., Riemannian geometry, 2nd ed., Princeton University Press, Princeton, N.J., 1949. [14] Eisenhart L.P., Separable systems of Stäckel, Ann. of Math. (2) 35 (1934), 284–305. [15] 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. [16] Kalnins E.G., Miller Jr. W., Conformal Killing tensors and variable separation for Hamilton–Jacobi equa- tions, SIAM J. Math. Anal. 14 (1983), 126–137. [17] Kalnins E.G., Miller Jr. W., Intrinsic characterization of variable separation for the partial differential equations of mechanics, Atti Accad. Sci. Torino Cl. Sci. Fis. Mat. Natur. 117 (1983), suppl. 2, 511–533. [18] 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. [19] Kalnins E.G., Miller Jr. W., The theory of orthogonal R-separation for Helmholtz equations, Adv. Math. 51 (1984), 91–106. [20] Kalnins E.G., Miller Jr. W., Williams G.C., Recent advances in the use of separation of variables methods in general relativity, Philos. Trans. Roy. Soc. London Ser. A 340 (1992), 337–352. [21] Levi D., Winternitz P., Nonclassical symmetry reduction: example of the Boussinesq equation, J. Phys. A: Math. Gen. 22 (1989), 2915–2924. [22] Levi-Civita T., Sulla integrazione della equazione di Hamilton-Jacobi per separazione di variabili, Math. Ann. 59 (1904), 383–397. http://dx.doi.org/10.1063/1.530365 http://dx.doi.org/10.1063/1.1862325 http://dx.doi.org/10.1063/1.1862325 http://dx.doi.org/10.1063/1.1695447 http://dx.doi.org/10.1063/1.1695447 http://arxiv.org/abs/math-ph/0311051 http://dx.doi.org/10.1512/iumj.1969.18.18074 http://dx.doi.org/10.1007/978-0-387-73831-4_14 http://dx.doi.org/10.1142/S021988780600120X http://dx.doi.org/10.1142/S021988780600120X http://arxiv.org/abs/nlin.SI/0512033 http://dx.doi.org/10.1137/S0036139993251846 http://dx.doi.org/10.1137/S0036139993251846 http://dx.doi.org/10.1063/1.2747611 http://arxiv.org/abs/nlin.SI/0610012 http://dx.doi.org/10.1103/PhysRev.74.87 http://dx.doi.org/10.1103/PhysRev.74.87 http://dx.doi.org/10.2307/1968433 http://dx.doi.org/10.1137/0514009 http://dx.doi.org/10.1063/1.527088 http://dx.doi.org/10.1016/0001-8708(84)90004-5 http://dx.doi.org/10.1098/rsta.1992.0071 http://dx.doi.org/10.1088/0305-4470/22/15/010 http://dx.doi.org/10.1088/0305-4470/22/15/010 http://dx.doi.org/10.1007/BF01445149 http://dx.doi.org/10.1007/BF01445149 Nonregular Separation of Variables 31 [23] Miller Jr. W., Mechanisms for variable separation in partial differential equations and their relationship to group theory, in Symmetries and Nonlinear Phenomena (Paipa, 1988), CIF Ser., Vol. 9, World Sci. Publ., Teaneck, NJ, 1988, 188–221. [24] Miller Jr. W., The technique of variable separation for partial differential equations, in Nonlinear Phenomena (Oaxtepec, 1982), Lecture Notes in Phys., Vol. 189, Springer, Berlin, 1983, 184–208. [25] Miller Jr. W., Kalnins E.G., Separation of variables methods for systems of differential equations in math- ematical physics, in Lie Theory, Differential Equations and Representation Theory (Montreal, PQ, 1989), Univ. Montréal, Montreal, QC, 1990, 283–300. [26] Nucci M.C., Clarkson P.A., The nonclassical method is more general than the direct method for symmetry reductions. An example of the FitzHugh–Nagumo equation, Phys. Lett. A 164 (1992), 49–56. [27] Olver P.J., Rosenau P., Group-invariant solutions of differential equations, SIAM J. Appl. Math. 47 (1987), 263–278. [28] Olver P.J., Rosenau P., The construction of special solutions to partial differential equations, Phys. Lett. A 114 (1986), 107–112. [29] Ovsiannikov L.V., Group analysis of differential equations, Academic Press Inc., New York, 1982. [30] Prus R., Sym A., Non-regular and non-Stäckel R-separation for 3-dimensional Helmholtz equation and cyclidic solitons of wave equation, Phys. Lett. A 336 (2005), 459–462. [31] Shapovalov V.N., Separation of variables in a second-order linear differential equation, Differ. Equ. 16 (1981), 1212–1220. [32] Shapovalov V.N., Stäckel spaces, Sib. Math. J. 20 (1979), 790–800. [33] Stäckel P., Über die Integration der Hamilton–Jacobischen differential Gleichung mittelst Separation der Variabelen, Habilitationsschrift, Halle, 1891. [34] Sym A., Solitons of wave equation, J. Nonlinear Math. Phys. 12 (2005), suppl. 1, 648–659. [35] Sym A., Szereszewski A., On Darboux’s approach to R-separability of variables, SIGMA 7 (2011), 095, 21 pages, arXiv:1102.2637. [36] Thirring W., A course in mathematical physics. Vol. 3. Quantum mechanics of atoms and molecules, Lecture Notes in Physics, Vol. 141, Springer-Verlag, New York, 1981. http://dx.doi.org/10.1007/3-540-12730-5_7 http://dx.doi.org/10.1016/0375-9601(92)90904-Z http://dx.doi.org/10.1137/0147018 http://dx.doi.org/10.1016/0375-9601(86)90534-7 http://dx.doi.org/10.1016/j.physleta.2005.01.017 http://dx.doi.org/10.1007/BF00971844 http://dx.doi.org/10.2991/jnmp.2005.12.s1.50 http://dx.doi.org/10.3842/SIGMA.2011.095 http://arxiv.org/abs/1102.2637 1 Introduction 2 Review of regular orthogonal separation for the Hamilton-Jacobi equation H=E 3 Review of the Stäckel procedure for the Helmholtz or Schrödinger equation 4 A generalization of Stäckel form 5 Generalized Stäckel form for the Helmholtz equation 6 Maximal nonregular separation as regular separation with a side condition 7 Maximal nonregular separation generalized Stäckel form 7.1 Differential conditions for nonregular separation on a quadratic first integral leaf 8 Examples of nonregular separability for Hamilton-Jacobi and Helmholtz equations 8.1 Examples of restricted regular separation 8.2 Nonregular separation in 2D and a ``no go'' theorem 8.3 Quantum particle in R2 with a magnetic field 8.4 Solute transport 8.5 Nonregular separation in more than 2 dimensions 9 Conclusions References
id	nasplib_isofts_kiev_ua-123456789-148652
institution	Digital Library of Periodicals of National Academy of Sciences of Ukraine
issn	1815-0659
language	English
last_indexed	2025-12-07T16:00:09Z
publishDate	2012
publisher	Інститут математики НАН України
record_format	dspace
spelling	Broadbridge, P. Chanu, C.M. Miller Jr., Willard 2019-02-18T17:33:50Z 2019-02-18T17:33:50Z 2012 Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables / P. Broadbridge, C.M. Chanu, Willard Miller Jr. // Symmetry, Integrability and Geometry: Methods and Applications. — 2012. — Т. 8. — Бібліогр.: 36 назв. — англ. 1815-0659 2010 Mathematics Subject Classification: 35Q40; 35J05 DOI: http://dx.doi.org/10.3842/SIGMA.2012.089 https://nasplib.isofts.kiev.ua/handle/123456789/148652 Olver and Rosenau studied group-invariant solutions of (generally nonlinear) partial differential equations through the imposition of a side condition. We apply a similar idea to the special case of finite-dimensional Hamiltonian systems, namely Hamilton-Jacobi, Helmholtz and time-independent Schrödinger equations with potential on N-dimensional Riemannian and pseudo-Riemannian manifolds, but with a linear side condition, where more structure is available. We show that the requirement of N−1 commuting second-order symmetry operators, modulo a second-order linear side condition corresponds to nonregular separation of variables in an orthogonal coordinate system, characterized by a generalized Stäckel matrix. The coordinates and solutions obtainable through true nonregular separation are distinct from those arising through regular separation of variables. We develop the theory for these systems and provide examples. This paper is a contribution to the Special Issue “Symmetries of Dif ferential Equations: Frames, Invariants and Applications”. The full collection is available at http://www.emis.de/journals/SIGMA/SDE2012.html. This work was partially supported by a grant from the Simons Foundation (#208754 to Willard Miller, Jr.) and by the Australian Research Council (grant DP1095044 to G.E. Prince and P. Broadbridge). en Інститут математики НАН України Symmetry, Integrability and Geometry: Methods and Applications Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables Article published earlier
spellingShingle	Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables Broadbridge, P. Chanu, C.M. Miller Jr., Willard
title	Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables
title_full	Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables
title_fullStr	Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables
title_full_unstemmed	Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables
title_short	Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables
title_sort	solutions of helmholtz and schrödinger equations with side condition and nonregular separation of variables
url	https://nasplib.isofts.kiev.ua/handle/123456789/148652
work_keys_str_mv	AT broadbridgep solutionsofhelmholtzandschrodingerequationswithsideconditionandnonregularseparationofvariables AT chanucm solutionsofhelmholtzandschrodingerequationswithsideconditionandnonregularseparationofvariables AT millerjrwillard solutionsofhelmholtzandschrodingerequationswithsideconditionandnonregularseparationofvariables

Solutions of Helmholtz and Schrödinger Equations with Side Condition and Nonregular Separation of Variables

Репозитарії

Схожі ресурси