The Exponential Map for Hopf Algebras

We give an analogue of the classical exponential map on Lie groups for Hopf ∗-algebras with differential calculus. The major difference with the classical case is the interpretation of the value of the exponential map, classically an element of the Lie group. We give interpretations as states on the...

Повний опис

Збережено в:
Бібліографічні деталі
Опубліковано в: :Symmetry, Integrability and Geometry: Methods and Applications
Дата:2022
Автори: Alhamzi, Ghaliah, Beggs, Edwin
Формат: Стаття
Мова:Англійська
Опубліковано: Інститут математики НАН України 2022
Онлайн доступ:https://nasplib.isofts.kiev.ua/handle/123456789/211528
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 Exponential Map for Hopf Algebras. Ghaliah Alhamzi and Edwin Beggs. SIGMA 18 (2022), 017, 17 pages

Репозитарії

Digital Library of Periodicals of National Academy of Sciences of Ukraine
_version_ 1859566488892997632
author Alhamzi, Ghaliah
Beggs, Edwin
author_facet Alhamzi, Ghaliah
Beggs, Edwin
citation_txt The Exponential Map for Hopf Algebras. Ghaliah Alhamzi and Edwin Beggs. SIGMA 18 (2022), 017, 17 pages
collection DSpace DC
container_title Symmetry, Integrability and Geometry: Methods and Applications
description We give an analogue of the classical exponential map on Lie groups for Hopf ∗-algebras with differential calculus. The major difference with the classical case is the interpretation of the value of the exponential map, classically an element of the Lie group. We give interpretations as states on the Hopf algebra, elements of a Hilbert 𝐶∗-bimodule of 1/2 densities, and elements of the dual Hopf algebra. We give examples for complex-valued functions on the groups 𝑆₃ and ℤ, Woronowicz's matrix quantum group ℂq[𝑆𝑈₂], and the Sweedler-Taft algebra.
first_indexed 2026-03-13T16:53:20Z
format Article
fulltext Symmetry, Integrability and Geometry: Methods and Applications SIGMA 18 (2022), 017, 17 pages The Exponential Map for Hopf Algebras Ghaliah ALHAMZI a and Edwin BEGGS b a) Imam Mohammad Ibn Saud Islamic University (IMSIU), Riyadh, Saudi Arabia E-mail: gyalhamzi@imamu.edu.sa b) Department of Mathematics, Swansea University, Wales, UK E-mail: e.j.beggs@swansea.ac.uk URL: https://www.swansea.ac.uk/staff/science/maths/beggs-e-j/ Received June 15, 2021, in final form February 16, 2022; Published online March 09, 2022 https://doi.org/10.3842/SIGMA.2022.017 Abstract. We give an analogue of the classical exponential map on Lie groups for Hopf ∗-algebras with differential calculus. The major difference with the classical case is the interpretation of the value of the exponential map, classically an element of the Lie group. We give interpretations as states on the Hopf algebra, elements of a Hilbert C∗-bimodule of 1 2 densities and elements of the dual Hopf algebra. We give examples for complex valued functions on the groups S3 and Z, Woronowicz’s matrix quantum group Cq[SU2] and the Sweedler–Taft algebra. Key words: Hopf algebra; differential calculus; Lie algebra; exponential map 2020 Mathematics Subject Classification: 16T05; 46L87; 58B32 1 Introduction For a Lie group G with Lie algebra g we have the exponential map exp: g → G [13]. We wish to give a Hopf algebra generalisation of this map. The first thing is to decide what maps to what, and the easy bit is what we map out of. Classically we have an algebra of smooth functions C∞(G,C) with a ∗-structure which is just pointwise conjugation. This has a differential calculus of 1-forms Ω1 G (again we take complex valued with a ∗-operation). Then the left invariant 1- forms Λ1 G (equivalently the cotangent space at the identity) has a dual gC, the complexified Lie algebra, and g is the real part of gC. We apologise to real differential geometers for the seemingly unnecessary diversion through the complex numbers, but for Hopf algebras over C this will become necessary. Algebraically, from the group G we take the algebra of complex valued smooth functions C∞(G), and then the differential calculus Ω1 G for this algebra. The target for the exponential map is more of a problem. To explain, we ignore analytic complications (after all we will consider finite and discrete groups), and just take CG to be the complex group algebra. We can view g ∈ G in two ways, first as g ∈ CG and secondly as the state “evaluate at g” on the function algebra C(G). To complicate things further, the state can be written using a GNS construction using a Hilbert space (12 densities) and we could consider g to lie in this Hilbert space. All of these points of view will appear. To motivate the exponential map for Hopf algebras we first look at the Lie group case. Here the motivation is obvious, we know an enormous amount about Lie algebras and their classification, and the exponential map allows us to use this to study Lie groups. Hopf algebras are at a different stage of their history; we know little about the classification theory. We know almost nothing about the non-bicovariant case, in fact the associative algebra U(g) in this case was only worked out recently in terms of higher-order differential operators on Hopf algebras (see [1] or [5, p. 515]). (To clarify, here U(g) denotes the algebra generated by invariant vector fields, and does not refer to a deformation of a classical enveloping algebra for which mailto:gyalhamzi@imamu.edu.sa mailto:e.j.beggs@swansea.ac.uk https://www.swansea.ac.uk/staff/science/maths/beggs-e-j/ https://doi.org/10.3842/SIGMA.2022.017 2 G. Alhamzi and E. Beggs we use Uq(su2) below.) Of course, for a Hopf algebra H the dual of g is a quotient of H+ by an ideal, but this just illustrates the problem in that we start with H to find g. Published material on quantum Lie algebras concentrates on their properties (e.g., [12, 17]), especially in the braided case, and not on a classification theory. We would hope that giving a very general construction for the exponential map for Hopf algebras would motivate the study of the corresponding quantum Lie algebras in their own right, including their classification theory. In turn, this could be used in classification of Hopf algebras. We begin by using the Kasparov, Stinespring, Gel’fand, Năımark and Segal (KSGNS) con- struction (see [15]) in a case which (for C∗-algebras) gives a time varying state ψt on the function algebra C(G). The KSGNS construction works by using a bimodule as half densities, and we have m(t) an element of this bimodule. The construction for m(t) reduces to a differential equa- tion, and solving this equation uses an actual exponential (Taylor series) in the group algebra CG. In Section 3 we give a construction for the dynamics of states on an algebra from [2]. This uses a vector field on an algebra, and for a Hopf algebra we can use the left invariant vector fields in an exact correspondence with the classical case using Woronowicz’s calculi on Hopf algebras [24]. We will do all this for four examples, the discrete group S3 in the simplest case, and the group Z is not much more complicated (though it gives an interesting non-diffusion evolution of states on Z). The quantum group Cq[SU2] has much more complicated formulae, so we carry out calculations only in special cases. The Sweedler–Taft Hopf algebra is included to stress that the method is more general than its initial motivation in C∗-algebras. In solving the dynamics of the states we come on a much simpler and very direct interpretation of the exponential as a power series in the dual Hopf algebra. If we had simply written this in the beginning then questions would have been asked about its role and whether it was just writing a power series to look like the classical case. But now we can be more clear about its role, it plays a fundamental part in the dynamics of the states in the Hopf algebra using Hilbert C∗-bimodules. Given an invariant vector field (i.e., an element of the “Lie algebra” of the Hopf algebra) we get an exponential path in time t ∈ R lying in the dual Hopf algebra starting at ϵ (i.e., evaluation at the identity). Of course, for many Hopf algebras the exponential will not lie in the original dual Hopf algebra as it is an infinite series, but in a completion or formal extension. Since we are exponentiating a vector field X, the reader may be puzzled about why in various places (e.g., Proposition 6.7) we get an exponential of minus X. The simple explanation is that the “weight” defining the functional moves in the opposite direction to what the functional applies to. Thus for a functional Tt : C0(R) → R and weight w(x) Tt(f) = ∫ R f(x+ vt)w(x)dx = ∫ R f(x)w(x− vt)dt. As pointed out by one of the referees, the reader should note the similarities in the construc- tion here with Lévy processes on bialgebras [11]. The reader may ask why we continue to use an exponential with parameter in R in a noncom- mutative setting. The differential setting of the KSGNS construction is very general, and could be used with other Hopf algebras replacing C∞(R). However in [4] it is shown that using C∞(R) is sufficient to describe quantum mechanics (the Schrödinger and Klein–Gordon equations) as auto parallel paths using the proper time as parameter. This shows that the C∞(R) parameter case is of interest, thought not the most general case. The importance of paths on C∗-algebras parametrised by the reals is illustrated by the definition of suspension of an algebra. 2 Preliminaries A first-order differential calculus Ω1 B on an algebra B is a B-bimodule with a derivation d: B → Ω1 B, and so that Ω1 B is spanned by cdb where b, c ∈ B. For a ∗-algebra B, this will The Exponential Map for Hopf Algebras 3 be a ∗-differential calculus if there is an antilinear map ∗ : Ω1 B → Ω1 B so that (c.db)∗ = d(b∗).c∗. The right vector fields χR B consist of right module maps from Ω1 B to B, with evaluation ev : χR B ⊗Ω1 B → B. For a left B-module M a left connection is a linear map ∇M : M → Ω1 B ⊗BM with the left Leibniz rule for b ∈ B and m ∈M ∇M (b.m) = db⊗m+ b.∇M (m). (2.1) In the case where M is a B-A-bimodule we have a left bimodule connection (M,∇M , δM ) when there is a bimodule map σM : M ⊗ A Ω1 A → Ω1 B ⊗ B M for which we have the modified right Leibniz rule for a ∈ A ∇M (m.a) = ∇M (m).a+ σM (m⊗da). Bimodule connections were introduced in [8, 9, 19] and extensively used in [10, 16]. For a Hopf algebra H we use the Sweedler notation ∆h = h(1)⊗h(2). A differential calculus is called left covariant if there is a left H-coaction ∆L : Ω 1 H → H ⊗Ω1 H where ∆L(h.dk) = h(1)k(1)⊗h(2)dk(2) for h, k ∈ H [23]. Similarly to the Sweedler notation, for a left coaction write ∆L(ξ) = ξ[−1]⊗ ξ[0] for ξ ∈ Ω1 H . We call Λ1 H the vector space of left invariant forms (i.e., ξ such that ∆Lξ = 1⊗ ξ ). We now suppose that H has an invertible antipode, required by our choice of right vector fields and left coactions. The left coaction on χR H is defined to make the evaluation ev : χR H ⊗H Ω1 H → H a left comodule map, and is given by, for X ∈ χR H and η ∈ Ω1 H X[−1]⊗X[0](η) = X(η[0])(1)S −1 ( η[−1] ) ⊗X ( η[0] ) (2) . Definition 2.1 ([18]). Two Hopf algebras H and H ′ are dually paired if there is a map ev : H ′⊗H → C which obey, for all α, β ∈ H ′ and h, k ∈ H ev⟨α, hk⟩ = ev⟨α(1), h, ⟩ev⟨α(2), k⟩, ev⟨αβ, h⟩ = ev⟨α, h(1)⟩ev⟨β, h(2)⟩, ev⟨1H′ , h⟩ = ϵH(h), ev⟨α, 1H⟩ = ϵH′(h), ev⟨Sα, h⟩ = ev⟨α, Sh⟩. They are a strictly dual pair if this pairing is nondegenerate. If H is finite-dimensional the idea of dual is quite simply the linear dual. However for infinite- dimensional Hopf algebras we must take more care. Notably the Hopf algebras Cq[SU2] and the deformed enveloping algebra Uq(su2) are dually paired, but Uq(su2) is much smaller than the continuous dual vector space of the C∗-algebra Cq[SU2]. Definition 2.2. A right integral ϕ : H → C is a linear map such that ϕ ( h(1) ) h(2) = 1H .ϕ(h). It is said to be normalised if ϕ(1H) = 1. Definition 2.3 ([18]). A Hopf algebra H which is also a ∗-algebra is called a Hopf ∗-algebra if ∆(h∗) = h(1) ∗⊗h(2) ∗, ϵ(h∗) = ϵ(h)∗, (S ◦ ∗)2 = id. For a Hopf ∗-algebra we call a Haar right integral ϕ Hermitian if ϕ(h∗) = ϕ(h)∗. 4 G. Alhamzi and E. Beggs 3 The KSGNS construction and paths The KSGNS construction [15] for a completely positive map from C∗-algebras A to B is given by an B-A bimodule M and a Hermitian inner product ⟨ , ⟩ : M ⊗ A M −→ B. (3.1) Recall that the conjugate A-B-bimodule M is the conjugate C-vector space, with elements m ∈ M for m ∈ M and m+ n = m + n and λm = λm for m,n ∈ M and λ ∈ C. The actions of the algebras are a.m = ma∗ and m.b = b∗m for a ∈ A and b ∈ B [3]. If we forget about completeness under a norm and positivity we can restate this in terms of more general ∗-algebras. We shall take B = C∞(R), and then in this case we just assume that ⟨ , ⟩ in (3.1) is Hermitian, i.e., ⟨m,n⟩∗ = ⟨n,m⟩. We get ψ : A→ C∞(R) given by ψt(a) = ⟨ma,m⟩, (3.2) which is a time dependent linear functional, and in good cases a time dependent state. We define the time evolution of ψt by imposing the condition ∇Mm = 0 on m in (3.2) where∇M is a left B-connection as in (2.1). An obvious condition to place on the connection∇M is that it preserves the inner product, i.e., that d⟨m,n⟩ = (id⊗⟨ , ⟩) ( ∇Mm⊗n ) + (⟨ , ⟩⊗ id) ( m⊗∇Mn ) with∇Mn = p⊗ ξ∗ where∇Mn = ξ⊗ p. Note that this is just the usual preserving inner product condition used in Riemannian geometry [5]. As special case we consider M = C∞(R)⊗A with actions given by product making it into a C∞(R)-A bimodule. We define the inner product ⟨ , ⟩ on M ⟨f1⊗ a1, f2⊗ a2⟩ = f1f ∗ 2ϕ(a1a ∗ 2) for fi ∈ C∞(R) and ai ∈ A, where ϕ : A → C is Hermitian map (i.e., ϕ(a∗) = ϕ(a)∗) and in nice cases a positive map. In terms of the A valued function of time approach, this is just ⟨m,n⟩(t) = ϕ(m(t)n(t)∗) for m,n ∈M . We consider the special case where ∇M is a bimodule connection. In [2] this is used to recover classical geodesics, but we use this assumption as it gives us a role for vector fields. It also would allow us to define a velocity for the paths, but we do not go into this. We now take the C∞(R)-A bimodule C∞(R)⊗A in the previous theory. However, we quickly find out that this bimodule will not in general contain the solution of the differential equations, and so pass to a larger bimodule C∞(R, A), the infinitely differentiable functions from R to A. Outside the case where A is finite-dimensional (and the two definitions are the same) we would require some topology to define differentiable, but our infinite-dimensional examples are C∗- algebras. Proposition 3.1 ([2]). For a unital algebra A with calculus ΩA and C∞(R) with its usual calculus Ω(R) we set M = C∞(R, A). Then a general left bimodule connection on M is of the form, for m ∈ C∞(R)⊗A and ξ ∈ Ω1 A ∇M (m) = dt⊗ ( pm+ ∂m ∂t +X(dm) ) , σM (1⊗ ξ) = dt⊗X(ξ) for some p ∈ C∞(R, A) and X ∈ C∞(R, χR ) where d is the derivation d: A → Ω1 A. ( Note that explicitly including time evaluation we have X(η)(t) = X(t)(η(t)) for η ∈ C∞(R,Ω1 A ) . ) Further the connection preserves the inner product on M if for all a ∈ A and ξ ∈ Ω1 A. ⟨ ( pa+X(da) + ap∗ ) , 1⟩ = 0 = ⟨X(ξ∗)−X(ξ)∗, 1⟩. (3.3) The Exponential Map for Hopf Algebras 5 Following from the classical theory, we shall call the first equality of the equation (3.3) the divergence condition for p and the second the reality condition for X. In this case the divergence div(X) ∈ A of X ∈ χR for all a ∈ A is given by ϕ(div(X).a+X(da)) = 0. In [2] it is shown that we can set p = 1 2 div(X) in Proposition 3.1. In this paper we only consider the case of a Hopf algebra H and a left invariant right vector field X. Now if both X ∈ χR and ξ ∈ Ω1 H are left invariant we find that X(ξ) ∈ H is left invariant, so it is a multiple of the identity. We use the invariant derivative ω : H → Λ1 H which is defined so that dh = ω ( h(2) ) .h(1), ω(a) = da(2)S −1 ( a(1) ) ∈ Λ1 H . (3.4) Proposition 3.2. If ϕ is a Hermitian right Haar integral on H and X ∈ χR is left invariant, then div(X) = 0. Proof. As X ( ω(a(2)) ) is just a number in the following expression X(da) = X ( da(3)S −1 ( a(2) ) a(1) ) = X ( ω ( a(2) )) a(1), so ϕ(X(da)) = X ( ω ( a(2) )) ϕ ( a(1) ) . For the Hermitian right Haar integral ϕ ( a(1) ) a(2) = ϕ(a).1 so ϕ(X(da)) = X(ω(1))ϕ(a) = 0. ■ Proposition 3.3. For a Hopf ∗-algebra H with a left invariant ∗-calculus and ϕ is a Hermitian right Haar integral, to show that a left invariant right vector field X ∈ χR is real, it is sufficient to check that X(η∗) = X(η)∗ for all η ∈ Λ1 H . Proof. Recalling the property ϕ(a(1))a(2) = ϕ(a).1 for a right Haar integral, we have for ξ ∈ Ω1 1.ϕ(X(ξ∗)) = ϕ ( X(ξ∗)(1) ) X(ξ∗)(2), 1.ϕ(X(ξ)∗) = ϕ ( X(ξ)∗(1) ) X(ξ)∗(2) = ϕ ( X(ξ)(1) ∗)X(ξ)(2) ∗, and as X is left invariant and ev : χR ⊗Ω1 → H is a left comodule map we have 1.ϕ(X(ξ∗)) = ϕ ( ξ∗[−1] ) X ( ξ∗[0] ) = ϕ ( ξ[−1] ∗)X(ξ[0]∗), 1.ϕ(X(ξ)∗) = ϕ ( ξ[−1] ∗)X(ξ[0])∗. For h ∈ H, ϕ(h) is in the field and ϕ(h∗) = ϕ(h)∗ and so 1.ϕ(X(ξ∗)) = X ( ϕ ( ξ[−1] )∗ ξ[0] ∗) = X (( ϕ ( ξ[−1] ) ξ[0] )∗) , 1.ϕ(X(ξ)∗) = X ( ϕ ( ξ[−1] ) ξ[0] )∗ . Finally note that η = ϕ ( ξ[−1] ) ξ[0] is left invariant. ■ Theorem 3.4. The connection ∇Mm = dt⊗(ṁ + X(dm)) for left invariant X ∈ χR has solutions of ∇Mm = 0 given by m(t) = m(0)(1) exp(−t(X ◦ ω)) ( m(0)2 ) , where we take the exponential as a power series in elements of H ′. 6 G. Alhamzi and E. Beggs Proof. We solve ṁ = −X(dm) by using dm = ω(m(2))m(1), so ṁ = −(X ◦ ω) ( m(2) ) m(1), m̈ = −(X ◦ ω) ( ṁ(2) ) m(1) − (X ◦ ω) ( m(2) ) ṁ(1). (3.5) As ∆ and d dt on M commute d dt ( m(1)⊗m(2) ) = ṁ(1)⊗m(2) +m(1)⊗ ṁ(2) = −m(1)⊗m(2)(X ◦ ω) ( m(3) ) , and substituting this back into (3.5) gives m̈ = (X ◦ ω) ( m(3) ) (X ◦ ω) ( m(2) ) m(1) = (X ◦ ω)2 ( m(2) ) m(1) using the product in H ′. Continuing with higher derivatives and using Taylor’s theorem to get the answer, recalling that the first term in the exponential, the identity in H ′, is ϵ. ■ We can use this formula for m(t) in ψt(a) = ϕ(m(t)am(t)∗) to give ψt(a) = exp(−tX ◦ ω) ( m(0)(2) ) exp(−tX ◦ ω) ( n(0)(2) )∗ ϕ ( m(0)(1)an(0) ∗ (1) ) , (3.6) where n is an independent copy of m. Note that for a classical geodesic on a group starting at the identity element we would have H = C∞(G) and m(0) would be a δ-function (or more accurately 1 2 density) at the identity e ∈ G, giving ψ0 = ϵ : C∞(G) → R. 4 Functions on a finite group G We take H = C[G], the functions on a finite group G. A basis is δg for g ∈ G, the function taking value 1 at g and zero elsewhere. This is a Hopf algebra with ϵ(δg) = δg,e, ∆δg = ∑ x,y∈G : xy=g δx⊗ δy, S(δg) = δg−1 . The first-order left covariant differential calculi onH=C[G], correspond to subsets C⊆G\{e} [6]. The basis as a left module for the left invariant 1-forms is ea for a ∈ C, with relations and exterior derivative for f ∈ C[G] being ea.f = Ra(f)e a, df = ∑ a∈C (Ra(f)− f)ea, where Ra(f)(g) = f(ga) denotes right-translation. We take eb for b ∈ C to be the dual basis to ea ∈ Λ1, i.e., ev(eb⊗ ea) = eb(e a) = δa,b. Now from (3.4) ω(δg) =  ∑ ea if g = e, −ea if g−1 = a ∈ C, 0 otherwise, so if we set X = ∑ Xaea ∈ h = (Λ1 H)′, for some Xa ∈ H, then (X ◦ ω)(δg) =  ∑ Xa if g = e, −Xa if g−1 = a ∈ C, 0 otherwise. (4.1) We set ϕ to be the normalised Haar measure ϕ(f) = 1 |G| ∑ g∈G f(g). The Exponential Map for Hopf Algebras 7 Now (X ◦ ω) is an element of the dual of H = C(G), which is the group algebra H ′ = CG. To write elements of the dual Hopf algebra we first list the elements of G as g1, g2, . . . , gn and then for β ∈ H ′ = CG we use a column vector notation β =  β(δg1) β(δg2) ... β(δgn)  . (4.2) It will be convenient to turn the calculation of the exponential on CG into a matrix exponential using a differential equation. We set αt = exp(tX ◦ ω) so dαt dt = αt.(X ◦ ω) and dαt(δg) dt = ∑ xy=g αt(δx)(X ◦ ω(δy)). (4.3) Now we can write (4.3) as matrix equation α̇t = Tαt α̇t(δgi) = ∑ k αt(δgk).(X ◦ ω) ( δg−1 k gi ) , (4.4) where Tik = (X ◦ ω) ( δg−1 k gi ) =  ∑ Xa if k = i, −Xa if g−1 i gk = a ∈ C, 0 otherwise. (4.5) Now we have αt = exp(tT )β where β is the column vector corresponding to the identity in CG and we use the matrix exponential. The calculus on H has a ∗-structure given by ea ∗ = −ea −1 , so by Proposition 3.3 the left invariant vector field X = ∑ Xaea is real if X(a)∗ = −Xa−1 . Example 4.1. Let G = S3, set a ∈ C = {u, v, w} where u = (1, 2), v = (2, 3) and w = (1, 3), and write X = Xueu +Xvev +Xwew where Xu, Xv, Xw ∈ C and the elements of S3 are listed as g1 = e, g2 = (1, 2, 3), g3 = (1, 3, 2), g4 = u, g5 = v, g6 = w. (4.6) Now the matrix T in (4.5) becomes T =  T11 0 0 −Xu −Xv −Xw 0 T22 0 −Xv −Xw −Xu 0 0 T33 −Xw −Xu −Xv −Xu −Xv −Xw T44 0 0 −Xv −Xw −Xu 0 T55 0 −Xw −Xu −Xv 0 0 T66  , where the diagonal elements of T are Tii = Xu+Xv+Xw. Now the solution to α̇t = Tαt is αt = exp(tT )α0. Set α0 = e, the identity in S3, which is the column vector ( 1 0 0 0 0 0 )T , and time t = 1. We set Xu = ip, Xv = iq and Xw = ir and γ = √ p2 + q2 + r2 − pq − pr − qr for p, q, r ∈ C to get exp(i(peu + qev + rew) ◦ ω) 8 G. Alhamzi and E. Beggs = 1 3 ei(p+q+r)  2 cos(γ) + cos(p+ q + r) cos(p+ q + r)− cos(γ) cos(p+ q + r)− cos(γ) −i ( sin(γ)(2p− q − r) γ + sin(p+ q + r) ) −i ( i sin(γ)(2q − p− r) γ + sin(p+ q + r) ) −i ( i sin(γ)(2r − p− q) γ + sin(p+ q + r) )  ∈ CS3. (4.7) In our case C = {u, v, w} so a−1 = a for a ∈ C, and the reality condition is that p, q, r ∈ R and as a result γ ∈ R. Note that the vector does not depend on the sign of the square root and that the L2 norm of the vector in (4.7) is equal to 1. Now we look at time dependence of the state ψt given by equations ψt(a) = ϕ(m(t)am(t)∗) and (3.6). We start with m(0) = δe at t = 0 where everything is concentrated at the identity. We have ∆m(0) = ∆δe = ∑ g δg−1 ⊗ δg, so (3.6) gives ψt(a) = ∑ g,h ϕ(δg−1aδh−1)⟨exp(−X ◦ ω), δg⟩⟨exp(−X ◦ ω), δh⟩∗ = ∑ g ϕ(δg−1a) |⟨exp(−X ◦ ω), δg⟩|2 , (4.8) so ϕ is given by a probability density ∑ g δg−1 |⟨exp(−X ◦ω), δg⟩|2. In terms of the group algebra, which is dual to the functions, 9ψt = e (2 cos(γ) + cos(p+ q + r))2 + ( (132) + (123) ) (cos(p+ q + r)− cos(γ))2 + u ( (2p− q − r) sin(γ) γ + sin(p+ q + r) )2 + v ( (2q − p− r) sin(γ) γ + sin(q + p+ r) )2 + w ( (2r − q − p) sin(γ) γ + sin(r + q + p) )2 . To plot some example exponential of states we refer back to the ordering of group elements in (4.6), and plot the weight of each element against time for 0 ≤ t ≤ 7. We display some cases in Figure 1. This illustrates the conversion of the solution in (p, q, r) in particular cases to a function of the parameter t. Note in general the exponential map will not be periodic as the ratio between γ and p+ q + r is likely not to be rational. Figure 1. The states ψt given by exponentials for vector fields X = iteu + 1 3 itev + 1 2 itew and X = iteu + itev and X = iteu respectively. The Exponential Map for Hopf Algebras 9 5 Functions on the integers Z We shall apply the finite group methods of Section 4 to the group Z, which needs to be treated with care. We shall use rapidly decreasing functions and an un-normalised Haar measure ϕ(f) =∑ n∈Z f(n) and infinite matrices. The column vector notation of (4.2) becomes, truncating the infinite vectors β =  β(δ2) β(δ1) β(δ0) β(δ−1) β(δ−2)  , z−1 =  0 0 0 1 0  , z0 =  0 0 1 0 0  , z1 =  0 1 0 0 0  . (5.1) We have used the pairing of the group algebra CZ with the functions to give the vectors corre- sponding to Zn ∈ CZ, the basis elements corresponding to n ∈ Z. We look at equation (4.4) in the case of the integers, which becomes α̇t(δi) = ∑ k αt(δk)(X ◦ ω)(δi−k). (5.2) We use C = {+1,−1} for the calculus, giving two generators e+1, e−1 with dual invariant vector fields e+1 and e−1. Now for the vector field X = X+1e+1 +X−1e−1 (4.1) becomes (X ◦ ω)(δg) =  X+1 +X−1 if g = 0, −X−1 if g = 1, −X+1 if g = −1, 0 otherwise, so (5.2) becomes α̇t(δi) = −αt(δi+1)X +1 − αt(δi−1)X −1 + ( X+1 +X−1 ) αt(δi). (5.3) To describe this more easily we use matrices Nn (infinite in both directions, we only consider a part centred on the 0, 0 entry) N1 =  0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0  , N−1 =  0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0  , N2 =  0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0  , etc., and NnNm = Nn+m. If we write αt as a column vector similarly to (5.1), then we can write the differential equation α̇t = ( −X+1N−1 + ( X+1 +X−1 ) N0 −X−1N1 ) αt. To find the exponential we need to use a generalised hypergeometric function [22] 0F1(; a;x) = 1 + x a1! + x2 a(a+ 1)2! + · · · . Proposition 5.1. exp(−aN1 + (a+ b)N0 − bN−1) = ea+b ( 0F1(; 1; ab)N0 + ∑ n>0 0F1(;n+ 1; ab) n! ( (−a)nNn + (−b)nN−n )) , 10 G. Alhamzi and E. Beggs so αt = exp(tX ◦ ω) = exp ( −tX−1z1 + ( tX−1 + tX+1 ) z0 − tX+1z−1 ) = et(X +1+X−1) ( 0F1 ( ; 1; t2X−1X+1 ) z0 + ∑ n>0 0F1 ( ;n+ 1; t2X−1X+1 ) n! (( −tX−1 )n zn + (−tX+1)nz−n )) . Proof. Using the trinomial theorem (−aN1 + (a+ b)N0 − bN−1) n = ∑ i,j,k≥0:i+j+k=n n! i!j!k! (−aN1) i((a+ b)N0) j(−bN−1) k = ∑ i,j,k≥0:i+j+k=n n! i!j!k! (−a)i(a+ b)j(−b)kNi−k, so exp(−aN1 + (a+ b)N0 − bN−1) = ∑ i,j,k≥0 (−a)i(a+ b)j(−b)k i!j!k! Ni−k = ea+b ∑ i,k≥0 (−a)i(−b)k i!k! Ni−k. (5.4) If we set i − k = n we get sums depending on the sign of n. For n ≥ 0 we have the coefficient of Nn in (5.4) being ea+b ∑ k≥0 (−a)k+n(−b)k (k + n)!k! = ea+b (−a)n n! ∑ k≥0 n! (k + n)!k! (ab)k = ea+b (−a)n n! 0F1(;n+ 1; ab). For n < 0 we have (putting m = −n) k = i+m and the coefficient of N−m in (5.4) is ea+b ∑ i≥0 (−a)i(−b)i+m i!(i+m)! = ea+b (−a)m m! ∑ i≥0 m! (i+m)!i! (ab)i = ea+b (−b)m n! 0F1(; 1 +m; ab). To calculate exp(tX ◦ ω) we put a = tX−1 and b = tX+1 and apply the matrix exponential to z0 ∈ CZ to find the answer. ■ Now we follow the equation (4.8) in finding the state corresponding to the initial state which is the dual element z0, corresponding to m(0) = δ0, ψt(a) = ϕ(m(t)am(t)∗) = ∑ r∈Z ϕ(δra) |exp(−tX ◦ ω)(δ−r)|2 . As ϕ(δra) = a(r) we see that the state is the element of the dual ∑ n zn| exp(−tX ◦ ω)(δ−n)|2. Now we restrict to the real vector field case ( X+1 )∗ = −X−1, where X+1 +X−1 is imaginary, so ∣∣et(X++X−1) ∣∣ = 1, and ∣∣X+1 ∣∣2 = −X+1X−1 > 0. To plot a graph we take the case ∣∣X+1 ∣∣ = 1 ψt = z0 ∣∣ 0F1 ( ; 1;−t2 )∣∣2 +∑ n≥1 (zn + z−n) ∣∣∣∣∣0F1 ( ;n+ 1;−t2 ) n! ∣∣∣∣∣ 2 |t|2n. The Exponential Map for Hopf Algebras 11 Figure 2. The time evolution of states on the integers for the exponential and diffusion respec- tively. We plot this for integers −4 ≤ n ≤ 4 in the range 0 ≤ t ≤ 5 in the first graph in Figure 2, plotted using standard functions in Mathematica, and there it can be seen that there is a damped oscil- latory behaviour. We should compare this geodesic calculation with the usual diffusion equation on Z which also gives a time dependent state. Diffusion is defined in terms of a Lagrangian ∆ and Lagrangians on graphs have been studied for some time (e.g., [7]). We use the special case for diffusion of a density f : Z× [0,∞) → R given by, for λ > 0, df(n) dt = −λ(∆f)(n) = −λ(2f(n)− f(n− 1)− f(n+ 1)). This is just (5.3) with f(n) = αt(δn) but with X +1 = X−1 = −λ < 0, i.e., with an “imaginary” vector field satisfying X−1 = X+1. We start at t = 0 with f(n) = 0 for n ̸= 0 and f(0) = 1, and this is just the same as the initial condition for αt previously. Now the Proposition 5.1 gives the solution for f as a function of t in the case λ = 1 as f = exp(tX ◦ ω) = ( 0F1 ( ; 1; t2 ) z0 + ∑ n>0 0F1 ( ;n+ 1; t2 ) n! ( tnzn + tnz−n )) e−2t ∈ CZ. We plot this for −4 ≤ n ≤ 4 and 0 ≤ t ≤ 5 as before in the second graph in Figure 2. For both the exponential and the diffusion we have one real parameter, ∣∣X+1 ∣∣ and λ respectively. We can see from the graphs that the behaviour of the states in the two cases is different, with the exponential giving “damped oscillations” and the diffusion giving a monotonic decrease at z0. 6 The exponential map on quantum SU2 We use the matrix quantum group Cq[SU2] as given by Woronowicz [24] and quantum envelop- ing algebra Uq(su2) as given in [14, 20]. There is a dual pairing between H = Cq[SU2] and H ′ = Uq(su2). (We just say dual pairing as H is infinite-dimensional and we need to be careful about duals.) Definition 6.1. For q ∈ C∗ with q2 ̸= −1, we define the quantum group Cq[SU2] to have generators a, b, c, d with relations ba = qab, ca = qac, db = qbd, dc = qcd, cb = bc, da− ad = q ( 1− q−2 ) bc, ad− q−1bc = 1. 12 G. Alhamzi and E. Beggs This is a Hopf algebra with coproduct, antipode and counit ∆(a) = a⊗ a+ b⊗ c, ∆(b) = b⊗ d+ a⊗ b, ∆(c) = c⊗ a+ d⊗ c, ∆(d) = d⊗ d+ c⊗ b, S(a) = d, S(b) = −qb, S(c) = −q−1c, S(d) = a, ϵ(a) = ϵ(d) = 1, ϵ(b) = ϵ(c) = 0. This is a Hopf ∗-algebra with a∗ = d, d∗ = a, c∗ = −qb and b∗ = −q−1c for q real. We define a grade on monomials in generations by |a| = |c| = 1 and |b| = |d| = −1. Definition 6.2. Uq(su2) has generators X+, X−, q ±H 2 , where we have relations q− H 2 q H 2 = q H 2 q− H 2 = 1, q H 2 X±q −H 2 = q±X±, [X+, X−] = qH − q−H q − q−1 , and comultiplication, counit and antipode ∆q± H 2 = q± H 2 ⊗ q± H 2 , ∆X± = X±⊗ q H 2 + q− H 2 ⊗X±, ϵ ( q± H 2 ) = 1, ϵ(X±) = 0, S(X±) = −q±X±, S ( q± H 2 ) = q∓ H 2 . As in Definition 2.1, these are dually paired by ⟨α, tij⟩ = ρ(α)ij ∈ C where α ∈ Uq(su2) and t11 = a, t12 = b, t21 = c and t22 = d and ρ : Uq(su2) → M2(C) is the representation (where r = √ q) ρ(q H 2 ) = ( r 0 0 1 r ) , ρ(X+) = ( 0 1 0 0 ) , ρ(X−) = ( 0 0 1 0 ) . Definition 6.3 ([24]). The left covariant 3D calculus for the quantum group Cq[SU2] has gen- erators e0 and e±. The relations are e±a = qae±, e±b = q−1be±, e±c = qce±, e±d = q−1de±, e0a = q2ae0, e0b = q−2be0, e0c = q2ce0, e0d = q−2de0, and exterior derivative and the ∗-operator da = ae0 + qbe+, db = ae− − q−2be0, dc = ce0 + qde+, dd = ce− − q−2de0, e0∗ = −e0, e+∗ = −q−1e−, e−∗ = −qe+. Now using ω : H → Λ1 H from (3.4) we calculate ω(a) = q−2e0, ω(b) = q−1e−, ω(c) = q2e+, ω(d) = −e0. (6.1) We define e0, e+, e− to be the dual basis of e0, e+, e−, i.e., ⟨ei, ej⟩ = δij . Now every ei ◦ω gives a map from Cq[SU2] to C. We shall identify ei ◦ ω as an element of Uq(su2). The first step is to apply ei ◦ ω to a product. Lemma 6.4. For all g, h ∈ Cq[SU2] (e± ◦ ω)(gh) = (e± ◦ ω)(g)ϵ(h) + q−|g|ϵ(g)(e± ◦ ω)(h), (e0 ◦ ω)(gh) = (e0 ◦ ω)(g)ϵ(h) + q−2|g|ϵ(g)(e0 ◦ ω)(h). The Exponential Map for Hopf Algebras 13 Proof. By definition ω(gh) = d ( g(2)h(2) ) S−1 ( g(1)h(1) ) = ω(g)ϵ(h) + g(2)ω(h)S −1 ( g(1) ) . Now ejg(2)e iS−1 ( g(1) ) = 0 unless i = j, so we need to show g(2)e ±S−1 ( g(1) ) = ϵ(g)q−|g|e±, g(2)e 0S−1 ( g(1) ) = ϵ(g)q−2|g|e0. It is enough to do this on the generators a(2)e ±S−1(a(1)) = ae±S−1(a) + ce±S−1(b) = ae±d− q−1ce±b = q−1 ( ad− q−1cb ) e± = q−1e±, and similarly for e0 and b, c, d. ■ We can use Lemma 6.4 to identify the coproduct of ei ◦ ω, where the linear map g 7→ ϵ(g) is just 1 ∈ Uq(su2). To do this we need to identify the map g 7→ qs|g|ϵ(g). Lemma 6.5. For s ∈ R and g ∈ Cq[SU2] we have ⟨qsH , g⟩ = qs|g|ϵ(g). Proof. As ∆qsH = qsH ⊗ qsH where 〈 qsH , hg 〉 = 〈 qsH , h 〉〈 qsH , g 〉 we only have to check the formula on the generators, and this is 〈 qsH , tij 〉 = ρ ( qsH )i j using ρ ( qsH ) = ( qs 0 0 q−s ) . ■ Proposition 6.6. In Uq(su2) we have (ei ◦ ω)(h) = ev(νi⊗h) where νi ∈ Uq(su2) is given by, where r = √ q, ν0 = e0 ◦ ω = 1− q−2H q2 − 1 , ν+ = e+ ◦ ω = r3q −H 2 X−, ν− = e− ◦ ω = r−1q −H 2 X+. Proof. First we check that νi = ei ◦ ω on the generators ρ(ν0) = ( q−2 0 0 −1 ) , ρ(ν+) = ( 0 0 q2 0 ) , ρ(ν−) = ( 0 q−1 0 0 ) . Thus ν0 applied to b, c gives zero, and this is also true for e0 ◦ ω by (6.1). Also ν+ applied to a, b, d gives zero, and this is also true for e+ ◦ ω by (6.1). Lastly ν− applied to a, c, d gives zero, and this is also true for e− ◦ ω. We are left with the cases, for (e0 ◦ ω)(a) = e0 ( q−2e0 ) = q−2 = ⟨ν0, a⟩, (e0 ◦ ω)(d) = e0 ( −e0 ) = −1 = ⟨ν0, d⟩, (e+ ◦ ω)(c) = q2 = ⟨ν+, c⟩, (e− ◦ ω)(b) = q−1 = ⟨ν−, b⟩. (6.2) Now we need to show that we get equality on products of generators, which we show by induction. Suppose that is the dual ei ◦ ω = νi when applied to products of ≤ n generators. If g, h are product of ≤ n generators, then ν±(gh) = ν±(g)ϵ(h) + 〈 q−H , g 〉 ν±(h) = ν±(g)ϵ(h) + q−|g|ν±(h), which is (e± ◦ ω)(gh) by Lemma 6.4. Also from Lemma 6.5 ν0(gh) = 1 q2 − 1 ϵ(g)ϵ(h)− 1 q2 − 1 〈 q−2H , gh 〉 = ϵ(g)ϵ(h) q2 − 1 ( 1− q−2|g|q−2|h|), whereas (e0 ◦ ω)(gh) = ϵ(h) q2 − 1 ( ϵ(g)− q−2|g|ϵ(g) ) + q−2|g| q2 − 1 ( ϵ(h)− q−2|h|ϵ(h) ) ϵ(g) = ϵ(g)ϵ(h) q2 − 1 ( 1− q−2|g|q−2|h|) = ν0(gh) as required. ■ 14 G. Alhamzi and E. Beggs The previously calculated exponentials in this case give a formal exponential of a linear combination of the ν in Uq(su2). We consider that there is little reason to just write out such a formal sum. However, we can calculate the time evolution of certain states on Cq[SU2]. We use states of the form ψt(a) = ϕ(m(t)am(t)∗) where (m(t)) is defined in Theorem 3.4. First we look at the real vector field X = ie0. Proposition 6.7. For any monomial y in the generators a, b, c, d we have ν0(y) = −[−|y|]q2ϵ(y) using the q2 integer [n]q2 = ( 1− q2n ) / ( 1− q2 ) , and exp(itν0)(y) = exp ( −it[−|y|]q2 ) ϵ(y). Proof. The first result is given by using the definition of ν0 in Proposition 6.6 with Lemma 6.5. Then ν20(y) = ν0(y(1))ν0(y(2)) = [−|y(1)|]q2 [−|y(2)|]q2ϵ(y(1))ϵ(y(2)). As ϵ annihilates b and c we must have y containing no b, c for this to be nonzero. As ∆a = a⊗ a+b⊗ c and ∆d = d⊗ d+c⊗ b we have ν20(y) = ν0(y) 2. This continues with higher powers. ■ Proposition 6.8. For the real vector field X = ie0 we have (summing over monomials in m(0)) m(t) = m(0) exp ( it[−|m(0)|]q2 ) . Proof. We have from Theorem 3.4 and Proposition 6.7 m(t) = m(0)(1) exp(−ite0 ◦ ω) ( m(0)(2) ) = m(0)(1) exp ( it[−|m(0)(2)|]q2 ) ϵ ( m(0)(2) ) . Note that for a monomial m(0), we have |m(0)(2)| = |m(0)| giving the answer. ■ We now look at the X = γe+ + δe− case. In the previous examples we had ψ0(a) = ϕ(m(0)am(0)∗) = ϵ(a). However, because of the complexity of the calculation we will simply calculate m(t) for m(0) a generator. Proposition 6.9. For X = γe+ + δe− we calculate m(t) for m(0) being generator to be m(0) = a, m(t) = a cosh ( t √ qγδ ) − b q2γ√ qγδ sinh ( t √ qγδ ) , m(0) = b, m(t) = b cosh ( t √ qγδ ) − a q−1δ√ qγδ sinh ( t √ qγδ ) , m(0) = c, m(t) = c cosh ( t √ qγδ ) − d q2γ√ qγδ sinh ( t √ qγδ ) , m(0) = d, m(t) = d cosh ( t √ qγδ ) − c q−1δ√ qγδ sinh ( t √ qγδ ) . Proof. We get the corresponding element of the dual x = X ◦ ω = γν+ + δν−. Then xn(h) = x ( h(1) ) xn−1 ( h(2) ) , so from (6.2) xn(a) = q−1δxn−1(c), xn(b) = q−1δxn−1(d), xn(c) = q2γxn−1(a), xn(d) = q2γxn−1(b). These give xodd(a) = xodd(d) = 0, xeven(b) = xeven(c) = 0, x2n(a) = (qγδ)n = x2n(d), x2n+1(b) = qn−1γnδn+1, x2n+1(c) = qn+2γn+1δn, The Exponential Map for Hopf Algebras 15 so for the formal exponential e−tx(h) = ∑ n≥0 (−tx)n n! (h) e−tx(a) = ∑ n≥0 t2n(−x)2n 2n! (a) = ∑ n≥=0 t2n(qγδ)n 2n! = cosh ( t √ qγδ ) = e−tx(d), e−tx(b) = ∑ n≥0 t2n+1(−x)2n+1 (2n+ 1)! (b) = − ∑ n≥0 t2n+1qn−1γn+1δn (2n+ 1)! = −q−1δ√ qγδ sinh ( t √ qγδ ) , e−tx(c) = ∑ n≥0 t2n+1(−x)2n+1 (2n+ 1)! (c) = − ∑ n≥0 t2n+1qn+2γn+1δn (2n+ 1)! = −q2γ√ qγδ sinh ( t √ qγδ ) , and we use Theorem 3.4 to get the answer. ■ Example 6.10. For X = γe+ + δe− to be real by Proposition 3.3 we have γ∗ = −q−1δ, so√ qγδ = √ −q2|γ|2 = ±iq|γ|. Then a special case of Proposition 6.9 gives m(0) = a, m(t) = a cosh(iq|γ|t)− b q2γ iq|γ| sinh(iq|γ|t) = a cos(q|γ|t)− b qγ |γ| sin(q|γ|t). Using an exponential map on an initial state will give a time evolution preserving the normali- sation as long as we use a real invariant vector field. We now check this in our case of γe++δe−. We begin with the Haar integral on Cq[SU2] which is zero on all basis elements anbrcs and dnbrcs except ϕ((bc)r) = (−1)rqr [r + 1]q2 . Starting from m(0) = a we find from ϕ(m(t)m(t)∗) that ψt(1) = q2 1+q2 which is independent of t as required. 7 Sweedler–Taft algebra We will now look at the Sweedler–Taft algebra H of dimension 4. This Hopf algebra does not have a normalised Haar integral, and since we use the integral in finding the “state” we shall get an algebraic construction which is nothing like the C∗-algebra framework. The Sweedler–Taft algebra [21] is a unital algebra with generators x, t and relations t2 = 1, x2 = 0, xt = −tx, so it is 4-dimensional with basis {1, t, x, xt}. The following operations make it into a Hopf algebra ∆1 = 1⊗ 1, ∆t = t⊗ t, ∆x = x⊗ t+ 1⊗x, ∆(tx) = tx⊗ 1 + t⊗ tx, ϵ(t) = ϵ(1) = 1, ϵ(x) = ϵ(tx) = 0, St = t, Sx = tx. We make the Sweedler–Taft algebra into a Hopf ∗-algebra by t∗ = t and x∗ = x. There is a unique 2D bicovariant calculus with right ideal I ⊂ H+ = ker ϵ generated by x−xt and the ∗ operation above has S((x− xt)∗) = x− xt and so gives a ∗-calculus [5]. We take the basis e1 = [x] ∈ H+/I = Λ1 and e2 = [t− 1] of Λ1 and relations dt = te2, dx = xe2 + e1, d(xt) = te1, e2t = −te2, e1t = te1, e1x = xe1, e2x = −xe2 − 2e1, e1 ∧ e1 = e2 ∧ e2 = 0, e1 ∧ e2 = e2 ∧ e1, de2 = 0, de1 = −e1 ∧ e2. 16 G. Alhamzi and E. Beggs Then dt∗ = −e2t gives e2∗ = −e2 and dx∗ = xe2 + e1 gives e1∗ = −e1. We introduce a basis of left invariant right vector fields e1 and e2 dual to e1, e2 ∈ Λ1. From (3.4) we find ω(t) = −e2, ω(x) = −e1 and ω(tx) = −e1. Now we find ei ◦ ω ∈ H ′. We get ei ◦ ω(1) = 0 and e2 ◦ ω(t) = −1, e2 ◦ ω(x) = e2 ◦ ω(tx) = 0, e1 ◦ ω(t) = 0, e1 ◦ ω(x) = e1 ◦ ω(tx) = −1. The dual basis elements corresponding to 1, t, x, tx are δ1, δt, δx, δtx and their products are δ1.δ1 = δ1, δt.δt = δt, δx.δt = δx = δ1.δx, δtx.δ1 = δtx = δt.δtx, and any other product is zero. Then e1 ◦ ω = −δt and e2 ◦ ω = −δx − δtx. The vector field X = ae1 + be2 for a, b ∈ C has X ◦ ω = −aδt − b(δx + δtx), (X ◦ ω)2 = a2δ2t + b2(δx + δtx) 2 + abδt(δx + δtx) + ab(δx + δtx)δt = −a(X ◦ ω), so adding a time parameter s as usual (t being used already) exp(−s(X ◦ ω)) = ϵ− s(X ◦ ω) + (−s(X ◦ ω))2 2! + (−s(X ◦ ω))3 3! + · · · = ϵ− eas − 1 a (X ◦ ω), so from Theorem 3.4 m(s) = m(0) +m(0)(1) 1− esa a (X ◦ ω)(m(0)(2)). (7.1) We now turn to the inner product and definition of real vector fields. The Haar integral [5] is given by ϕ(tx) = λ where λ is arbitrary and ϕ of all other basis elements, including 1, to be zero. For Proposition 3.2 we require ϕ to be Hermitian, so λ is imaginary. Applying this ϕ in the formula for the inner product gives all left invariant vector fields being real, as ϕ(1) = 0. Note that this definition of reality corresponds to preserving the inner product in Proposition 3.1, rather than what in this case is the stronger condition on duals in Proposition 3.3. Example 7.1. Set m(0) = t+ x and X = ae1 + be2 as above. Then by (7.1) m(s) = t+ x+ 1− esa a (t(X ◦ ω)(t) + 1(X ◦ ω(x)) + x(X ◦ ω)(t)) = t+ x+ 1− esa a (−tb− xb− a). To calculate the value of the “state” ψs in ϕ(m(t)am(t)∗) we setm(s) = m11+mtt+mxx+mtxtx for mi ∈ C then for the Haar measure ϕ the map a 7→ ϕ(mam∗) is the element of the dual λδ1(−m1mtx +mtmx −mxmt +mtxm1) + λδt(m1mx −mtmtx −mxm1 +mtxmt) + λδx(−m1mt +mtm1) + λδtx ( |m1|2 − |mt|2 ) . Substituting the values for m(s) gives ψs = λ(δt − δx) ( esa − esa + ∣∣esa − 1 ∣∣2( b a − b a )) + λδtx (∣∣esa − 1 ∣∣2(1− ∣∣∣∣ ba ∣∣∣∣2 ) − 1− b a ( esa − 1 ) − b a ( esa − 1 )) . The Exponential Map for Hopf Algebras 17 Acknowledgements We would like to thank the editor and referees for many useful comments. Computer algebra and graphs were done on Mathematica. References [1] Beggs E.J., Differential and holomorphic differential operators on noncommutative algebras, Russ. J. Math. Phys. 22 (2015), 279–300, arXiv:1209.3900. [2] Beggs E.J., Noncommutative geodesics and the KSGNS construction, J. Geom. Phys. 158 (2020), 103851, 14 pages, arXiv:1811.07601. [3] Beggs E.J., Majid S., Bar categories and star operations, Algebr. Represent. Theory 12 (2009), 103–152, arXiv:math.QA/0701008. [4] Beggs E.J., Majid S., Quantum geodesics in quantum mechanics, arXiv:1912.13376. [5] Beggs E.J., Majid S., Quantum Riemannian geometry, Grundlehren der mathematischen Wissenschaften, Vol. 355, Springer, Cham, 2020. [6] Bresser K., Müller-Hoissen F., Dimakis A., Sitarz A., Non-commutative geometry of finite groups, J. Phys. A: Math. Gen. 29 (1996), 2705–2735. [7] Chung F.R.K., Spectral graph theory, CBMS Regional Conference Series in Mathematics, Vol. 92, Amer. Math. Soc., Providence, RI, 1997. [8] Dubois-Violette M., Masson T., On the first-order operators in bimodules, Lett. Math. Phys. 37 (1996), 467–474, arXiv:q-alg/9507028. [9] Dubois-Violette M., Michor P.W., Connections on central bimodules in noncommutative differential geom- etry, J. Geom. Phys. 20 (1996), 218–232, arXiv:q-alg/9503020. [10] Fiore G., Madore J., Leibniz rules and reality conditions, Eur. Phys. J. C Part. Fields 17 (2000), 359–366, arXiv:math.QA/9806071. [11] Franz U., Lévy processes on quantum groups, in Probability on Algebraic Structures (Gainesville, FL, 1999), Contemp. Math., Vol. 261, Amer. Math. Soc., Providence, RI, 2000, 161–179. [12] Gomez X., Majid S., Braided Lie algebras and bicovariant differential calculi over co-quasitriangular Hopf algebras, J. Algebra 261 (2003), 334–388, arXiv:math.QA/0112299. [13] Hausner M., Schwartz J.T., Lie groups; Lie algebras, Gordon and Breach Science Publishers, New York – London – Paris, 1968. [14] Kulish P.P., Reshetikhin N.Yu., Quantum linear problem for the sine-Gordon equation and higher represen- tations, J. Sov. Math. 23 (1983), 2435–2441. [15] Lance E.C., Hilbert C∗-modules. A toolkit for operator algebraists, London Mathematical Society Lecture Note Series, Vol. 210, Cambridge University Press, Cambridge, 1995. [16] Madore J., An introduction to noncommutative differential geometry and its physical applications, 2nd ed., London Mathematical Society Lecture Note Series, Vol. 257, Cambridge University Press, Cambridge, 1999. [17] Majid S., Quantum and braided-Lie algebras, J. Geom. Phys. 13 (1994), 307–356, arXiv:hep-th/9303148. [18] Majid S., Foundations of quantum group theory, Cambridge University Press, Cambridge, 1995. [19] Mourad J., Linear connections in non-commutative geometry, Classical Quantum Gravity 12 (1995), 965– 974. [20] Sklyanin E.K., Some algebraic structures connected with the Yang–Baxter equation, Funct. Anal. Appl. 16 (1982), 263–270. [21] Taft E.J., The order of the antipode of finite-dimensional Hopf algebra, Proc. Nat. Acad. Sci. USA 68 (1971), 2631–2633. [22] Weisstein E.W., Generalized hypergeometric function, FromMathWorld – AWolframWeb Resource, https: //mathworld.wolfram.com/GeneralizedHypergeometricFunction.html. [23] Woronowicz S.L., Twisted SU(2) group. An example of a noncommutative differential calculus, Publ. Res. Inst. Math. Sci. 23 (1987), 117–181. [24] Woronowicz S.L., Differential calculus on compact matrix pseudogroups (quantum groups), Comm. Math. Phys. 122 (1989), 125–170. https://doi.org/10.1134/S1061920815030012 https://doi.org/10.1134/S1061920815030012 https://arxiv.org/abs/1209.3900 https://doi.org/10.1016/j.geomphys.2020.103851 https://arxiv.org/abs/1811.07601 https://doi.org/10.1007/s10468-009-9141-x https://arxiv.org/abs/math.QA/0701008 https://arxiv.org/abs/1912.13376 https://doi.org/10.1007/978-3-030-30294-8 https://doi.org/10.1088/0305-4470/29/11/010 https://doi.org/10.1088/0305-4470/29/11/010 https://doi.org/10.1007/BF00312677 https://arxiv.org/abs/q-alg/9507028 https://doi.org/10.1016/0393-0440(95)00057-7 https://arxiv.org/abs/q-alg/9503020 https://doi.org/10.1007/s100520000470 https://arxiv.org/abs/math.QA/9806071 https://doi.org/10.1090/conm/261/04139 https://doi.org/10.1016/S0021-8693(02)00580-X https://arxiv.org/abs/math.QA/0112299 https://doi.org/10.1007/BF01084171 https://doi.org/10.1017/CBO9780511526206 https://doi.org/10.1017/CBO9780511569357 https://doi.org/10.1016/0393-0440(94)90014-0 https://arxiv.org/abs/hep-th/9303148 https://doi.org/10.1017/CBO9780511613104 https://doi.org/10.1088/0264-9381/12/4/007 https://doi.org/10.1007/BF01077848 https://doi.org/10.1073/pnas.68.11.2631 https://mathworld.wolfram.com/GeneralizedHypergeometricFunction.html https://mathworld.wolfram.com/GeneralizedHypergeometricFunction.html https://doi.org/10.2977/prims/1195176848 https://doi.org/10.2977/prims/1195176848 https://doi.org/10.1007/BF01221411 https://doi.org/10.1007/BF01221411 1 Introduction 2 Preliminaries 3 The KSGNS construction and paths 4 Functions on a finite group G 5 Functions on the integers Z 6 The exponential map on quantum SU_2 7 Sweedler-Taft algebra References
id nasplib_isofts_kiev_ua-123456789-211528
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
issn 1815-0659
language English
last_indexed 2026-03-13T16:53:20Z
publishDate 2022
publisher Інститут математики НАН України
record_format dspace
spelling Alhamzi, Ghaliah
Beggs, Edwin
2026-01-05T12:26:18Z
2022
The Exponential Map for Hopf Algebras. Ghaliah Alhamzi and Edwin Beggs. SIGMA 18 (2022), 017, 17 pages
1815-0659
2020 Mathematics Subject Classification: 16T05; 46L87; 58B32
arXiv:2203.04549
https://nasplib.isofts.kiev.ua/handle/123456789/211528
https://doi.org/10.3842/SIGMA.2022.017
We give an analogue of the classical exponential map on Lie groups for Hopf ∗-algebras with differential calculus. The major difference with the classical case is the interpretation of the value of the exponential map, classically an element of the Lie group. We give interpretations as states on the Hopf algebra, elements of a Hilbert 𝐶∗-bimodule of 1/2 densities, and elements of the dual Hopf algebra. We give examples for complex-valued functions on the groups 𝑆₃ and ℤ, Woronowicz's matrix quantum group ℂq[𝑆𝑈₂], and the Sweedler-Taft algebra.
We would like to thank the editor and referees for many useful comments. Computer algebra and graphs were done on Mathematica.
en
Інститут математики НАН України
Symmetry, Integrability and Geometry: Methods and Applications
The Exponential Map for Hopf Algebras
Article
published earlier
spellingShingle The Exponential Map for Hopf Algebras
Alhamzi, Ghaliah
Beggs, Edwin
title The Exponential Map for Hopf Algebras
title_full The Exponential Map for Hopf Algebras
title_fullStr The Exponential Map for Hopf Algebras
title_full_unstemmed The Exponential Map for Hopf Algebras
title_short The Exponential Map for Hopf Algebras
title_sort exponential map for hopf algebras
url https://nasplib.isofts.kiev.ua/handle/123456789/211528
work_keys_str_mv AT alhamzighaliah theexponentialmapforhopfalgebras
AT beggsedwin theexponentialmapforhopfalgebras
AT alhamzighaliah exponentialmapforhopfalgebras
AT beggsedwin exponentialmapforhopfalgebras