Skip to main content
Log in

Langlands duality and Poisson–Lie duality via cluster theory and tropicalization

  • Published:
Selecta Mathematica Aims and scope Submit manuscript

Abstract

Let G be a connected semisimple Lie group. There are two natural duality constructions that assign to G: its Langlands dual group \(G^\vee \), and its Poisson–Lie dual group \(G^*\), respectively. The main result of this paper is the following relation between these two objects: the integral cone defined by the cluster structure and the Berenstein–Kazhdan potential on the double Bruhat cell \(G^{\vee ; w_0, e} \subset G^\vee \) is isomorphic to the integral Bohr–Sommerfeld cone defined by the Poisson structure on the partial tropicalization of \(K^* \subset G^*\) (the Poisson–Lie dual of the compact form \(K \subset G\)). By Berenstein and Kazhdan (in: Contemporary mathematics, vol. 433. American Mathematical Society, Providence, pp 13–88, 2007), the first cone parametrizes the canonical bases of irreducible G-modules. The corresponding points in the second cone belong to integral symplectic leaves of the partial tropicalization labeled by the highest weight of the representation. As a by-product of our construction, we show that symplectic volumes of generic symplectic leaves in the partial tropicalization of \(K^*\) are equal to symplectic volumes of the corresponding coadjoint orbits in \({{\,\mathrm{Lie}\,}}(K)^*\). To achieve these goals, we make use of (Langlands dual) double cluster varieties defined by Fock and Goncharov (Ann Sci Ec Norm Supér (4) 42(6):865–930, 2009). These are pairs of cluster varieties whose seed matrices are transpose to each other. There is a naturally defined isomorphism between their tropicalizations. The isomorphism between the cones described above is a particular instance of such an isomorphism associated to the double Bruhat cells \(G^{w_0, e} \subset G\) and \(G^{\vee ; w_0, e} \subset G^\vee \).

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3

Similar content being viewed by others

Notes

  1. For a n-dimensional real vector space V and \(D\subset V\), by a lattice in D or simply a lattice, we mean a subset of D of the form \(L\cap D\), where \(L\cong {\mathbb {Z}}^n\) is a lattice in V. We use this terminology throughout this paper.

  2. The terminology flip of a triangulations is used in [5, 12, 13]. For a triangulation of n-gon, one may consider the dual graph of it, which is a tree. Then the flip of a triangulation induces a “move” of the corresponding tree. This move is called Whitehead move in [30]. In Teichmüller Theory, the terminology “Whitehead move” is usually presented by a similar move of trees, see e.g., [29]. The authors of [15] use the name Whitehead move for the flip of a diagonal of a triangulation.

References

  1. Alekseev, A., Berenstein, A., Hoffman, B., Li, Y.: Poisson structures and potentials. In: Lie Groups, Geometry, and Representation Theory. Progress in Mathematics, pp. 1–40. Birkhäuser/Springer, Cham (2018)

  2. Alekseev, A., Hoffman, B., Lane, J., Li, Y.: Concentration of symplectic volumes on Poisson homogeneous spaces. JSG 18(5), 1197–1220 (2020)

    MathSciNet  MATH  Google Scholar 

  3. Alekseev, A., Hoffman, B., Lane, J., Li, Y.: Action-angle coordinates on coadjoint orbits and multiplicity free spaces from partial tropicalization. arXiv: 2003.13621

  4. Alekseev, A., Lane, J., Li, Y.: The \(U(n)\) Gelfand–Zeitlin system as a tropical limit of Ginzburg–Weinstein diffeomorphisms. Philos. Trans. R. Soc. A 376, 20170428 (2018)

    Article  MathSciNet  Google Scholar 

  5. Burman, Y.M.: Triangulations of surfaces with boundary and the homotopy principle for functions without critical points. Ann. Glob. Anal. Geom. 17, 221–238 (1999)

    Article  MathSciNet  Google Scholar 

  6. Berenstein, A., Fomin, S., Zelevinsky, A.: Cluster algebras III: upper bounds and double Bruhat cells. Duke Math. J. 126(1), 1–52 (2005)

    Article  MathSciNet  Google Scholar 

  7. Berenstein, A., Kazhdan, D.: Geometric and unipotent crystals II: from unipotent bicrystals to crystal bases quantum groups. Contemporary Mathematics, vol. 433, pp 13–88. American Mathematical Society, Providence (2007)

  8. Berenstein, A., Zelevinsky, A.: Tensor product multiplicities, canonical bases and totally positive varieties. Invent. Math. 143, 77–128 (2001)

    Article  MathSciNet  Google Scholar 

  9. Frenkel, E., Hernandez, D.: Langlands duality for finite-dimensional representations of quantum affine algebras. Lett. Math. Phys. 96(1–3), 217–261 (2011)

    Article  MathSciNet  Google Scholar 

  10. Fock, V.V., Goncharov, A.B.: Cluster ensembles, quantization and the dilogarithm. Ann. Sci. Ec. Norm. Supér (4) 42(6), 865–930 (2009)

    Article  MathSciNet  Google Scholar 

  11. Fock, V.V., Goncharov, A.B.: The quantum dilogarithm and representations of quantum cluster varieties. Invent. Math. 175(2), 223–286 (2009)

    Article  MathSciNet  Google Scholar 

  12. Fomin, S., Reading, N.: Root systems and seneralized associahedra. Geometric Combinatorics, IAS/Park City Math. Ser., vol. 13, pp. 63–131. American Mathematical Society, Providence, RI (2007)

  13. Fomin, S., Shapiro, M., Thurston, D.: Cluster algebras and triangulated surfaces. Part I: cluster complexes. Acta Math. 201(1), 83–146 (2008)

    Article  MathSciNet  Google Scholar 

  14. Fomin, S., Zelevinsky, A.: Double Bruhat cells and total positivity. J. Am. Math. Soc. 12(2), 335–380 (1999)

    Article  MathSciNet  Google Scholar 

  15. Gekhtman, M., Shapiro, M., Vainshtein, A.: Cluster algebras and Poisson geometry. In: Mathematical Surveys and Monographs, vol. 167. American Mathematical Society (2010)

  16. Ginzburg, V.L., Weinstein, A.: Lie–Poisson structure on some Poisson Lie groups. J. Am. Math. Soc. 5(2), 445–453 (1992)

    Article  MathSciNet  Google Scholar 

  17. Goodearl, K.R., Yakimov, M.T.: The Berenstein–Zelevinsky quantum cluster algebra conjecture. J. Eur. Math. Soc. (JEMS) 22(8), 2453–2509 (2020)

    Article  MathSciNet  Google Scholar 

  18. Humphreys, J.: Introduction to Lie Algebras and Representation Theory. Springer, Berlin (1972)

    Book  Google Scholar 

  19. Kac, V.: Infinite Dimensional Lie Algebras, 3rd edn. Cambridge Univ Press, Cambridge (1990)

    Book  Google Scholar 

  20. Kamnitzer, J.: Mirković–Vilonen cycles and polytopes. Ann. Math. 2(171), 731–777 (2010)

    MATH  Google Scholar 

  21. Kashiwara, M.: The crystal base and Littelman’s refined Demazure character formula. Duke Math. J. 71(3), 839–858 (1993)

    Article  MathSciNet  Google Scholar 

  22. Kashiwara, M.: On crystal bases. Representations of groups (Banff, AB, 1994). CMS Conference Proceedings, vol. 16, pp. 155–197. American Mathematical Society (1995)

  23. Kashiwara, M., Similarity of crystal bases. Lie algebras and their representations (Seoul, 1995), Contemporary Mathematics, vol. 194, pp. 177–186. American Mathematical Society (1996)

  24. Kirillov, A.A.: Lectures on the Orbit Method. American Mathematical Society, Providence (2004)

    Book  Google Scholar 

  25. Lusztig, G.: Canonical bases arising from quantized enveloping algebras. J. Am. Math. Soc. 3(2), 447–498 (1990)

    Article  MathSciNet  Google Scholar 

  26. Lu, J.H., Weinstein, A.: Poisson–Lie groups, dressing transformations and Bruhat decompositions. J. Differ. Geom. 31(2), 501–526 (1990)

    Article  MathSciNet  Google Scholar 

  27. McGerty, K.: Langlands duality for representations and quantum groups at a root of unity. Commun. Math. Phys. 296(1), 89–109 (2010)

    Article  MathSciNet  Google Scholar 

  28. Nakashima, T.: Decorated geometric crystals, polyhedral and monomial realizations of crystal bases. J. Algebra 399, 712–769 (2014)

    Article  MathSciNet  Google Scholar 

  29. Papadopoulos, A.: Handbook of Teichmüller Theory, vol. 1. European Mathematical Society, Zurich (2007)

    Book  Google Scholar 

  30. William Moore, G., Goodman, M., Barnabas, J.: An iterative approach from the standpoint of the additive hypothesis to the dendrogram problem posed by molecular data sets. J. Theor. Biol. 38(3), 423–457 (1973)

    Article  Google Scholar 

Download references

Acknowledgements

We are grateful to B. Elek, V. V. Fock, A. Goncharov, J. Lane, J. H. Lu and M. Semenov-Tian-Shansky for useful discussions, and to D. R. Youmans for his helpful comments on an earlier draft. We are indebted to an anonymous referee of this paper for careful remarks and suggestions. Research of A.A. and Y.L. was supported in part by the grant MODFLAT of the European Research Council (ERC), by the grants number 178794 and 178828 of the Swiss National Science Foundation (SNSF) and by the NCCR SwissMAP of the SNSF. B.H. was supported by the National Science Foundation Graduate Research Fellowship under Grant Number DGE-1650441. A.B. and B.H. express their gratitude for hospitality and support during their visits to Switzerland in 2017 and 2018.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Arkady Berenstein.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendices

Appendices

Example: duality between \(B_2\) and \(C_2\)

Note \(\mathrm{SO}_{2n+1}^\vee =\mathrm{Sp}_{2n}\). Let us focus on the case \(n=2\). Here we use an alternative description of \(\mathrm{SO}_5\). Denote

The group \(\mathrm{SO}_5\) is isomorphic to

$$\begin{aligned} G=\{X\in \mathrm{GL}_5\mid XJ_5X^T=J_5\}, \end{aligned}$$

with Lie algebra:

$$\begin{aligned} {\mathfrak {g}}=\{x\in \mathfrak {gl}(5)\mid x+J_5x^TJ_5=0\}. \end{aligned}$$

Cartan subalgebra:

$$\begin{aligned} {\mathfrak {h}}=\{{{\,\mathrm{diag}\,}}(x_1,x_2,0,-x_2,-x_1)\}. \end{aligned}$$

Borel subalgebra:

$$\begin{aligned} {\mathfrak {b}}={\mathfrak {g}}\cap \{\text {upper-triangular matrices}\}. \end{aligned}$$

Cartan matrix and a symmetrizer:

$$\begin{aligned} A= \begin{bmatrix} 2 &{}\quad -1 \\ -2 &{}\quad 2 \end{bmatrix}= \begin{bmatrix} 1 &{} \\ &{}\quad 2 \end{bmatrix} \begin{bmatrix} 2 &{}\quad -1 \\ -1 &{}\quad 1 \end{bmatrix}, \quad D= \begin{bmatrix} 1 &{} \\ &{}\quad 2 \end{bmatrix}, \end{aligned}$$

Orthonormal basis in \({\mathfrak {h}}^*\):

$$\begin{aligned} \zeta _i: {{\,\mathrm{diag}\,}}(x_1,x_2,0,-x_2,-x_1)\mapsto x_i. \end{aligned}$$

Simple roots:

$$\begin{aligned} \alpha _1=\zeta _1-\zeta _2,\quad \alpha _2=\zeta _2. \end{aligned}$$

Positive roots:

$$\begin{aligned} \alpha _1,\quad \alpha _2,\quad \alpha _3:=\alpha _1+\alpha _2,\quad \alpha _4:=\alpha _1+2\alpha _2. \end{aligned}$$

Simple coroots:

$$\begin{aligned} \alpha _1^\vee ={{\,\mathrm{diag}\,}}(1,-1,0,1,-1); \quad \alpha _2^\vee ={{\,\mathrm{diag}\,}}(0,2,0,-2,0). \end{aligned}$$

Simple root vectors:

$$\begin{aligned} F_1=E_{21}-E_{54}; \quad F_2=E_{32}-E_{43}. \end{aligned}$$

Fundamental weights:

$$\begin{aligned} \omega _1=\alpha _3, \quad \omega _2=\frac{1}{2}\alpha _4. \end{aligned}$$

Fundamental coweights:

$$\begin{aligned} \omega _1^\vee =\alpha _1^\vee +\frac{1}{2}\alpha _2^\vee , \quad \omega _2^\vee =\alpha _1^\vee +\alpha _2^\vee . \end{aligned}$$

Character lattice of the maximal torus:

$$\begin{aligned} X^*(H)={\mathbb {Z}}\{\alpha _1,\alpha _2\}. \end{aligned}$$

Cocharacter lattice of the maximal torus:

$$\begin{aligned} X_*(H)={\mathbb {Z}}\{\omega _1^\vee ,\omega _2^\vee \}. \end{aligned}$$

Weyl group:

$$\begin{aligned} W=S_2 < imes {\mathbb {Z}}_2, \, \text { with generator }\, s_1, s_2 \, \text {satisfying}\, (s_1s_2)^4=1. \end{aligned}$$

The longest element:

$$\begin{aligned} w_0=(s_1s_2)^2=(s_2s_1)^2. \end{aligned}$$

Now let us compute the BK potential and the BK cone. Note that the lift of \(s_i\) to G is given by:

$$\begin{aligned} \overline{s_1}=P_1P_4;\quad \overline{s_2}=P_2P_3P_2; \quad \text {where}\, P_i = E_{i,i+1}-E_{i+1,i}. \end{aligned}$$

And note \((\overline{s_1}\overline{s_2})^2=P_1P_2P_3P_4P_1P_2P_3P_1P_2P_1\). Let

$$\begin{aligned} \varvec{x}=\exp \left( \ln (x_1)\omega _1^\vee +\ln (x_2)\omega _2^\vee \right) \end{aligned}$$

and

$$\begin{aligned} x_{-1}(t)= \begin{bmatrix} t^{-1} &{} &{} &{} &{} \\ 1 &{}\quad t &{} &{} &{}\\ &{} &{}\quad 1 &{}&{} \\ &{}&{}&{}\quad t^{-1} &{}\\ &{}&{}&{}\quad -1&{}\quad t \end{bmatrix};\quad x_{-2}(t)= \begin{bmatrix} 1 &{} &{} &{} &{} \\ &{}\quad t^{-2} &{} &{} &{}\\ &{}\quad t^{-1} &{}\quad 1 &{} &{} \\ &{}\quad -\frac{1}{2}&{}\quad -t &{}\quad t^2 &{}\\ &{}&{}&{}&{} 1 \end{bmatrix}. \end{aligned}$$

Then for the longest word \((s_1s_2)^2\), generic elements of the double Bruhat cell \(G^{w_0,e}\) can be written:

$$\begin{aligned} \varvec{x}x_{-1}(t_1)x_{-2}(t_2)x_{-1}(t_3)x_{-2}(t_4)\in G^{w_0,e}. \end{aligned}$$

This element is equal to

$$\begin{aligned} \varvec{x} \begin{bmatrix} \dfrac{1}{t_1 t_3} &{} &{} &{} &{} \\ \dfrac{t_1}{t_2^2}+\dfrac{1}{t_3} &{}\quad \dfrac{t_1 t_3}{t_2^2 t_4^2} &{} &{} &{} \\ \dfrac{1}{t_2} &{}\quad \dfrac{t_3}{t_2t_4^2}+\dfrac{1}{t_4} &{}\quad 1 &{} &{} \\ -\dfrac{1}{2 t_1} &{}\quad -\dfrac{\left( t_3+t_2 t_4\right) {}^2}{2 t_1 t_3 t_4^2} &{}\quad -\dfrac{t_2 \left( t_3+t_2 t_4\right) }{t_1 t_3} &{}\quad \dfrac{t_2^2 t_4^2}{t_1 t_3} &{} \\ \dfrac{1}{2} &{}\quad \dfrac{1}{2} \left( \dfrac{\left( t_3+t_2 t_4\right) {}^2}{t_3 t_4^2}+t_1\right) &{}\quad \dfrac{t_4 t_2^2}{t_3}+t_2+t_1 t_4 &{}\quad -\dfrac{\left( t_2^2+t_1 t_3\right) t_4^2}{t_3} &{}\quad t_1 t_3 \\ \end{bmatrix}. \end{aligned}$$

Thus the potential is

$$\begin{aligned} \left( t_1+\frac{\left( t_3+t_2 t_4\right) ^2}{t_3t_4^2}\right) +t_4+ x_1\cdot \frac{1}{t_1}+x_2\left( \frac{1}{t_4}+\frac{t_2^2+t_1t_3}{t_2t_3}\right) , \end{aligned}$$

which gives us the cone cut out by the inequalities:

$$\begin{aligned} \begin{aligned} x_1\geqslant t_1&\geqslant 0;\\ x_2 \geqslant t_4&\geqslant 0;\\ 2t_2\geqslant t_3 \geqslant 2t_4&\geqslant 0;\\ x_2&\geqslant t_2-t_1;\\ x_2&\geqslant t_3-t_2. \end{aligned} \end{aligned}$$
(40)

Now let us describe \(\mathrm{Sp}_4\) as the dual of \(\mathrm{SO}_5\). Denote

$$\begin{aligned} J'_{2n}= \begin{bmatrix} &{} J_n \\ -J_n &{} \end{bmatrix}. \end{aligned}$$

The group \(\mathrm{Sp}_4\) is isomorphic to

$$\begin{aligned} G^\vee =\{X\in \mathrm{GL}_4\mid XJ'_4X^T=J'_4\} \end{aligned}$$

with Lie algebra:

$$\begin{aligned} {\mathfrak {g}}^\vee =\{x\in \mathfrak {gl}(4)\mid x-J'x^TJ'=0\}. \end{aligned}$$

Cartan subalgebra:

$$\begin{aligned} {\mathfrak {h}}=\{{{\,\mathrm{diag}\,}}(x_1,x_2,-x_2,-x_1)\}. \end{aligned}$$

The Borel subalgebra:

$$\begin{aligned} {\mathfrak {b}}^\vee ={\mathfrak {g}}^\vee \cap \{\text {upper-triangular matrices}\}. \end{aligned}$$

Orthonormal basis in \(({\mathfrak {h}}^\vee )^*\):

$$\begin{aligned} \zeta _i^\vee : {{\,\mathrm{diag}\,}}(x_1,x_2,-x_2,-x_1)\mapsto x_i. \end{aligned}$$

Simple roots:

$$\begin{aligned} \beta _1=\zeta _1^\vee -\zeta _2^\vee ,\quad \beta _2=2\zeta _2^\vee . \end{aligned}$$

Positive roots:

$$\begin{aligned} \beta _1,\quad \beta _2,\quad \beta _3:=2\beta _1+\beta _2,\quad \beta _4:=\beta _1+\beta _2. \end{aligned}$$

Simple coroots of \({\mathfrak {g}}^\vee \) are given by:

$$\begin{aligned} \beta _1^\vee ={{\,\mathrm{diag}\,}}(1,-1,1,-1); \quad \beta _2^\vee ={{\,\mathrm{diag}\,}}(0,1,-1,0); \end{aligned}$$

Simple root vectors:

$$\begin{aligned} F_1=E_{21}-E_{43}; \quad F_2=E_{32}. \end{aligned}$$

Fundamental weights:

$$\begin{aligned} \kappa _1=\frac{1}{2}\beta _3, \quad \kappa _2=\beta _4. \end{aligned}$$

Fundamental coweights:

$$\begin{aligned} \kappa _1^\vee =\beta _1^\vee +\beta _2^\vee , \quad \kappa _2^\vee =\frac{1}{2}\beta _1^\vee +\beta _2^\vee . \end{aligned}$$

Character lattice of the maximal torus:

$$\begin{aligned} X^*(H^\vee )={\mathbb {Z}}\{\kappa _1,\kappa _2\}. \end{aligned}$$

Cocharacter lattice of the maximal torus:

$$\begin{aligned} X_*(H)={\mathbb {Z}}\{\beta _1^\vee ,\beta _2^\vee \}. \end{aligned}$$

To calculate the potential for \(G^\vee \), we need the lift of \(s_i\) to \(G^\vee \):

$$\begin{aligned} \overline{s_1}=P_1P_3;\quad \overline{s_2}=P_2; \quad \text { where }\, P_i = E_{i,i+1}-E_{i+1,i}. \end{aligned}$$

Note \((\overline{s_1}\overline{s_2})^2=P_1P_2P_3P_1P_2P_1\). Let

$$\begin{aligned} \varvec{y}^\vee =\exp \left( \ln (y_1)\beta _1^\vee +\ln (y_2)\beta _2^\vee \right) \end{aligned}$$

and

$$\begin{aligned} x_{-1}^\vee (t)= \begin{bmatrix} t^{-1} &{} &{} &{} \\ 1 &{}\quad t &{} &{} \\ &{}&{}\quad t^{-1} &{}\\ &{}&{}\quad -1&{}\quad t \end{bmatrix};\quad x_{-2}^\vee (t)= \begin{bmatrix} 1 &{} &{} &{} \\ &{}\quad t^{-1} &{} &{} \\ &{}\quad 1 &{}\quad t &{}\\ &{}&{}&{}\quad 1 \end{bmatrix}. \end{aligned}$$

Then for the longest word \((s_1s_2)^2\), generic elements of the double Bruhat cell \(G^{\vee ;e,w_0}\) can be written:

$$\begin{aligned} \varvec{y}^\vee x_{-1}^\vee (t_1)x_{-2}^\vee (t_2)x_{-1}^\vee (t_3)x_{-2}^\vee (t_4)\in G^{\vee ;e,w_0}. \end{aligned}$$

This element is equal to

$$\begin{aligned} \varvec{y}^\vee \begin{bmatrix} \dfrac{1}{t_1 t_3} &{} &{} &{} \\ \dfrac{t_1}{t_2}+\dfrac{1}{t_3} &{}\quad \dfrac{t_1 t_3}{t_2 t_4} &{} &{} \\ \dfrac{1}{t_1} &{}\quad \dfrac{t_2}{t_1 t_3}+\dfrac{t_3}{t_1 t_4} &{}\quad \dfrac{t_2 t_4}{t_1 t_3} &{} \\ -1 &{} -t_1-\dfrac{t_2}{t_3}-\dfrac{t_3}{t_4} &{}\quad \left( -t_1-\dfrac{t_2}{t_3}\right) t_4 &{}\quad t_1 t_3 \\ \end{bmatrix}. \end{aligned}$$

Thus the potential is

$$\begin{aligned} \left( t_1+\frac{t_2}{t_3}+\frac{t_3}{t_4}\right) +t_4+ \frac{y_1^2}{y_2}\cdot \frac{1}{t_1}+ \frac{y_2^2}{y_1^2}\left( \frac{(t_1t_3+t_2)^2}{t_2t_3^2}+\frac{1}{t_4}\right) , \end{aligned}$$

which gives us the cone cut out by the following inequalities:

$$\begin{aligned} \begin{aligned} 2y_1-y_2\geqslant t_1&\geqslant 0;\\ 2y_2-2y_1 \geqslant t_4&\geqslant 0;\\ t_2\geqslant t_3 \geqslant t_4&\geqslant 0;\\ 2y_2-2y_1&\geqslant t_2-2t_1;\\ 2y_2-2y_1&\geqslant 2t_3-t_2. \end{aligned} \end{aligned}$$
(41)

Recall that \(\psi : X_*(H)\rightarrow X^*(H)\) is given by:

$$\begin{aligned} x_1\omega _1^\vee +x_2\omega _2^\vee \mapsto (x_1+x_2)\alpha _1+(x_1+2x_2)\alpha _2. \end{aligned}$$

Then the map \(\psi _{{\mathbf {i}}}: {\mathcal {L}} \rightarrow {\mathcal {L}}^\vee \) is given by:

$$\begin{aligned} (x_1,x_2;t_1,t_2,t_3,t_4)\mapsto (x_1+x_2,x_1+2x_2;t_1,2t_2,t_3,2t_4). \end{aligned}$$

Thus it easy to see, after replacing \((y_1,y_2;t_1,t_2,t_3,t_4)\) by \((x_1+x_2,x_1+2x_2;t_1,2t_2,t_3,2t_4)\), that the real cone defined by (41) is the real cone defined by (40).

Bohr–Sommerfeld lattices and tropical Poisson varieties

In this section we extend the notion of a Bohr–Sommerfeld lattice to the category of tropical Poisson varieties, which we called \(\mathbf {PTrop}\) in [1]. We will actually consider the category \(\mathbf {DecPTrop}\) of decorated tropical Poisson varieties. An object of \(\mathbf {DecPTrop}\) is a tuple \(({\mathcal {C}}\times T, X^t, \pi , {{\,\mathrm{hw}\,}}, P)\), where

  • T is a connected subgroup of \((S^1)^n\);

  • \(X^t = {{\,\mathrm{Hom}\,}}(S^1,(S^1)^n)\cong {{\,\mathrm{Hom}\,}}({\mathbb {C}}^\times ,({\mathbb {C}}^\times )^n)=(({\mathbb {C}}^\times )^n)^t\);

  • \({\mathcal {C}}\subset X^t\otimes {\mathbb {R}}\) is an open rational polyhedral cone in \(X^t\otimes {\mathbb {R}}\);

  • \(\pi \) is a constant Poisson bivector on \((X^t\otimes {\mathbb {R}})\times T\), and the projections to \(X^t \otimes {\mathbb {R}}\) and T (both equipped with the zero Poisson structure) are Poisson maps;

  • \(P\cong {\mathbb {Z}}^{n-\dim (T)}\) is a free abelian group;

  • \({{\,\mathrm{hw}\,}}:X^t\rightarrow P\) is a \({\mathbb {Z}}\)-linear map, so that fibers of the induced map \({{\,\mathrm{hw}\,}}_{{\mathbb {R}}}\circ {{\,\mathrm{pr}\,}}_1:{\mathcal {C}}\times T\rightarrow P\otimes {\mathbb {R}}\) are the symplectic leaves of \(({\mathcal {C}}\times T,\pi )\).

An example of a decorated tropical Poisson variety is

$$\begin{aligned} \left( PT(K^*),(G^{w_0,e})^t, \pi _{PT}^{\varsigma ({\mathbf {i}})}, {{\,\mathrm{hw}\,}}^t, X_*(H)\right) . \end{aligned}$$

where tropicalization is taken with respect to the chart \(\varsigma ({\mathbf {i}})\) given by (30). Here the formula for \(\pi _{PT}^{\varsigma ({\mathbf {i}})}\) is given by Theorem 6.1, and we extend \(\pi _{PT}^{\varsigma ({\mathbf {i}})}\) to be a constant bracket on all of \(((G^{w_0,e}, \varsigma ({\mathbf {i}}))^t \otimes {\mathbb {R}}) \times (S^1)^m \cong {\mathbb {R}}^{m+r} \times (S^1)^m\).

An arrow in \(\mathbf {DecPTrop}\) is defined to be a pair

$$\begin{aligned} (f,g):({\mathcal {C}}\times T,X^t, \pi , {{\,\mathrm{hw}\,}}, P) \rightarrow ({\mathcal {C}}'\times T',{X'}^t, \pi ', {{\,\mathrm{hw}\,}}', P') \end{aligned}$$

where \(f:X^t\rightarrow {X'}^t\) is a piecewise \({\mathbb {Z}}\)-linear map which is homogeneous in the sense that \(f(nx)=nf(x)\) for \(n\in {\mathbb {Z}}_{\geqslant 0}\), and \(g:P\rightarrow P'\) is a \({\mathbb {Z}}\)-linear map so that \({{\,\mathrm{hw}\,}}'\circ f=g\circ {{\,\mathrm{hw}\,}}\). We require that f induces a map of cones \(f:{\mathcal {C}}\rightarrow {\mathcal {C}}'\), and, on each open linearity chamber \(C\subset {\mathcal {C}}\) of f, the naturally induced map \(C\times (S^1)^n\rightarrow f(C)\times (S^1)^{n'}\) restricts to a Poisson map \(f:C\times T\rightarrow f(C)\times T'\).

There is an obvious forgetful functor from \(\mathbf {DecPTrop}\) to \(\mathbf {PTrop}\).

Definition B.1

For a point \(\lambda \in P\), consider the fiber \({{\,\mathrm{hw}\,}}_{{\mathbb {R}}}^{-1}(\lambda ) \subset X^t\otimes {\mathbb {R}}\). For each \(x\in {{\,\mathrm{hw}\,}}_{{\mathbb {R}}}^{-1}(\lambda )\cap X^t\), there is a lattice coming from \(\pi \) in the tangent space \(T_x ({{\,\mathrm{hw}\,}}_{{\mathbb {R}}}^{-1}(\lambda ))\), as in (34). Using the canonical identification \(T_x ({{\,\mathrm{hw}\,}}_{{\mathbb {R}}}^{-1}(\lambda )) \cong {{\,\mathrm{hw}\,}}_{{\mathbb {R}}}^{-1}(\lambda )\), we can realize this lattice as a subset \(\Lambda _x\) of \({{\,\mathrm{hw}\,}}_{{\mathbb {R}}}^{-1}(\lambda )\). A decorated tropical Poisson variety is quantizable if, for any \(\lambda \in P\) and any \(x,y\in {{\,\mathrm{hw}\,}}_{{\mathbb {R}}}^{-1}(\lambda )\cap X^t\), one has \( \Lambda _x=\Lambda _y\). If \({\mathcal {C}}\times T\) is quantizable, define the Bohr–Sommerfeld lattice as

$$\begin{aligned} \Lambda := \bigcup _{x\in {{\,\mathrm{hw}\,}}^{-1}(P)} \Lambda _x. \end{aligned}$$

Our example \(PT(K^*)\) is quantizable, and this definition of \(\Lambda \) agrees with the one in (36), after forming the intersection with the cone \({\mathcal {C}}^G_{\varsigma ({\mathbf {i}})}({\mathbb {R}})\).

Because \(\pi \) is assumed to be constant, to check that \({\mathcal {C}}\times T\) is quantizable it is enough to check that

$$\begin{aligned} X^t \cap {{\,\mathrm{hw}\,}}^{-1}_{\mathbb {R}}({{\,\mathrm{hw}\,}}(z))\subset \Lambda _z, \end{aligned}$$
(42)

for any one \(z \in X^t\). Indeed, let \(\lambda \in P\) and let \(x,y\in {{\,\mathrm{hw}\,}}_{{\mathbb {R}}}^{-1}(\lambda )\cap X^t\). Then, because \(\pi \) is constant, \(\Lambda _x = \Lambda _z + (x-z)\) and so \(X^t \cap {{\,\mathrm{hw}\,}}^{-1}_{\mathbb {R}}(\lambda )\subset \Lambda _x\). Again because \(\pi \) is constant, \(\Lambda _y = \Lambda _x + (x-y)\). So since \(x-y\in X^t\), and since we have shown \(X^t \cap {{\,\mathrm{hw}\,}}^{-1}_{\mathbb {R}}(\lambda )\subset \Lambda _x\), it follows that \(\Lambda _y= \Lambda _x\).

Lemma B.2

Let (fg) be an isomorphism of decorated tropical Poisson varieties:

$$\begin{aligned} (f,g):({\mathcal {C}}\times T,X^t, \pi , {{\,\mathrm{hw}\,}}, P) \rightarrow ({\mathcal {C}}'\times T',{X'}^t, \pi ', {{\,\mathrm{hw}\,}}', P'). \end{aligned}$$

Assume \({\mathcal {C}}\times T\) is quantizable. Then \({\mathcal {C}}'\times T'\) is quantizable. If \(\Lambda '\) denotes the Bohr–Sommerfeld lattice of \({\mathcal {C}}'\times T'\), then \(f(\Lambda \cap \overline{{\mathcal {C}}})=\Lambda ' \cap \overline{{\mathcal {C}}'}\).

Proof

Without loss of generality assume that \(P=P'\) and \(g={{\,\mathrm{Id}\,}}\). First, we show that \({\mathcal {C}}'\times T'\) is quantizable. Let \(\lambda \in {{\,\mathrm{hw}\,}}({\mathcal {C}}\cap X^t)\) and let \(x\in {{\,\mathrm{hw}\,}}_{{\mathbb {R}}}^{-1}(\lambda ) \cap X^t\) be a point which is inside an open linearity chamber C of f. Note that C is a cone because f is assumed to be homogeneous. We may assume we have chosen \(\lambda \) and x such that the lattice generators of \(\Lambda _x\) are contained in C.

Let \(C_\lambda = C\cap {{\,\mathrm{hw}\,}}_{{\mathbb {R}}}^{-1}(\lambda )\), and let \(L=C_\lambda \times T\). Then f induces a symplectomorphism of L onto its image, and we have \(f(\Lambda _x\cap C_\lambda )=\Lambda _{f(x)}'\cap f(C_\lambda )\). Since f is an isomorphism, one has \(f(X^t\cap C_\lambda )={X'}^t\cap f(C_\lambda )\). And since \({\mathcal {C}}\times T\) is quantizable, one has \(X^t \cap C_\lambda \subset \Lambda _x \cap C_\lambda \). Therefore \({X'}^t \cap f(C_\lambda )\subset \Lambda _{f(x)}'\cap f(C_\lambda )\). Since we chose x so that the lattice generators of \(\Lambda _x\) are contained in C, we can extend to all of \({{{\,\mathrm{hw}\,}}_{\mathbb {R}}'}^{-1}(\lambda )\). We then have

$$\begin{aligned} {X'}^t\cap {{{\,\mathrm{hw}\,}}'_{{\mathbb {R}}}}^{-1}(\lambda ) \subset \Lambda _{f(x)}'. \end{aligned}$$

By the criterion (42), since \(\pi '\) is constant, this tells us that \({\mathcal {C}}'\times T\) is quantizable.

Now, let \(\Lambda '\) be the Bohr–Sommerfeld lattice of \({\mathcal {C}}'\times T'\). We will show that \(f(\Lambda )=\Lambda '\). It is enough to check, for each open linearity cone C of f, that \(f(\Lambda \cap C)=\Lambda '\cap f(C)\); one can then extend to the boundary of C and f(C) by linearity. For this it is enough to check that \(f(\Lambda _x\cap C)=\Lambda '_{f(x)}\cap f(C)\) for all \(x\in C\cap X^t\). But this follows because f induces a Poisson isomorphism from \(C\times T\) to its image. \(\square \)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Alekseev, A., Berenstein, A., Hoffman, B. et al. Langlands duality and Poisson–Lie duality via cluster theory and tropicalization. Sel. Math. New Ser. 27, 69 (2021). https://doi.org/10.1007/s00029-021-00682-x

Download citation

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s00029-021-00682-x

Keywords

Mathematics Subject Classification

Navigation