← All problems

Generic Semisimplicity of the Big Quantum Cohomology of Flag Varieties

28 years old · traced to

The big quantum cohomology of G/P is generically semisimple.

References

N. Perrin and M. N. Smirnov, On the big quantum cohomology of coadjoint varieties, Selecta Math. 31 (2025), Article 30. https://doi.org/10.1007/s00029-025-01017-w

J. A. Cruz Morales, A. Mellit, N. Perrin, and M. Smirnov, with an appendix by A. Kuznetsov, On quantum cohomology of Grassmannians of isotropic lines, unfoldings of An-singularities, and Lefschetz exceptional collections, Ann. Inst. Fourier (Grenoble) 69 (2019), no. 3, 955–991; doi:10.5802/aif.3263.

S. Galkin, A. Mellit, and M. Smirnov, Dubrovin’s conjecture for IG(2, 6), Int. Math. Res. Not. IMRN (2015), no. 18, 8847–8859; doi:10.1093/imrn/rnu205.

H. Iritani and Y. Koto, ⁠Quantum cohomology of projective bundles⁠, arXiv:2307.03696⁠.

https://www.grassmannian.info/

Progress summary

Refreshed
Open · solution reported

A temporary manuscript claims a complete proof for all flag varieties, but it is explicitly unverified and the established literature covers only several families.

The conjecture, recorded by Perrin and Smirnov, asserts generic semisimplicity of the big quantum cohomology of every homogeneous space G/PG/P. A temporary manuscript dated August 13, 2026 claims this in full generality, but its argument has not been independently verified.

Known results

  • Ordinary and symplectic partial flag varieties: formal generic semisimplicity (Iritani–Koto, 2023).
  • Coadjoint varieties: generic semisimplicity in all Dynkin types, with new cases in types DnD_n, E6E_6, E7E_7, and E8E_8 (Perrin–Smirnov, 2025).
  • IG⁡(2,2n)\operatorname{IG}(2,2n): regularity and hence generic semisimplicity, despite non-semisimple small quantum cohomology (Cruz Morales–Mellit–Perrin–Smirnov, 2017).

Posted attempt — August 2026

Changzheng Li, Zhaoyang Liu, Tony Yue Yu, and Chi Zhang claim a formal proof for arbitrary flag varieties, via a Landau–Ginzburg unfolding whose spectral cover is regular and generically finite étale. The claim is a complete result, but the temporary manuscript says it is still being checked and revised; it has not been independently verified.

Current status (as of August 2026): Several substantial families are proved, while the full assertion for all G/PG/P is supported only by an unverified temporary manuscript claim.

Sources

Solutions 1

The proof is now posted on arXiv: https://arxiv.org/abs/2610.28418v1See full solutionHide full solution

Convergence and semisimplicity of big quantum cohomology of flag varieties

Changzheng Li, Zhaoyang Liu, Tony Yue Yu, and Chi Zhang

The current preprint is available at https://arxiv.org/abs/2610.28418. It establishes convergence and generic semisimplicity, together with regularity of the formal big quantum algebra after extending the Novikov coefficients to a field. The statement and proof outline below update the earlier preliminary account.

Main result. Let X=G/PX=G/P, where GG is a simply connected complex simple Lie group and PP is a parabolic subgroup. Put

r=dim⁡H2(X,C),N=dim⁡H∗(X,C).r=\dim H^2(X,\mathbb C),\qquad N=\dim H^*(X,\mathbb C).

Then the big quantum product of XX converges near the large-radius limit. On every sufficiently small large-radius polydisc, the semisimple locus of the resulting analytic quantum algebra is open and dense.

More precisely, set

SX=C(q1,…,qr)[[τ1,…,τN]],KX=Frac⁡(SX),S_X=\mathbb C(q_1,\ldots,q_r)[[\tau_1,\ldots,\tau_N]], \qquad K_X=\operatorname{Frac}(S_X),

and equip

QX=H∗(X,C)⊗CSX\mathcal Q_X=H^*(X,\mathbb C)\otimes_{\mathbb C}S_X

with the big quantum product. The ring QX\mathcal Q_X is regular and finite free of rank NN over SXS_X. Its generic fiber

QX⊗SXKX\mathcal Q_X\otimes_{S_X}K_X

is finite étale over KXK_X. Over an algebraic closure K‾X\overline K_X, it is therefore isomorphic to K‾XN\overline K_X^N.

Convergence and generic semisimplicity, both formally and analytically, also hold when GG is any connected complex semisimple Lie group.

Proof outline.

  1. Convergence by equivariant reconstruction. Choose a generic line λ=ζξ0\lambda=\zeta\xi_0 in the equivariant parameter space. For ζ≠0\zeta\ne0, a suitable divisor DD generates the specialized equivariant cohomology, so that
1,D,…,DN−11,D,\ldots,D^{N-1}

form a basis. A parameter-uniform version of Iritani’s reconstruction theorem then gives convergence of the equivariant big quantum product on a common polydisc, uniformly for ζ\zeta on a fixed circle ∣ζ∣=δ|\zeta|=\delta.

After absorbing divisor coordinates into qaeτaq_ae^{\tau_a}, each multiplication-matrix entry has an expansion

∑JfJ(ζ)xJ,fJ(ζ)∈C[ζ].\sum_J f_J(\zeta)x^J, \qquad f_J(\zeta)\in\mathbb C[\zeta].

Cauchy’s formula gives

fJ(0)=12πi∫∣ζ∣=δfJ(ζ)dζζ,∣fJ(0)∣≤sup⁡∣ζ∣=δ∣fJ(ζ)∣.f_J(0)=\frac{1}{2\pi i} \int_{|\zeta|=\delta}f_J(\zeta)\frac{d\zeta}{\zeta}, \qquad |f_J(0)|\leq\sup_{|\zeta|=\delta}|f_J(\zeta)|.

The uniform convergence estimates therefore pass to the non-equivariant coefficients fJ(0)f_J(0). This proves convergence near the large-radius limit.

  1. Regularity from big mirror symmetry. Work over k=C(q1,…,qr)k=\mathbb C(q_1,\ldots,q_r). Let WW be Rietsch’s mirror superpotential and let FF be its formal unfolding supplied by the big quantum DD-module mirror theorem of Hinault–Li–Yu–Zhang–Zhang. Write
S=k[[y1,…,yN]],A=Jac⁡(F).S=k[[y_1,\ldots,y_N]],\qquad \mathcal A=\operatorname{Jac}(F).

The mirror theorem gives finite freeness of A\mathcal A over SS. The central parameter derivatives

gi=∂yiF∣y=0g_i=\left.\partial_{y_i}F\right|_{y=0}

have classes forming a basis of Jac⁡(W)\operatorname{Jac}(W).

At a critical point pp, differentiation induces a surjection

Jac⁡(W)⊗kκ(p)⟶coker⁡(Hess⁡p(W)).\operatorname{Jac}(W)\otimes_k\kappa(p) \longrightarrow \operatorname{coker}\bigl(\operatorname{Hess}_p(W)\bigr).

Indeed, differentiating an element of the Jacobian ideal at pp gives an element of the Hessian image, while differentials of regular functions span the cotangent space. Consequently,

TpY⊕κ(p)N⟶Tp∗Y,(v,c)⟼Hess⁡p(W)(v)+∑ici,dgi(p)T_pY\oplus\kappa(p)^N\longrightarrow T_p^*Y, \qquad (v,c)\longmapsto \operatorname{Hess}_p(W)(v)+\sum_i c_i,dg_i(p)

is surjective.

Thus the critical equations of FF have independent linear terms in the completed ambient local ring. They form part of a regular system of parameters, and their quotient is regular. This proves regularity of A\mathcal A. Its generic fiber is a finite-dimensional regular algebra over a characteristic-zero field, hence finite étale.

The multiplication compatibility in the big quantum DD-module mirror theorem identifies A\mathcal A with the pullback of QX\mathcal Q_X under the formal mirror map. Regularity and generic finite étaleness therefore pass to QX\mathcal Q_X.

  1. Analytic generic semisimplicity. In a fixed cohomology basis σ1,…,σN\sigma_1,\ldots,\sigma_N, let CiC_i denote multiplication by σi\sigma_i. Consider the trace discriminant
D(q,τ)det⁡(Tr⁡(CiCj))1≤i,j≤N.\mathcal D(q,\tau) \det\bigl(\operatorname{Tr}(C_iC_j)\bigr)_{1\leq i,j\leq N}.

Generic finite étaleness implies that this formal series is nonzero. Convergence makes it holomorphic near the large-radius limit. Its nonvanishing locus is open and dense, and is precisely the semisimple locus.

For a general semisimple group, the flag variety is a product of flag varieties for simple groups. Convergence follows from the product theory of quantum cohomology. On the parameter locus pulled back from the factors, the quantum algebra is their tensor product. Semisimple points on this locus show that the discriminant of the full family is nonzero.

References.

  • C. Li, Z. Liu, T. Y. Yu, and C. Zhang, Convergence and semisimplicity of big quantum cohomology of flag varieties, arXiv:2610.28418. https://arxiv.org/abs/2610.28418
  • T. Hinault, C. Li, T. Y. Yu, C. Zhang, and S. Zhang, Unfolding of equivariant F-bundles and application to the mirror symmetry of flag varieties, arXiv:2505.09950. https://arxiv.org/abs/2505.09950
  • H. Iritani, Convergence of quantum cohomology by quantum Lefschetz, J. Reine Angew. Math. 610 (2007), 29–69.

AI disclosure. ChatGPT assisted with the proofs at an early stage, particularly the formal semisimplicity argument, and with final proofreading, as disclosed in the preprint.