pith. sign in

arxiv: 2603.21009 · v2 · pith:6ECZR5ENnew · submitted 2026-03-22 · 🪐 quant-ph · math.RT

Lie-algebraic incompleteness of symmetry-adapted VQE for non-Abelian molecular point groups

Leon D. da Silva , Marcelo P. Santos This is my paper

Pith reviewed 2026-05-21 10:23 UTC · model grok-4.3

classification 🪐 quant-ph math.RT
keywords symmetry-adapted VQEnon-Abelian point groupsdynamical Lie algebraunitary coupled clustervariational quantum eigensolverirreducible representationsgradient plateaumolecular symmetry
0
0 comments X

The pith

Symmetry-adapted VQE for non-Abelian point groups confines the dynamical Lie algebra to an Abelian subalgebra and restricts reachable states to a measure-zero torus.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that symmetry-adapted Unitary Coupled-Cluster VQE fails for non-Abelian molecular point groups when adapted only to Abelian subgroups. This practice induces a spurious splitting of multidimensional irreducible representations, which discards cross-component excitations before they can be reached. At the Lie-algebraic level the filter reduces the Dynamical Lie Algebra to the Abelian subalgebra u(1) raised to d_lambda, so the variational manifold collapses onto a torus of measure zero. A separate numerical obstruction arises because Abelian-adapted orbitals make cross-component integrals vanish, producing zero-gradient plateaus along the missing directions. The NH3 calculation with C3v symmetry illustrates both effects, converging to an energy 21.8 mHa above full configuration interaction despite optimizer convergence.

Core claim

The Abelian-subgroup restriction in symmetry-adapted UCCSD induces a spurious splitting of multidimensional irreps, confining the DLA to the Abelian subalgebra u(1)^{d_lambda} and the reachable state manifold to the torus T^{d_lambda}. Molecular orbitals adapted solely to an Abelian subgroup produce cross-component integrals that vanish identically, creating a zero-gradient plateau along non-Abelian algebraic directions. A proof-of-principle experiment on NH3/STO-3G confirms both the predicted DLA confinement and the gradient plateau, with SymUCCSD converging to an error of 21.8 mHa above the FCI energy despite full optimizer convergence.

What carries the argument

The dynamical Lie algebra confinement to the Abelian subalgebra u(1)^{d_lambda} induced by filtering the symmetry-adapted ansatz through an Abelian subgroup.

If this is right

  • Cross-component excitations between split irrep components are systematically excluded from the variational manifold.
  • The reachable states lie on a lower-dimensional torus whose dimension is set by the Abelian subalgebra rank.
  • Optimizer trajectories encounter flat regions along any direction outside the Abelian subalgebra.
  • Recovering full symmetry-equivariant dynamics requires both complete off-diagonal generators and independent parameters for cross-component excitations.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same Abelian filtering may limit other variational ansatze that rely on partial symmetry projection.
  • Full non-Abelian adaptation would enlarge the reachable manifold and remove the gradient plateau in symmetric molecules.
  • Direct computation of DLA dimension before and after adding off-diagonal generators provides a practical test for any given point group.

Load-bearing premise

Molecular orbitals adapted only to an Abelian subgroup of a non-Abelian point group make all cross-component integrals vanish identically.

What would settle it

Measure the dimension of the generated dynamical Lie algebra and check whether the gradient with respect to non-Abelian generators remains exactly zero throughout optimization on a small C3v system such as NH3/STO-3G.

Figures

Figures reproduced from arXiv: 2603.21009 by Leon D. da Silva, Marcelo P. Santos.

Figure 1
Figure 1. Figure 1: Conceptual graphical abstract diagnosing the gradient trap. [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
read the original abstract

Symmetry-adapted variational quantum eigensolvers (VQE) based on the Unitary Coupled-Cluster ansatz (SymUCCSD) effectively reduce the parameter count for Abelian molecular point groups. For non-Abelian groups, they systematically fail, without a theoretical explanation. In this work, we prove that the Abelian-subgroup restriction induces a spurious splitting of multidimensional irreducible representations, prematurely discarding cross-component excitations. At the Lie-algebraic level, this filter confines the Dynamical Lie Algebra (DLA) to the Abelian subalgebra $\mathfrak{u}(1)^{d_\lambda}$, restricting the reachable state manifold to a measure-zero torus $\mathbb{T}^{d_\lambda}$. However, completing the algebra is insufficient on its own, due to a numerical obstruction. Molecular orbitals adapted solely to an Abelian subgroup produce cross-component integrals that vanish identically, creating a zero-gradient plateau along non-Abelian algebraic directions. A proof-of-principle experiment on NH$_3$/STO-3G ($C_{3v}$, 16 qubits) confirms both the predicted DLA confinement and the gradient plateau, with SymUCCSD converging to an error of $21.8$ mHa above the FCI energy despite full optimizer convergence. Our analysis provides an algebraic and geometric diagnosis of the observed numerical breakdown, establishing that recovering full equivariant dynamics requires both the inclusion of complete off-diagonal generators and the independent parametrization of cross-component excitations.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proves that symmetry-adapted UCCSD ansatze for non-Abelian molecular point groups, when orbitals are adapted only to an Abelian subgroup, induce a spurious splitting of multidimensional irreps. This confines the dynamical Lie algebra to the Abelian subalgebra u(1)^{d_λ} and restricts the reachable manifold to a measure-zero torus T^{d_λ}. The authors further argue that completing the algebra is insufficient because Abelian-adapted orbitals cause cross-component integrals to vanish identically, producing a persistent zero-gradient plateau along non-Abelian directions. A proof-of-principle experiment on NH3/STO-3G (C_{3v}, 16 qubits) reports that SymUCCSD converges to an error of 21.8 mHa above FCI despite optimizer convergence, confirming both DLA confinement and the gradient plateau.

Significance. If the central claims hold, the work supplies a precise Lie-algebraic and geometric diagnosis of why symmetry-adapted VQE systematically fails for non-Abelian groups. It isolates two distinct obstructions—algebraic incompleteness and a numerical integral-vanishing effect—and thereby indicates the minimal requirements (full off-diagonal generators plus independent cross-component parametrization) for recovering equivariant dynamics. The explicit experiment on a small but representative system provides concrete evidence that the diagnosed mechanisms are operative in practice.

major comments (2)
  1. The numerical experiment (described in the abstract and presumably §4) demonstrates DLA confinement and the gradient plateau only for the restricted SymUCCSD ansatz. The central claim that the zero-gradient obstruction is independent of algebra completion requires an explicit additional run with the completed non-Abelian generator set on the same Abelian-adapted orbitals; without it, it remains possible that the observed plateau is an artifact of the restricted parametrization rather than a property of the orbital basis itself.
  2. The abstract states that the authors 'prove' the DLA confinement to u(1)^{d_λ}. The provided description indicates only a sketch; the full derivation (including the precise embedding of the symmetry-adapted generators into the Lie algebra and the explicit computation of the commutator closure) must be supplied in the main text with all intermediate steps, as this step is load-bearing for the geometric claim that the reachable manifold is a torus.
minor comments (2)
  1. Notation: the symbol d_λ is introduced without an explicit definition in the abstract; a short sentence clarifying that it denotes the dimension of the irrep λ would improve readability.
  2. The error value 21.8 mHa is reported to three significant figures; stating the number of optimizer iterations or the convergence threshold used would allow readers to assess whether the plateau is truly optimizer-independent.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading of the manuscript and for the constructive comments, which help clarify the presentation of our results. We address each major comment below and indicate the corresponding revisions.

read point-by-point responses

  1. Referee: The numerical experiment (described in the abstract and presumably §4) demonstrates DLA confinement and the gradient plateau only for the restricted SymUCCSD ansatz. The central claim that the zero-gradient obstruction is independent of algebra completion requires an explicit additional run with the completed non-Abelian generator set on the same Abelian-adapted orbitals; without it, it remains possible that the observed plateau is an artifact of the restricted parametrization rather than a property of the orbital basis itself.

    Authors: We agree that an explicit numerical run with the completed non-Abelian generator set on the same Abelian-adapted orbitals is required to fully substantiate the independence of the zero-gradient obstruction. In the revised manuscript we will add this experiment for the NH3/STO-3G system. The new results will demonstrate that the gradients along non-Abelian directions remain identically zero even after algebra completion, confirming that the obstruction originates from the vanishing cross-component integrals induced by the orbital basis rather than from the restricted parametrization. revision: yes

  2. Referee: The abstract states that the authors 'prove' the DLA confinement to u(1)^{d_λ}. The provided description indicates only a sketch; the full derivation (including the precise embedding of the symmetry-adapted generators into the Lie algebra and the explicit computation of the commutator closure) must be supplied in the main text with all intermediate steps, as this step is load-bearing for the geometric claim that the reachable manifold is a torus.

    Authors: We acknowledge that the current version presents the DLA confinement argument in outline form. In the revised manuscript we will expand the relevant section to supply the complete derivation, including the precise embedding of the symmetry-adapted generators into the Lie algebra and the explicit, step-by-step computation of the commutator closure. This will rigorously establish that the reachable manifold is confined to the torus T^{d_λ}. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected in derivation chain

full rationale

The paper derives the DLA confinement to the Abelian subalgebra u(1)^{d_lambda} and the measure-zero torus directly from standard representation theory applied to the Abelian-subgroup restriction on non-Abelian irreps. The numerical obstruction (vanishing cross-component integrals under Abelian-adapted orbitals) follows as a symmetry consequence of the orbital choice itself, without reducing to a fitted parameter, self-definition, or load-bearing self-citation. The NH3 experiment confirms the predicted effects for the restricted ansatz but does not create a circular loop, as the algebraic claims rest on independent group-theoretic facts rather than the numerical results. The derivation chain is self-contained against external benchmarks in Lie algebra and quantum chemistry symmetry.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The paper relies on standard concepts from dynamical Lie algebras in variational quantum algorithms and representation theory of point groups without introducing new free parameters or postulated entities.

axioms (2)
  • standard math The Dynamical Lie Algebra generated by the ansatz controls the reachable state manifold in VQE.
    Invoked to show confinement to the Abelian subalgebra u(1)^{d_lambda}.
  • domain assumption Adapting molecular orbitals only to an Abelian subgroup causes cross-component integrals to vanish.
    This is the stated source of the zero-gradient plateau along non-Abelian directions.

pith-pipeline@v0.9.0 · 5796 in / 1451 out tokens · 62126 ms · 2026-05-21T10:23:47.079570+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

  1. [1]

    A variational eigenvalue solver on a photonic quantum processor

    Alberto Peruzzo, Jarrod McClean, Peter Shadbolt, Man-Hong Yung, Xiao-Qi Zhou, Peter J. Love, Al´ an Aspuru-Guzik, and Jeremy L. O’Brien. “A variational eigenvalue solver on a photonic quantum processor”. Nat. Commun.5, 4213 (2014)

    work page 2014

  2. [2]

    Quan- tum computational chemistry

    Sam McArdle, Suguru Endo, Al´ an Aspuru-Guzik, Simon C. Benjamin, and Xiao Yuan. “Quan- tum computational chemistry”. Rev. Mod. Phys.92, 015003 (2020)

    work page 2020

  3. [3]

    Strategies for quantum computing molecular energies using the unitary coupled cluster ansatz

    Jonathan Romero, Ryan Babbush, Jarrod R. McClean, Cornelius Hempel, Peter J. Love, and Al´ an Aspuru-Guzik. “Strategies for quantum computing molecular energies using the unitary coupled cluster ansatz”. Quantum Sci. Technol.4, 014008 (2019)

    work page 2019

  4. [4]

    Barren plateaus in quantum neural network training landscapes

    Jarrod R. McClean, Sergio Boixo, Vadim N. Smelyanskiy, Ryan Babbush, and Hartmut Neven. “Barren plateaus in quantum neural network training landscapes”. Nat. Commun.9, 4812 (2018)

    work page 2018

  5. [5]

    Progress toward larger molecular simulation on a quantum computer: Simulating a system with up to 28 qubits accelerated by point-group symmetry

    Changsu Cao, Jiaqi Hu, Wengang Zhang, Xusheng Xu, Dechin Chen, Fan Yu, Jun Li, Han-Shi Hu, Dingshun Lv, and Man-Hong Yung. “Progress toward larger molecular simulation on a quantum computer: Simulating a system with up to 28 qubits accelerated by point-group symmetry”. Phys. Rev. A105, 062452 (2022)

    work page 2022

  6. [6]

    Hamiltonian-informed point group symmetry-respecting ansatz for variational quantum eigensolver

    Rongxin He, Arzigul Ablimit, Xiangjian Hong, Qiankun Chai, Jingwei Zhou, Ji Guan, Guan- glei Cui, and Shengyu Ying. “Hamiltonian-informed point group symmetry-respecting ansatz for variational quantum eigensolver” (2025). arXiv:2512.21087

    work page arXiv 2025

  7. [7]

    Introduction to quantum control and dynamics

    Domenico D’Alessandro. “Introduction to quantum control and dynamics”. Chapman and Hall/CRC. (2007)

    work page 2007

  8. [8]

    Linear representations of finite groups

    Jean-Pierre Serre. “Linear representations of finite groups”. Springer-Verlag. New York (1977)

    work page 1977

  9. [9]

    Group theoretical methods and their applications

    Albert F¨ assler and Eduard Stiefel. “Group theoretical methods and their applications”. Birkh¨ auser. Boston, MA (1992)

    work page 1992

  10. [10]

    Recent developments in the PySCF program package

    Q. Sun et al. “Recent developments in the PySCF program package”. J. Chem. Phys.153, 024109 (2020)

    work page 2020

  11. [11]

    Theory for equivariant quantum neural networks

    Quynh T. Nguyen, Louis Schatzki, Paolo Braccia, Michael Ragone, Patrick J. Coles, Fr´ ed´ eric Sauvage, Mart´ ın Larocca, and M. Cerezo. “Theory for equivariant quantum neural networks”. PRX Quantum5, 020328 (2024)

    work page 2024

  12. [12]

    Exploiting symmetry in variational quantum machine learning

    Johannes Jakob Meyer, Marian Mularski, Elies Gil-Fuster, Antonio Anna Mele, Francesco Arzani, Alena Wilms, and Jens Eisert. “Exploiting symmetry in variational quantum machine learning”. PRX Quantum4, 010328 (2023)

    work page 2023

  13. [13]

    Diagnosing barren plateaus with tools from quantum optimal control

    Martin Larocca, Piotr Czarnik, Kunal Sharma, Gopikrishnan Muraleedharan, Patrick J. Coles, and Marco Cerezo. “Diagnosing barren plateaus with tools from quantum optimal control”. Quantum6, 824 (2022)

    work page 2022

  14. [14]

    A Lie algebraic theory of barren plateaus for deep parameterized quantum circuits

    Michael Ragone, Bojko Bakalov, Marco Cerezo, and Martin Larocca. “A Lie algebraic theory of barren plateaus for deep parameterized quantum circuits”. Nat. Commun.15, 7352 (2024)

    work page 2024

  15. [15]

    An adaptive variational algorithm for exact molecular simulations on a quantum computer

    Harper R. Grimsley, Sophia E. Economou, Edwin Barnes, and Nicholas J. Mayhall. “An adaptive variational algorithm for exact molecular simulations on a quantum computer”. Nat. Commun.10, 3007 (2019)

    work page 2019

  16. [16]

    Constructing compact ADAPT unitary coupled-cluster ansatz with parameter-based criterion

    Runhong He, Xin Hong, Qiaozhen Chai, Ji Guan, Junyuan Zhou, Arapat Ablimit, Guolong Cui, and Shenggang Ying. “Constructing compact ADAPT unitary coupled-cluster ansatz with parameter-based criterion” (2026). arXiv:2602.04253. 10

    work page arXiv 2026