arXiv:2609.39452
From Wavefunction Regularity to Eigenvector Conditioning: Accuracy--Cost Trade-offs of Quantum Algorithms for Non-Hermitian Transcorrelated Hamiltonians
Cheng-Lin Hong, Emiel Koridon, Paul K. Faehrmann, Thomas D. Kühne, Stefano Polla, Werner Dobrautz
Abstract
The electron--electron Coulomb singularity forces cusp structure in electronic wavefunctions. This cusp structure limits wavefunction regularity and slows the convergence of expansions in smooth bases, requiring large basis sets to reach a target accuracy. The transcorrelated (TC) method incorporates this short-range behavior into the Hamiltonian, thereby improving the regularity of the transformed wavefunction and accelerating convergence toward the complete-basis-set limit. The resulting Hamiltonian is non-Hermitian and non-normal, which limits the direct applicability of standard quantum algorithms. For such non-normal operators, the complexity of several recent quantum algorithms depends explicitly on quantities related to eigenvector conditioning and eigenvalue sensitivity. Understanding these quantities is essential for assessing both algorithmic complexity and quantum resource requirements. In this work, we address these issues in two stages. We first study a solvable model, which provides a controlled setting for isolating how transcorrelation affects basis convergence and eigenvector conditioning. Within this setting, we investigate two distinct one-parameter Jastrow correlators. We derive the asymptotic energy-error scaling: $N^{-1/2}$ for the bare projection and $N^{-3/2}$ for both TC correlators at generic fixed parameters, with $N$denoting the basis size. Furthermore, we show that this faster convergence can yield a polynomial improvement in the asymptotic query-complexity upper bound. We then extend the analysis to second-quantized electronic-structure Hamiltonians and use particle-number-sector diagnostics to distinguish target-sector from auxiliary-sector conditioning. Our work provides a framework for understanding the central accuracy--cost trade-off in TC Hamiltonians.
AI Summary
Transcorrelation incorporates electron-coalescence cusps into a similarity-transformed Hamiltonian, accelerating basis convergence but introducing non-normality. This paper asks when the accuracy gain outweighs the eigenvector-conditioning costs entering non-Hermitian quantum eigenvalue estimation. The analysis starts with a harmonic oscillator plus a delta-function interaction, projected onto even oscillator states, comparing exponential and erfc-based Jastrow correlators. Local wavefunction expansions and a two-sided projection-error analysis give bare energy error $O(N^{-1/2})$ versus generically $O(N^{-3/2})$ after transcorrelation, conditional on uniform reduced-inverse stability. Explicit leading prefactors distinguish the correlators; the exponential form permits cancellation of the leading term. Assuming real, diagonalizable projected spectra, block-encoding normalization $O(N)$, and eigenvector conditioning $O(N^a)$, the resulting query bounds are $\widetilde O(\varepsilon^{-3})$ bare and $\widetilde O(\varepsilon^{-(5+2a)/3})$ transcorrelated, improving the exponent when $a<2$. Small-system calculations on He, Be, and LiH examine accuracy, Pauli one-norm, reference overlap, and conditioning. Orthonormalizing within degenerate eigenspaces removes arbitrary solver-basis inflation, while particle-number restriction separates physical-sector from auxiliary-sector conditioning. Be/STO-6G examples match bare cc-pVDZ accuracy with roughly tenfold smaller Pauli one-norm. The xTC approximation reduces Pauli-term counts with maximum tested energy deviations of $1.01\times10^{-4}$ Ha for Be and $1.2\times10^{-5}$ Ha for LiH. Asymptotic conditioning growth is not established. Non-nested molecular bases limit scaling conclusions, and larger He projections exhibit complex excited eigenvalues despite real ground energies. Target-sector xTC conditioning and end-to-end quantum resource advantages remain unresolved.
were used for technical discussion, literature searches, proof checking, writing support, and developing and checking analysis, verification, and plotting scripts.
Central claim may not hold: The abstract states the generic N^{-3/2} transcorrelated energy-error scaling without disclosing that its proof assumes, rather than establishes, convergence and uniform reduced-inverse stability of the projected ground-state eigenvalue branch.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 35 + 29 pages, 15 figures, 1 table