Comparing the Performance of Leading VQE Algorithms for Computing Ground-State Energies of Amino Acids

Jul 16, 2026

Simulating molecules is a central application of quantum computing, offering a route beyond the exponential scaling limits of classical simulation. The work introduces an integrated, reproducible benchmark repository covering 10+ variational ansatz and two truncation strategies for mapped Hamiltonians. The framework measures performance across multiple axes, including variance and computational time, and evaluates amino-acid ground-state energies using QMProt Dataset Hamiltonians via four dedicated experimental studies.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages26

Reading Time1 h 6 min

Introduction

Molecular simulation challenges and the Electronic Structure Problem

Classical benchmarks and accuracy targets

Quantum approaches in the NISQ era

Variational Quantum Eigensolver (VQE) motivation

arXiv :2607 .02620v1 [ quant-ph] 2 Jul 2026 Comparing the Performance of Leading VQE Algorithms for Computing Ground-State Energies of Amino Acids Sanskriti Shindadkar 1 , Clyde Villacrusis2,3 , Jasper Andrews4 , and Brandon Yan2 1 Department of Bioengineering, UCLA Samueli School of Engineering 2 Department of Computer Science, UCLA Samueli School of Engineering 3 Department of Linguistics, UCLA College of Letters and Science 4 Department of Mathematics, UCLA College of Letters and Science July 1, 2026 Abstract Simulating molecules is a major application of quantum computing, with the potential to overcome exponential scaling constraints of classical computation. Researchers use different methods in order to evaluate the readiness of NISQ computers in order to test current simulation capabilities. We present an integrated repository with reproducible benchmarks of over 10 different ansatz from published papers and two different truncation methods, applicable to any set of mapped hamiltonians, providing a single pipeline for comparing performance along multiple axes, including variance and computational time, among others. We apply them to simulate different amino acids, using hamiltonians taken from the QMProt Dataset. We then ran four separate experiments. First, we quantified noise resilience by optimizing the same hardware-efficient ansatz under identical initialization while sweeping PennyLane noise channels and strengths, and measuring parameter drift, cosine similarity of optimal parameters, and energies evaluated on noiseless versus noisy backends. We then studied barren-plateau–related trainability via gradient-variance diagnostics and optimization trajectories across initialization strategies and ansatz depth on small systems. We then compared adaptive versus fixed ans¨atze at matched parameter budgets, reporting outer-loop iterations, wall time, and especially total cost-function evaluations to fairly contrast greedy adaptive growth with layered hardware-efficient circuits. Lastly, we mapped accuracy versus expressive capacity by sweeping the number of retained adaptive operators and recording ground-state energy error relative to classical references. 1 Introduction The accurate simulation of molecular systems remains one of the most formidable challenges in computational science. At the heart of this challenge lies the Electronic Structure Problem (ESP): solving the Schr¨odinger equation to determine the ground-state energy (GSE) of a many-electron system. Classical methods, while sophisticated, face an exponential scaling wall; as the molecular system size increases, the number of possible electronic configurations grows combinatorially, rendering exact solutions for large biological molecules computationally intractable. While chemical accuracy (about 4.184 kJ/mol) is enough for many molecular calculations, some scientific questions require higher accuracy. For instance, reliable polymorph ranking can require sub-chemical accuracy (a few tenths of a kJ/mol)[1], making greater accuracy of great interest for pharmaceutical developers. A variety of classical Density Functional Theory (DFT) is being used for for many molecular ground-state energy calculations and datasets. Ramakrishnan et al. [2] computed equilibrium geometries and many properties, including free energy for about 134k stable CHONF molecules using the B3LYP density functional with the 6-31G basis set, making a reference dataset [2] . Later, Narayanan et al. [3] computed G4MP2 energies for about 133,000 QM9 organic molecules and benchmarked G4MP2 enthalpies of formation against 459 molecules with an accuracy of 0.79 kcal/mol for G4MP2 enthalpies of formation [3] . Researchers are also beginning to explore Large Language Models (LLMs) for molecular and materials property prediction. For crystalline materials, Niyongabo Rubungo et al. [4] proposed LLM-Prop, a T5-based model that predicts crystal properties from text description

FAQs

What problem does the document address in quantum molecular simulation?

It targets accurate ground-state energy computation for molecular electronic structure, focusing on the Electronic Structure Problem and the scaling limits of classical methods.

How does the proposed benchmark repository enable algorithm comparison?

It provides a single reproducible pipeline that supports 10+ ansatz from published work and two truncation methods, enabling fair comparisons across metrics like variance, computational time, noise effects, and cost-function evaluations.

What experiments are conducted to evaluate VQE performance?

The study runs four experiments: noise resilience via noise-channel sweeps, trainability via gradient-variance diagnostics and optimization trajectories, adaptive versus fixed ansatz under matched budgets, and accuracy versus expressive capacity by varying retained adaptive operators.

Jul 16, 2026

Upload by Emma Wilson