Enhancing Chemistry on Quantum Computers with Fermionic Linear Optical Simulation

Zack Hassman † , Oliver Reardon-Smith ‡ , Gokul Subramanian Ravi ∗ , Frederic T. Chong † , Kevin J. Sung § · 2025

Read Enhancing Chemistry on Quantum Computers with Fermionic Linear Optical Simulation by Zack Hassman † , Oliver Reardon-Smith ‡ , Gokul Subramanian Ravi ∗ , Frederic T. Chong † , Kevin J. Sung § (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-12416.txt.

Introduction

Zack Hassman and four co-authors present ExtraFerm, an open-source quantum circuit simulator tailored to chemistry problems. It computes the chances of measurement outcomes for circuits built from particle-number-conserving matchgates plus controlled-phase gates. Ordinary state vector methods cannot reach some of the systems this class covers.

ExtraFerm supports both exact calculation and an approximate mode. In that mode the cost grows only with the size of the controlled-phase angles. The team demonstrates it on a cluster-based wavefunction method combined with sample-based diagonalization, improving molecular ground-state energy estimates at negligible extra cost. Against tensor network and state vector simulators, they report far better memory scaling in its target regime.

Catalog record

CREATORZack Hassman † , Oliver Reardon-Smith ‡ , Gokul Subramanian Ravi ∗ , Frederic T. Chong † , Kevin J. Sung §
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE53 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-12416

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Enhancing Chemistry on Quantum Computers with Fermionic Linear Optical Simulation Zack Hassman † , Oliver Reardon-Smith ‡ , Gokul Subramanian Ravi ∗ , Frederic T. Chong † , Kevin J. Sung § Abstract We present and open source a quantum circuit simulator tailored to chemistry applications. More specifically, our simulator can compute the Born-rule probabilities of samples obtained from circuits containing passive fermionic linear optical elements and controlled-phase gates. We support both approximate and exact calculation of probabilities, and for approximate probability calculation, our simulator’s runtime is exponential only in the magnitudes of the circuit’s controlled-phase gate angles. This makes our simulator useful for simulating certain systems that are beyond the reach of conventional state vector methods. We demonstrate our simulator’s utility by simulating the local cluster unitary Jastrow (LUCJ) ansatz and integrating it with sample-based quantum diagonalization (SQD) to improve the accuracy of molecular ground-state energy estimates. Applied to a 52-qubit N 2 system, we observe accuracy improvements of up to 46% over the baseline SQD implementation with negligible computational overhead. More generally, we highlight a regime in which our simulator achieves substantially superior latency scaling and exponentially superior memory scaling over a tensor network simulator and a state vector simulator. As an efficient and flexible tool for simulating quantum chemistry circuits, our simulator enables new opportunities for enhancing near-term quantum algorithms in chemistry and related domains. I Introduction Among the most promising near-term applications of quantum computing is quantum computational chemistry [ 11 , 34 ] , which has attracted considerable attention due to potentially transformative applications in drug development [ 47 ] and materials science [ 1 ] , among other fields. Despite the excitement around this application, hardware demonstrations have been limited by the noise present in current quantum processors. Quantum error correction [ 18 , 17 , 49 , 51 , 10 , 28 ] offers a long-term solution to faulty hardware, but a fully fault-tolerant quantum computer has yet to be built. In the near-term, quantum error mitigation techniques [ 52 , 29 , 22 , 30 , 12 ] can improve results obtained from noisy quantum processors, in which many of these techniques involve classical simulation of quantum circuits [ 13 , 16 , 15 ] . Because classical simulation of quantum circuits is intractable in general, these simulations must introduce approximations, or are exact only when applied to specific narrow classes of circuits. Simulation methods tailored to circuits relevant to quantum chemistry may enable new techniques for enhancing the performance of chemistry workloads on quantum computers. Fig. 1: The green star indicates the class of circuits that ExtraFerm is designed to address: circuits containing m matchgates and k controlled-phase gates where k\ll m . This work concentrates on particle number-conserving circuits. Note that matchgates + controlled-phase gates form a universal gate set. In this work, we introduce ExtraFerm 1 1 1 https://github.com/zhassman/ExtraFerm , an open-source quantum circuit simulator that can be used to compute the Born-rule probabilities of samples (bitstrings) from circuits composed of passive fermionic linear optical elements, also known as particle number-conserving matchgates, and controlled-phase gates. These circuits are highly relevant to the field of quantum computational chemistry for simulating quantum many-body systems. Fig. 1 shows where these circuits sit relative to other well-studied gate families. ExtraFerm can perform both exact and approximate probability calculations for measurements in the output distributions of these circuits. For approximate calculation, the only exponential component of ExtraFerm’s runtime is a linear dependence in the extent of the quantum circuit it is simulating. The extent is larger for controlled phase gates with larger phases and is multiplicative as more controlled phase gates are added (see Refs. [ 42 , 14 , 43 ] ). For exact probability calculation, ExtraFerm’s runtime is exponential only in the number of controlled-phase gates. Notably, this means that for both approximate and exact calculation, ExtraFerm’s performance is polynomial in the number of qubits and matchgates. Unlike conventional state vector simulation, which tracks the amplitudes of all bitstrings, ExtraFerm computes Born-rule probabilities for a pre-specified subset of the output distribution. Indeed, for small systems, one can use ExtraFerm to simulate the entire support of a circuit, but we do not expect this to be ExtraFerm’s most compelling use case. Instead, we propose ExtraFerm as a unique tool that can be embedded in quantum-classical workflows to recover signal from noisy samples from large, application-scale quantum circuits. We apply ExtraFerm to the simulation of the local unitary cluster Jastrow (LUCJ) ansatz [ 36 ] , which has recently been adopted for diverse applications in quantum simulation of chemical systems [ 3 , 48 , 26 , 31 , 44 ] . When mapped to a quantum circuit using the Jordan-Wigner transformation [ 37 ] , the LUCJ ansatz decomposes into particle number-conserving matchgates and controlled-phase gates, making it amenable to simulation by ExtraFerm in certain regimes. To highlight the regimes in which ExtraFerm may be useful, we use LUCJ circuits to study the error and latency performance of ExtraFerm. Note that LUCJ circuits are flexible in the sense that repeated layers of them can be used to implement arbitrary particle number-conserving extended matchgate circuits, providing us with a generic but practically useful benchmark. Beyond this, Section V-A show that ExtraFerm offers significantly lower latency and exponentially better memory performance compared to both a tensor network and state vector simulator. Finally, as an end-to-end practical use case, we show how to use ExtraFerm to boost the performance of sample-based quantum diagonalization (SQD) [ 44 ] . SQD is an extension of quantum-selected configuration interaction (QSCI) [ 27 ] , an algorithm that samples configurations from a quantum computer and uses them to select a subspace in which to diagonalize a molecular Hamiltonian. SQD adds an error mitigation procedure called configuration recovery that attempts to correct sampled bitstrings that were affected by noise and do not satisfy symmetries of the system. Configuration recovery greatly improves performance on noisy quantum processors, leading to the adoption of SQD for some of the largest demonstrations of quantum chemistry on quantum processors to date [ 31 , 26 , 48 ] . We improve SQD by using ExtraFerm to select high-probability bitstrings during the early iterations of configuration recovery to provide better orbital occupancy information. We call this approach the “warm-start” variant of SQD and find that, compared to the original implementation of SQD, we obtain higher-accuracy energy estimates. For our main result, we consider a 52-qubit system sampled on an IBM Heron quantum processor and observe accuracy improvements of up to 46.09% with respect to the heat-bath configuration interaction (HCI) energy taken as a reference. Across repeated trials, we find that the warm-start variant enjoys a variance reduction of up to 98.34% . Furthermore, the overhead of the warm-start variant is minimal—at worst accounting for an extra 2.03% of runtime. For completeness, we also report results for a 28-qubit system simulated classically in the presence of global depolarizing noise and observe similar improvements. The main contributions of this work are the following: 1. We release ExtraFerm, an extended matchgate simulator for particle number-conserving circuits allowing for high-performance calculation of Born-rule probabilities with tunable accuracy. 2. We evaluate latency

Read the complete text