en.Wedoany.com Reported - IBM Quantum researchers have launched the open-source Python library ffsim, designed for efficient classical simulation of fermionic quantum circuits, aiming to provide faster verification and benchmarking tools for the quantum information community.
Unlike general-purpose quantum circuit simulators that rely on storing the full state vector, ffsim leverages the strictly conserved total particle number and spin z-component (Sz) symmetry in physical systems to significantly compress the computational space. While a general simulator requires maintaining a complex vector of dimension 2n for an n-qubit circuit, ffsim, for 2N spin orbitals, only tracks states containing a fixed number of spin-up electrons (Nα) and spin-down electrons (Nβ), thereby substantially reducing the numerical prefactor.
This engineering impact is demonstrated in benchmarks of the two-dimensional Hubbard model on a 4×8 lattice configuration. Under this model, the standard fermion-to-qubit mapping translates into a 64-qubit circuit. Executing this model requires approximately 256 exbibytes (EiB) of memory for a general state vector simulator, whereas ffsim, at 1/8 filling, compresses the active state vector to 19.3 gibibytes (GiB), enabling researchers to perform exact simulations on a single classical desktop workstation.
The library is based on a functional programming paradigm, uses NumPy arrays to store and evolve fermionic wavefunctions, integrates an optimized backend, supports a set of particle-conserving gates, and includes native interfaces with the Qiskit and PySCF ecosystems. Through the Qiskit compilation layer, ffsim can also serve as a simulation backend for any arbitrary qubit circuit composed of Hamming weight-preserving gates.
The development team benchmarked ffsim against the Fermionic Quantum Emulator (FQE) and Qiskit Aer. On a single-threaded M1 MacBook, tests of Trotterized time evolution for molecular Hamiltonians in double-factorized representation showed that ffsim simulates 11 times faster than FQE, with a 2.4x speedup for quadratic Hamiltonian evolution and an 8.4x speedup for molecular Hamiltonian operator application. When the system scale was extended to 16 orbitals (32 qubits), Qiskit Aer could not handle it due to classical memory constraints, while ffsim continued to run. A detailed technical preprint has been published on arXiv, with related code and tutorials available in the Qiskit Community GitHub repository, and an integration overview of the library has been published on the IBM Quantum Blog.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
