Accelerated Neural Network Quantum States with Probabilistic Computing
Project Description
The quantum many-body problem represents a grand challenge that is at the heart of many important problems in both basic and applied sciences including quantum chemistry and quantum computing [1]. The Hilbert space of a many-body quantum system grows exponentially with system size N, limiting traditional tensor network and quantum Monte Carlo methods to dozens of interacting particles. An important advancement in circumventing this scalability challenge is the introduction of neural quantum states (NQS) as variational wavefunctions [2] that can be optimized with powerful state-of-the-art machine learning approaches. Nevertheless, scale remains limited at this time to 100 – 1000 particles.
This project proposes to scale up NQS computations using state-of-the-art hardware acceleration approaches such as probabilistic (p-) computing [3]. A probabilistic computer of N bits naturally samples from a 2^N state space and has recently demonstrated remarkable time and energy efficiency advantages for quantum Monte Carlo problems [4] and transformer neural networks [5]. The selected candidate will evaluate the algorithmic advantage of NQS built with p-bits on a CPU and make projections about p-NQS performance on FPGA and clockless p-computers.
Start Date
January 2026
Postdoc Qualifications
PhD in Electrical and Computer Engineering, Materials Science and Engineering, Computer Science or closely related fields. Experience in quantum mechanics, quantum computing, and computational physics is preferred.
Co-advisors
Supriyo Datta (datta@purdue.edu), Elmore Family School of Electrical and Computer Engineering: https://nanohub.org/groups/supriyodatta
Arnab Banerjee (arnabb@purdue.edu), Department of Physics and Astronomy: https://www.physics.purdue.edu/people/faculty/arnabb.php
Bibliography
[1] A. Banerjee et al., Proximate Kitaev quantum spin liquid behaviour in a honeycomb magnet, Nat. Mater. 15, 733 (2016).
[2] G. Carleo and M. Troyer, Solving the quantum many-body problem with artificial neural networks, Science 355, 602 (2017).
[3] W. A. Borders, A. Z. Pervaiz, S. Fukami, K. Y. Camsari, H. Ohno, and S. Datta, Integer factorization using stochastic magnetic tunnel junctions, Nature 573, 390 (2019).
[4] S. Chowdhury, K. Y. Camsari, and S. Datta, Accelerated quantum Monte Carlo with probabilistic computers, Commun. Phys. 6, 1 (2023); Emulating quantum circuits with generalized Ising machines, IEEE Access 11, 116944 (2023).
[5] L. A. Ghantasala, M.-C. Li, R. Jaiswal, A. Ghosh, B. Behin-Aein, J. Makin, S. Sen, and S. Datta, Energy Efficient P-Circuits for Generative Neural Networks, arXiv:2507.07763 (2025); p-Circuits: Neither Digital Nor Analog, 2025 IEEE International Solid-State Circuits Conference (ISSCC, 2025).