Spiking neuromorphic chip learns entangled quantum states

Abstract

The approximation of quantum states with artificial neural networks has gained a lot of attention during the last years. Meanwhile, analog neuromorphic chips, inspired by structural and dynamical properties of the biological brain, show a high energy efficiency in running artificial neural-network architectures for the profit of generative applications. This encourages employing such hardware systems as platforms for simulations of quantum systems. Here we report on the realization of a prototype using the latest spike-based BrainScaleS hardware allowing us to represent few-qubit maximally entangled quantum states with high fidelities. Bell correlations of pure and mixed two-qubit states are well captured by the analog hardware, demonstrating an important building block for simulating quantum systems with spiking neuromorphic chips.

The BrainScaleS-2 Accelerated Neuromorphic System With Hybrid Plasticity

“…The accelerated nature of the system allows one to rapidly produce samples over long periods. Recently such a variational representation was used to represent POVM probability distributions of states in certain quantum systems on BrainScaleS-2 (Czischek et al, 2019 ; Klassert et al, 2021 ; Czischek et al, 2022 ).…”

A Scalable Approach to Modeling on Accelerated Neuromorphic Hardware

“…However, to deliver on this promise in reality, both, hardware and software need to be carefully designed, implemented, and applied. The publications building on BSS-2 are evidence of what is possible in terms of modeling on accelerated neuromorphic hardware (Bohnstingl et al, 2019 ; Cramer et al, 2020 , 2022 ; Wunderlich et al, 2019 ; Billaudelle et al, 2020 , 2021 ; Müller et al, 2020a ; Spilger et al, 2020 ; Weis et al, 2020 ; Göltz et al, 2021 ; Kaiser et al, 2022 ; Klassert et al, 2021 ; Klein et al, 2021 ; Stradmann et al, 2021 ; Czischek et al, 2022 ; Schreiber et al, in preparation).…”

“…This set of layers is feature-complete to formulate arbitrary hardware-compatible experiments and was used as basis for experiments in Schemmel et al ( 2020 ), Göltz et al ( 2021 ), Klassert et al ( 2021 ), Czischek et al ( 2022 ), and Cramer et al ( 2022 ).…”

Variational learning of quantum ground states on spiking neuromorphic hardware

“…These advantages could provide significant speedups for the emulation of large networks or quantum spin systems. For this reason we have tested the scalability of our approach, thereby expanding previous work by Czischek et al. (2022) from representing small entangled states to larger quantum spin systems of up to spins.…”

“…Because of the physical implementation the emulation becomes inherently parallel, rendering the sampling speed independent of the network size. We note that neuromorphic hardware has recently been used to represent entangled quantum states using a mapping of general mixed quantum states to a probabilistic representation and training the system to represent a given state by approximating its corresponding probability distribution ( Czischek et al., 2022 ). Here, instead, we directly encode the wave function of pure quantum states and use this approach for variational ground state search through minimization of the quantum system’s total energy.…”