Tackling the Challenge of a Huge Materials Science Search Space with Quantum‐Inspired Annealing

Abstract

Efficient screening of chemicals is essential for exploring new materials. However, the search space is astronomically large, making calculations with conventional computers infeasible. For example, an N-component system of organic molecules generates >10 60N candidates. Here, a quantum-inspired annealing machine is used to tackle the challenge of the large search space. The prototype system extracts candidate chemicals and their composites with desirable parameters, such as melting temperature and ionic conductivity. The system can be at least 10 4 -10 8 times faster than conventional approaches. Such dramatic acceleration is critical for exploring the enormous search space in virtual screening of materials.

Automated design of Li+-conducting polymer by quantum-inspired annealing

“…The prediction model should follow quadratic equation (1) for the subsequent quantum annealing. [19] 𝐸 reg = −𝑦 ̂= −𝑓 ML (𝒙) = ∑ 𝐽 reg,ij…”

“…Annealers are emerging massively parallel computing hardware that can find minimums for the quadratic formulas. [19,[21][22][23][24] The superposition principle or its computational equivalent enables efficient solution exploration within a practical time range. [19,[21][22][23][24] To acquire solutions, we examined quantum (developed by D-Wave), [22] quantum/digital hybrid (D-Wave), [22] our quantum-inspired DAU, [23,25] simulated bifurcation machine (SBM; Toshiba), [24] standard CPU-based simulated annealing, and conventional Markov chain Monte Carlo (MCMC) solvers.…”

“…[19,[21][22][23][24] The superposition principle or its computational equivalent enables efficient solution exploration within a practical time range. [19,[21][22][23][24] To acquire solutions, we examined quantum (developed by D-Wave), [22] quantum/digital hybrid (D-Wave), [22] our quantum-inspired DAU, [23,25] simulated bifurcation machine (SBM; Toshiba), [24] standard CPU-based simulated annealing, and conventional Markov chain Monte Carlo (MCMC) solvers. Except for the D-Wave (quantum mode) and MCMC solvers, all the solvers provided solutions with an energy around -40 and the norm ∑ 𝑥 𝑖 ≅ 700.…”

“…Weight ratio, molecular weight, and structural information for the polymers, and inorganic additives (e.g., SiO2) were ignored for 𝒙 because quantum annealing systems can input only binary variables, not continuous numbers. [19] Experimental ionic conductivity around room temperature was predicted from the MFPs of the two additives to the salt, that is, a 512 × 2 = 1024-dimensional binary. The PLS algorithm was used to obtain a linear regression model.…”

Continuous black-box optimization with quantum annealing and random subspace coding

“…For our method, instead of quantum annealer, we can use other Ising machines established using characteristic physics systems such as, digital circuit [13], CMOS [14], laser oscillator [15,16], GPU [17], and so on. Recently, Kitai et al [18] proposed a method that solves a discrete black-box optimization problem with the help of quantum annealing, which is called FMQA algo- [19,20].…”