Arclight Quantum Quantum Computing Inc., alongside researchers at the Chinese Academy of Sciences, has created fast ITE or FITE, a new probabilistic imaginary-time evolution scheme that simplifies quantum computation for ground state preparation and thermal equilibrium calculations. This method uses an optimised offline basis transformation to represent the Hamiltonian diagonally; its online computational cost remains constant regardless of how long the simulation runs. Numerical tests demonstrate reduced gate costs compared with existing methods when targeting equivalent accuracy levels.

Arclight Quantum Computing Inc., with Chinese Academy of Sciences researchers, has devised fast ITE or FITE, a new method for quantum simulation that simplifies calculations needed for understanding ground states and thermal equilibrium. The team optimised an initial process, a basis transformation, to represent complex systems in a more manageable way. Once prepared, the computational effort required remains constant regardless of how long the simulation runs, reducing demands on processing power when exploring extended scenarios or multiple possibilities.

Arclight Quantum Computing Inc., working with researchers at the Chinese Academy of Sciences, has developed fast ITE, or FITE, a new method for simplifying quantum simulations used in understanding ground states and thermal equilibrium. The team optimised an initial process; akin to building a custom cheat sheet before taking an exam, they performed complex calculations beforehand to speed up later stages. This allows them to represent complicated systems more efficiently.

Once this preparation is complete, the computational effort needed remains constant no matter how long the simulation runs, reducing demands on processing power when exploring extended scenarios. The technique relies on “imaginary-time evolution”, which can be visualised as rewinding a video game to find stable starting conditions rather than watching events unfold forward. Numerical tests show FITE reduces gate costs compared with existing methods at equivalent accuracy levels, but approximation errors and offline preparation may impact its overall benefit.

FITE lowers gate costs and boosts state acceptance rates in quantum

FITE, the fast imaginary-time evolution method, reduced online gate costs by approximately 21,123 times compared with first-order Pauli PITE methods on Hubbard, XY and XYZ models. Achieving these gains required surpassing an accuracy threshold where previous techniques frequently failed to yield accepted states after a single attempt. Optimising an offline basis transformation allows FITE to reuse circuit structures regardless of simulation length or initial conditions, a key advantage over approaches requiring recalculation at each time step.

Success probabilities reached 0.317 on a Hubbard model; 0.0641 on an XY chain; and 4.09x 10−4 on an XYZ chain. For the two-qubit Hubbard model, grouped filtering achieved similar success rates as FITE, but this advantage did not consistently extend across other models like the XY chain where FITE still outperformed it; however even with these improvements absolute success probability remained low on the XYZ chain meaning repeated attempts are costly. While demonstrating substantial resource reduction, benchmarks do not yet account for potential overheads associated with offline basis preparation or coefficient estimation in more complex systems.

Optimised System Representation Balances Preparation Costs and Simulation Speed

A technique used to find stable quantum states, imaginary-time evolution, sees demonstrably reduced computational cost thanks to this new method, though its overall benefit isn’t guaranteed. Arclight Quantum Computing and Chinese Academy of Sciences acknowledge significant offline preparation costs remain unresolved. This pre-processing stage optimizes how complex systems are represented for simpler calculations but introduces an initial investment that could outweigh gains depending on specific applications or system size.

Still, it offers valuable progress by tackling a core challenge in quantum simulation: reducing the demands of finding stable solutions. Further development will address remaining preparation costs for wider applicability across diverse quantum problems.

This research demonstrated a probabilistic imaginary-time evolution scheme with reduced computational cost compared to existing Pauli-based methods while maintaining comparable accuracy, reaching infidelity below 0.005, across Hubbard, XY and XYZ models. The method achieves higher success probabilities, up to 0.317 on the Hubbard model, by optimizing how complex systems are represented before calculations begin.

This allows for consistent circuit structures that can be reused regardless of simulation length or initial conditions, potentially improving efficiency in finding stable states within quantum simulations; however, benchmarks do not yet fully account for preparation costs associated with more complex systems.

👉 More information
🗞 Probabilistic imaginary-time evolution via compiled diagonal representations
✍️ Hengyue Li, Huazhe Lou, Yang Li, Mingyu Huang, Yusheng Yang, Jianyong Sheng and Jian Sun (Affiliation: Arclight Quantum Computing); Shenggang Ying (Chinese Academy of Sciences)
🧠 ArXiv: https://arxiv.org/abs/2610.39600

More like this