Quantum computers are expected to handle certain complex mathematical computations and advanced simulations that regular computers and supercomputers cannot. But today’s quantum hardware is error-prone, and errors pile up over the long sequences of operations that demanding simulations require.
One solution is to build a fault-tolerant quantum computer that uses quantum error correction to find and fix errors before they ruin the computations. However, this requires a substantial amount of hardware and time.
Assistant Professor Siyuan Niu of the UCF Department of Electrical and Computer Engineering is working with her research team, including UCF co-investigators Di Wu and Eduardo Mucciolo, to determine if a partially fault-tolerant quantum computer with less hardware can offer a workable middle ground between today’s noisy devices and fully fault-tolerant machines. The goal is to improve how simulations are translated into quantum operations, how quickly and accurately errors are identified and how the total quantum and classical computing resources needed to run a simulation are estimated. The project is supported through a $400,000 grant from the Department of Energy.
The broader aim is to determine whether, and when, these advanced machines can reach quantum utility, meaning they can perform scientifically useful computations at a practical cost.
“ For this project, the task is simulating how electrons behave in materials,” Niu says. “The team wants to find out whether an emerging type of quantum computer could do that accurately enough, quickly enough, and with reasonable quantum hardware and classical computing demands to be worthwhile.”
The team will use the Fermi-Hubbard model, a standard way to study interacting electrons, as a benchmark to see if a partially fault-tolerant computer can simulate how these electrons evolve over time. At the same time, the computer needs to keep errors in check through quantum error correction, so it can continue to make reliable calculations even though individual qubits and operations occasionally fail.
“It does this by spreading information across multiple physical qubits, repeatedly checking for signs of errors, and using a fast classical decoder to pinpoint and correct them before they spoil the result,” Niu says. “Quantum computers need this because qubits are fragile, and errors accumulate during the many operations a useful simulation requires.”
Niu’s project will target the Space-Time Efficient Analog Rotation, or STAR, architecture, a partially fault-tolerant design that uses full error correction to protect basic quantum operations, while the rotation operations that quantum simulations rely on heavily are applied directly as “analog” rotations with only partial protection.
“That could avoid some of the extra qubits and time required by fully fault-tolerant designs to perform such rotations,” Niu says. “The trade-off is that these rotations carry some residual error and may need to be repeated until they succeed, so they must be carefully managed and scheduled. The project is tailoring its compiler and cost estimates to that trade-off.”
By the end of the project, Niu’s team aims to deliver an open-source framework that combines a quantum simulation compiler, a fast and accurate GPU-based decoder and a resource estimator. Together, these will show how many qubits, how much time, how much classical computing power and the type of operations that are required for electron simulations on quantum computers. Because the framework is modular, other researchers will be able to plug in their own compilation or decoding methods.
“ Comparing those results with today’s noisy, unprotected devices and fully protected designs could help researchers decide which simulations are realistic, which hardware improvements matter most and when useful applications might become feasible,” Niu says. “It would be a roadmap and assessment, rather than a guaranteed demonstration of quantum advantage.”