Hybrid algorithms can accommodate limited qubits and provide error correction for real-world tasks.

According to an article in Nature Reviews Physics, Los Alamos National Laboratory, and other top institutions, have created hybrid classical/quantum algorithms that extract the best performance (and possibly quantum advantage) from today’s noisy and error-prone hardware. They are also known as variational quantum algorithms, using quantum boxes to manipulate quantum systems. However, classical computers take most of the workload. This allows them to do their best: solve the optimization problem.

Although quantum computers can perform certain tasks better than classical computers, they cannot run long algorithms on the currently available quantum hardware. They produce too much noise when they interact with the environment, which corrupts information being processed,” Marco Cerezo, a leading author of the paper and physicist at Los Alamos specializing in quantum computing and quantum machine learning and quantum information, said. “With the variational quantum algorithm, we can get the best of both. You can use quantum computers to accomplish difficult tasks for classical computers and then use classical computers to complement quantum computing power.

The current noisy, intermediate-scale quantum computers are between 50 and 100 qubits in size, lose their quantumness quickly and lack error correction, which can require more qubits. Theoreticians have worked on algorithms to run on a large, fault-tolerant, error-correcting quantum computer since the late 1990s.

These algorithms are too complex or give nonsense results, so we can’t use them yet. People realized that we needed an approach to adapt to the limitations of the hardware we had–an optimization problem,” stated Patrick Coles, a Los Alamos theoretical physicist who developed algorithms and was the senior author of this paper.

Coles stated that we found a way to turn all the problems in interest into optimization problems. This could potentially give the quantum computer a quantum advantage. These problems include material science and quantum chemicals simulations, factoring numbers and big-data analysis. They also cover virtually every possible application for quantum computers.

Variational algorithms are used because the optimization process changes the algorithm as needed. This is a form of machine learning. It alters logic gates and parameters to reduce a cost function. This mathematical expression indicates how well an algorithm has performed its task. The problem can be solved when the cost function is at its lowest value.

The variational quantum algorithm uses an iterative function to estimate the cost function. It then sends that result back to the classical computer. The input parameters are adjusted by the classical computer and sent to the quantum computer. It then runs the optimization again.

This review article serves as a guideline and introduction for those just starting to research this emerging field. The authors cover all aspects of algorithms, including how they work and the challenges and pitfalls. It also looks to the future and considers the best ways to achieve quantum computer advantage in the coming years.