What is NISQ computing?

We know that noise and decoherence are major problems for quantum computers. Even leveraging the power of quantum firmware, error will limit the scale of machine we can build. But the good news is that clever algorithm designers have started to discover new problems that can potentially be solved - problems that could have huge impacts - with noisy intermediate-scale quantum computers: NISQ machines.

Quantum computing in the NISQ era is here right now. We’ve already seen a tantalizing demonstration of “Quantum supremacy” where a small quantum computer solves a problem that, for all practical purposes, is impossible for a regular machine. Unfortunately, it isn’t a problem anyone actually wanted to solve for some commercial or scientific end. So we remain in search of quantum advantage - a demonstration that it’s actually useful to solve a problem with a quantum computer instead of a regular one.

For instance, quantum computers can be used to solve problems in chemistry, like understanding the chemical properties of or reactions between large molecules. This is because we can build a quantum mechanical model of a fundamentally quantum object - the electrons interacting in a molecule.

A lot is lost when we translate this problem to an algorithm run on a classical computer. In fact, make a chemical problem big enough and we can’t say very much about it at all. It intuitively makes more sense to do this quantum modelling on quantum hardware, and formal theories back up this idea. All we need to do is outperform the poor reproduction of the problem on conventional hardware to get a big, practically useful win for drug discovery, biology, and industry.

Moreover, using clever tricks to minimize the amount of quantum computation mechanics- through algorithms called VQE and QAOA - we can reduce the demands on the quantum hardware even further. In these approaches, a lot of the computation is still handled well by a conventional computer. A special small quantum co-processor or accelerator is then used as the key to solving very small chunks that are the most difficult for the regular computer. With these simplified quantum co-processors, quantum error correction is just out of the question, so leveraging approaches like quantum firmware appears to be one of the best pathways to achieving quantum advantage.

Remember, that’s the point at which we achieve a real benefit by using a quantum computer to solve a problem we really care about. At that point, everything changes - we stop researching how to build useful quantum computers and start producing them. Based on recent progress, our community reasonably believes that quantum advantage is achievable within the next 5-10 years.

That’s not the end of the story, of course - it’s really one of the first steps. We can still look to build exceptionally large general-purpose quantum computers. Getting there will require major advances in hardware system design, and combining quantum firmware with quantum error correction to push error rates down as low as possible.