Andrew Houck

But a lot of our best algorithms that we run on computers are what we call heuristics.

We run them.

They give us answers.

Those answers are things we didn't know.

We can't prove they're optimal, but they're better than anything we had.

And there's a lot of reasons to suspect that quantum computers will have vastly more impact in these heuristic kinds of algorithms than in things where I can prove down on pen and paper that it will take exactly this many steps to get an optimal answer.

There's all kinds of challenges there.

You need to start with something that can actually behave in a quantum mechanical way.

And the leading platforms either use single atoms or single ions trapped, floating in vacuum, held in place by lasers or electromagnetic fields, or superconducting circuits fabricated like the computer chips we have today.

The challenge is in the circuit models where you can build a lot of them, the information is incredibly fragile.

The very first superconducting qubit that anybody built, a qubit is a quantum bit, something that can store quantum information.

The very first superconducting qubit somebody built lasted for one nanosecond.

You can't do a lot of computation in a nanosecond.

In the 25 years since that time, we've gotten that number up just recently above a millisecond with work that came out of my lab in collaboration with my colleagues here at Princeton.

So it's very exciting to break a millisecond.

And that's long enough that you can start to do error correction and think about actually getting real algorithms done.

But there are so many things that can come and destroy this very fragile quantum state.

So it's hard to get to the point where you can do a lot with it.

It depends on what your system is sensitive to.

But if you're sensitive to magnetic fields, elevators are a real problem.