The Incredible Power of Quantum Memory

The archetypal version of this story appeared in Quanta Magazine.

It’s not casual to survey quantum systems—collections of particles that travel the counterintuitive rules of quantum mechanics. Heisenberg’s uncertainty principle, a cornerstone of quantum theory, says it’s intolerable to simultaneously measurement a particle’s nonstop presumption and its speed—pretty important accusation for knowing what’s going on.

In bid to study, say, a peculiar postulation of electrons, researchers person to beryllium clever astir it. They mightiness instrumentality a container of electrons, poke astatine it successful assorted ways, past instrumentality a snapshot of what it looks similar astatine the end. In doing so, they anticipation to reconstruct the interior quantum dynamics astatine work.

But there’s a catch: They can’t measurement each the system’s properties astatine the aforesaid time. So they iterate. They’ll commencement with their system, poke, past measure. Then they’ll bash it again. Every iteration, they’ll measurement immoderate caller acceptable of properties. Build unneurotic capable snapshots, and machine learning algorithms tin assistance reconstruct the afloat properties of the archetypal system—or astatine slightest get truly close.

This is simply a tedious process. But successful theory, quantum computers could help. These machines, which enactment according to quantum rules, person the imaginable to beryllium overmuch amended than mean computers astatine modeling the workings of quantum systems. They tin besides store accusation not successful classical binary memory, but successful a much analyzable signifier called quantum memory. This allows for acold richer and much close descriptions of particles. It besides means that the machine could support aggregate copies of a quantum authorities successful its moving memory.

A fewer years ago, a squad based astatine the California Institute of Technology demonstrated that definite algorithms that usage quantum representation necessitate exponentially less snapshots than algorithms that don’t usage it. Their method was a large advance, but it required a comparatively ample magnitude of quantum memory.

That’s thing of a deal-breaker, due to the fact that arsenic a applicable matter, quantum representation is hard to travel by. A quantum machine is made of interconnected quantum bits called qubits, and qubits tin beryllium utilized for computation oregon representation but not both.

Now, 2 autarkic teams person travel up with ways of getting by with acold little quantum memory. In the archetypal paper, Sitan Chen, a machine idiosyncratic astatine Harvard University, and his coauthors showed that conscionable 2 copies of the quantum authorities could exponentially trim the fig of times you request to instrumentality a snapshot of your quantum system. Quantum memory, successful different words, is astir ever worthy the investment.