Boson sampling is one of those tricky problems that can bog down an ordinary computer but is a breeze for a quantum computer to solve. Suppose I sold you a black box with the promise that it is a quantum computer that solves the boson sampling problem. How can you verify that I'm not trying to scam you? This task sounds like a catch 22, because in order to test that the box performs the calculations correctly you would need a quantum computer in the first place! Things get especially complicated when you recall that there are some classical algorithms (such as the mean field sampler) that do a terrific job at approximating a proper quantum boson sampler. Here's where Liu and colleagues come to the rescue: they found which questions (i.e. input states) yield very different answers (i.e. outputs) if asked to a classical mean field sampler or to a quantum boson sampler. This is very good news, especially because boson samplers come in countless different flavors (depending on which network of linear optics they contain) and the input states that Liu and colleagues found always work to tell apart a legit quantum box from an imposter classical box.

---

---

 Add Comment

You must log in to add comments.