 |
Center for Nonlinear Studies |
Experimentalists in quantum information science often validate their experiment by doing quantum state tomgraphy. They repeatedly produce a state and measure it (in various bases), then collate the data to arrive at a "best guess" for the system's density matrix. In this talk, I propose *interval estimates* as a powerful alternative. Instead of reporting a single state -- about which the best that can be said is that it is *probably* fairly close to the true state -- this method reports a convex region that contains the true state with (guaranteed) high probability. Thus, an interval estimate can form the basis of a rigorous logical statement about the state... which in turn can be used (e.g.) to design a fault tolerant quantum computer. I'll discuss how to design a confidence region estimator with guaranteed success probability, how to describe the result concisely, and how to derive a useful *point* estimator from it.
Host: Peter Loxley, loxley@lanl.gov