I’ve been running code to sample from a GBS classically and was curious why simulating PNR detection is so much slower than threshold detection? Sampling using threshold detectors is dependent on calculating the Torontonian which has similar complexity to calculating the Hafnian for PNR detection. So why the drastic increase in runtime for non-threshold detector simulation?
Hi Amanuel. You need to compute a lot more probabilities when you sample with PNRs because there are simply many more possible events. That alone may explain the difference without having to dig into complexities of torontonians and hafnians.
2 Likes