robust-period
robust-period copied to clipboard
code calculation time is long
Hi, I tested the code and the result is very accurate. But I would like to ask why the calculation time is very long, the average time to calculate the time series of length 1000 in the paper is 0.146s, this is more than 40 seconds with the calculation time in the code.
Hi, you're correct, and that is expected. That part is the one that I cannot really replicate due to incomplete details in the original paper. In fact, I used different optimization (iterative gradient-based) rather than the one in the paper (ADMM), so this might be the main culprit. The hardware specification might matter as well, but not sure about this.