Question about get_mask function

[NikolasMorshuis]

Hi,

Thanks for your great work and for sharing your code!

I was wondering about the get_mask function. It seems to me that the true acceleration that we get is somewhat lower than the one we expect: If I want e.g. an acceleration of 16x with a Gaussian uniform 1D sampling pattern and a center fraction of 0.04, I get a true acceleration of 11x using the get_mask function.

In the code below, more samples than necessary are sampled (Nsamp*1.2 while we only need Nsamp). Additionally to Nsamp, Nsamp_center has to be added, leading to even more samples in total.

https://github.com/HJ-harry/DDS/blob/509b140ee8ba76772ab7d98929270033d4306f2f/utils.py#L380-L392

Looking forward to hear your insights on the topic!