Matteo Ravasi
Matteo Ravasi
Regarding GitHub CI, this never fails (or at least not to my knowledge). The red stars you see in the latest commits are due to my impatience of merging without...
Yep, agree. As 1 and 4 are now very fast, I think if we can improve on 2 and 3 the overall timing of the test suite should become more...
Makes sense. I guess here the situation is reversed as cupy doesn’t seem to like tuples. I definitely see the point of using tuples for this parameter as it is...
Wait, I got this wrong. Actually if `nnfts` was stored as tuple everything would be fine, the issue is that it is stored as np.array. ``` a = (100, 100)...
Sounds good :) are you happy to take a look into this or shall I do it?
Sure, let me do it. I think I am going to make sure dims is always a tuple, then I’m sure it will work as this is what I do...
Great @pavane, this will make it a great addition. Carlos references are very good ones to understand how to implement the adjoint of the eikonal solver. We are here at...
Hi again, I was actually reading again your initial message and the comment of Carlos and I think there is something important to keep in mind here. If you are...
Nice :) I think the fact you can get rays from the eikonal map is very good and basically the last step would be to grid them into the tomographic...
Hi @eurunuela, currently it will not work as internally we call `matvec` and `rmatvec` but it should be an easy change to make it work for multiple right-hand sides at...