gh-486: run individual iternorms on ell blocks
#591
Conversation
Saransh-cpp
added
maintenance
There was a problem hiding this comment.
Still needs ncorr to be a list, and -
every block will need it's own stack of ncorr[i] previous iterations
But I don't think it's that bad — if you figure out the block boundaries in alm beforehand, you need to iterate over them (say k1, k2) zipped with each individual j, and you can update the stack of y with j and z[k1 : k2]
That's why we rearrange the alm — now the block between l1 and l2 is contiguous, so you can cut a slice out of z
There was a problem hiding this comment.
@ntessore could you look at this and see if this is the right direction? Thanks!
| alm = _multalm(z, s) | ||
|
|
||
| # add the mean of the conditional distribution | ||
| y = np.zeros((n * (n + 1) // 2, ncorr), dtype=np.complex128) |
There was a problem hiding this comment.
Given that y is has a dimension ncorr, should be initialised inside the loop or should it be given the dimension ncorrs.
| for i in range(ncorr): | ||
| # calculate ks | ||
| pass |
There was a problem hiding this comment.
I am thinking that this should be possible.
|
I'll take this one on now that fields has been refactored. |
2 participants
Description
Closes: #486
Checks