Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: meas_deblender
-
Labels:None
-
Story Points:4
-
Epic Link:
-
Sprint:DRP S18-5
-
Team:Data Release Production
Description
After DM-9584, the current bottleneck in the deblender is the translation and PSF operators, where (when PSF convolution is used) nearly all of the processing time is spent in a single function:
def apply_filter(X, weights, slices, inv_slices): |
"""Apply a filter to a 2D image X |
|
|
Parameters
|
----------
|
X: 2D numpy array
|
The image to apply the filter to
|
weights: 1D array
|
Weights corresponding to each slice in `slices`
|
slices: list of `slice` objects
|
Slices in the new `X` to store the filtered X
|
inv_slices: list of `slice` objects
|
Slices of `X` to apply each weight
|
|
Returns
|
-------
|
new_X: 2D numpy array
|
The result of applying the filter to `X`
|
"""
|
result = np.zeros(X.shape, dtype=X.dtype) |
for n, weight in enumerate(weights): |
result[slices[n]] += weight * X[inv_slices[n]] |
return result |
Moving this to C++ and using the Eigen package is likely to improve the performance, and is the last tall pole remaining in optimizing the deblender. After this any performance increases will have to come from more clever implementations of certain algorithms and small optimizations throughout the code.
Attachments
Activity
The changes are made in https://github.com/fred3m/scarlet/pull/43. Jim Bosch, the main parts I'd like you to check are that I made the best use of Eigen in operators_pybind11.cc and my implementation to download the Eigen headers if needed. I tested that this works on my machine, but it would be useful to have both Peter Melchior and Jim Bosch verify that this works for them too.
Fred Moolekamp
added a comment - The changes are made in https://github.com/fred3m/scarlet/pull/43. Jim Bosch , the main parts I'd like you to check are that I made the best use of Eigen in operators_pybind11.cc and my implementation to download the Eigen headers if needed. I tested that this works on my machine, but it would be useful to have both Peter Melchior and Jim Bosch verify that this works for them too. It works fine for me.
Peter Melchior
added a comment - It works fine for me. Peter Melchior I updated the branch and it should be ready to merge with your approval.
Fred Moolekamp
added a comment - Peter Melchior I updated the branch and it should be ready to merge with your approval. This is great work and a substantial speed-up, which allows us to routinely do the full PSF treatment.
Peter Melchior
added a comment - This is great work and a substantial speed-up, which allows us to routinely do the full PSF treatment. 
There is an extra attribute LinearFilter.cpp that is used to turn the C++ methods on/off for testing. Before this ticket is merged that will be removed, but I'm leaving it in until this passes the rest of the review.