fix concatenation for random statistics #147
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
There were two issues in the concatenation code for the local statistics for the parallel conditional permutation code. The original code is below: > rlocals = numpy.hstack(rlocals).flatten() The first big issue is that rlocals is a numpy array containing the random realizations of simulations for observations within each processing job. This code, however, will flatten that out into an array of chunk_size*p_permutations, which certainly is not correct. The second (slightly smaller) issue is that hstack fails when its chunks are heterogeneously sized, even when you just want to stack over the rows. For example: >>> a = numpy.arange(20).reshape(4,5) >>> b = numpy.arange(10).reshape(2,5) >>> c = numpy.hstack((a,b)) # Fails! That last numpy.hstack line fails with a ValueError, since "all of the input array dimensions for the concatenation axis must match exactly." That means that a and b have to have the same number of rows to be hstacked. But, since they're 2-dimensional, one can *row_stack* them just fine: >>> c = numpy.row_stack((a,b)) # Success! c contains the first 4 rows from a, and then the last 2 rows are b. This is what we need to do in the rlocals case, since each chunk might be slightly different in size, but still will always be (chunk_size, p_permutations).
Codecov Report
@@ Coverage Diff @@
## master #147 +/- ##
=======================================
Coverage 77.61% 77.61%
=======================================
Files 25 25
Lines 3274 3274
=======================================
Hits 2541 2541
Misses 733 733
Continue to review full report at Codecov.
|
Member
Author
|
This resolves #146. To verify this works, this is broken on master, but resolved here. I knew I had addressed this before in the import esda
import numpy
import libpysal
import time
time0 = time.time()
size = 100;
patch_eg = numpy.random.rand(size,size).flatten()
w = libpysal.weights.lat2W(size,size)
lm = esda.Moran_Local(patch_eg, w, transformation='r', keep_simulations=True,
permutations=1000, n_jobs=8) |
sjsrey
approved these changes
Nov 24, 2020
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There were two issues in the concatenation code for the local statistics
for the parallel conditional permutation code. The original code is
below:
The first big issue is that rlocals is a numpy array containing the
random realizations of simulations for observations within each
processing job. This code, however, will flatten that out into an array
of chunk_size*p_permutations, which certainly is not correct.
The second (slightly smaller) issue is that hstack fails when its chunks
are heterogeneously sized, even when you just want to stack over the
rows. For example:
That last numpy.hstack line fails with a ValueError, since "all of the
input array dimensions for the concatenation axis must match exactly."
That means that a and b have to have the same number of rows to be
hstacked. But, since they're 2-dimensional, one can row_stack them
just fine:
c contains the first 4 rows from a, and then the last 2 rows are b.
This is what we need to do in the rlocals case, since each chunk might be
slightly different in size, but still will always be (chunk_size,
p_permutations).