More missing numpy and matlab #649
Conversation
`union` returns the unique, unsorted Tensor of values that are found in either of the two input Tensors. Note that an equivalent function exists both in `numpy` (where it is called `union1d`) and in `Matlab` (where it is called `union1d`). However, those functions always sort the output, while Arraymancer's version does not. To replicate the same behavior, simply apply `sort` to the output of this function.
AngelEzquerra
force-pushed
the
more_missing_numpy_and_matlab
branch
4 times, most recently
from
02be765
to
28c8ec1
Compare
| let data = toSeq(a) | ||
| result = toTensor(data, shape) |
There was a problem hiding this comment.
I suppose we could also do:
result = newTensorUninit[T](a.card)
for i, x in a:
result[i] = xto avoid the double copy (set -> seq -> tensor)
There was a problem hiding this comment.
Makes sense. I'll make that change
There was a problem hiding this comment.
Interestingly a performance measurement using timeit showed that the original version was faster on Windows. We will leave this as is for now while we investigate what is going on (and if we find out why we'll update this in a separate PR).
There was a problem hiding this comment.
Yeah, let's investigate it further after this PR.
|
Maybe we can reuse the internal logic? In general |
You are right. Let me remove those from this PR. We can discuss offline and I can create another separate PR later. |
This will let us avoid having to convert HashSets into seqs before converting them into tensors. Note that this also improves a little the docstrings of a couple of the existing `toTensor` procedures.
`intersection` returns the "intersection" of 2 Tensors as an unsorted rank-1 Tensor. Note that an equivalent function exists both in `numpy` (where it is called `intersect1d`) and in `Matlab` (where it is called `intersect`). However, those functions always sort the output, while Arraymancer's version does not. To replicate the same behavior, simply apply `sort` to the output of this function. Also note that to implement this feature we moved (and made public) the existing, private toHashSet procedure from spatial/distances.nim into tensor/initialization.nim.
`setDiff` returns the (symmetric or non symmetric) "difference" between 2 Tensors as an unsorted rank-1 Tensor. Note that an equivalent function exists both in `numpy` (where it is called `setdiff1d`) and in `Matlab` (where it is called `setdiff`). However, those functions always sort the output, while Arraymancer's version does not. To replicate the same behavior, simply apply `sort` to the output of this function.
`find` (which is used to implement `contains`) was already supported (since `system.find` is generic and works with Tensors) but was untested, so this also adds a test for it.
This was a useful std/math function that we did not support yet.
AngelEzquerra
force-pushed
the
more_missing_numpy_and_matlab
branch
from
28c8ec1
to
381c5ba
Compare
I've updated the PR as discussed here. |
AngelEzquerra
force-pushed
the
more_missing_numpy_and_matlab
branch
from
381c5ba
to
2c7752b
Compare
|
I just pushed a fix to the original (non gemm) version which solves a problem with convolving a slice (was using |
| # The folling export is needed to avoid an compilation error in | ||
| # algorithms.nim/intersection() when running the test_algorithms test: | ||
| # `Error: type mismatch - Expression: items(s1)` | ||
| export sets |
There was a problem hiding this comment.
Note that we could also get away with adding a bind sets.items inside of intersection and setDiff. However, given that we now depend on sets, I don't see an issue exporting it too.
Fix typo in export comment & add alternative for reader
|
Merging, we can look into the |
This PR adds multiple functions that are available both in numpy and Matlab:
union: returns the unique, unsorted Tensor of values that are found in either of the two input Tensors.intersection: returns the "intersection" of 2 Tensors as an unsorted rank-1 Tensor.setDiff: returns the (symmetric or non symmetric) "difference" between 2 Tensors as an unsorted rank-1 Tensor.toHashSet: convert a Tensor into HashSet.contains(and thusinandnotin).sub2indandind2sub: convert "subscript" sequences into linear tensor indexes and vice-versa.almostEqual: missing from nim's std/math.