Tests are failing in my local copy

[tgkolda]

Hi, I have a local copy of the main branch. When I run pytest, I get a bunch of errors. I don't know if it's something with my local version or the code itself. I'm trying to make my own changes and run the tests, but then I see these failures that I don't think are caused by me. Even without my changes, the tests are failing. Is this expected? Are there just a subset of tests that I should be focusing on?

PS C:\Users\TammyKolda\Documents\GIT Repos\pyttb> pytest . --ignore=./pyttb/tgk_stuff.py
================================================ test session starts ================================================
platform win32 -- Python 3.11.11, pytest-8.3.4, pluggy-1.5.0
rootdir: C:\Users\TammyKolda\Documents\GIT Repos\pyttb
configfile: pyproject.toml
plugins: cov-6.0.0
collected 208 items                                                                                                  

pyttb\cp_als.py .                                                                                              [  0%]
pyttb\hosvd.py .                                                                                               [  0%]
pyttb\khatrirao.py .                                                                                           [  1%]
pyttb\ktensor.py ..............................                                                                [ 15%]
pyttb\pyttb_utils.py ....F...F...                                                                              [ 21%]
pyttb\sptenmat.py ..............                                                                               [ 28%]
pyttb\sptensor.py ......................................................                                       [ 54%]
pyttb\sumtensor.py .............                                                                               [ 60%]
pyttb\tenmat.py ....................                                                                           [ 70%]
pyttb\tensor.py ............................................................                                   [ 99%]
pyttb\ttensor.py ..                                                                                            [100%]

===================================================== FAILURES ====================================================== 
______________________________________ [doctest] pyttb.pyttb_utils.tt_ind2sub _______________________________________ 
465
466     Returns
467     -------
468     Multi-dimensional indices for the tensor.
469
470     Example
471     -------
472     >>> shape = (2, 2, 2)
473     >>> linear_indices = np.array([0, 1])
474     >>> tt_ind2sub(shape, linear_indices)
Expected:
    array([[0, 0, 0],
           [1, 0, 0]])
Got:
    array([[0, 0, 0],
           [1, 0, 0]], dtype=int64)

C:\Users\TammyKolda\Documents\GIT Repos\pyttb\pyttb\pyttb_utils.py:474: DocTestFailure
______________________________________ [doctest] pyttb.pyttb_utils.tt_sub2ind _______________________________________ 
521     subs:
522         Subscripts for tensor
523     order:
524         Memory layout
525
526     Examples
527     --------
528     >>> shape = (2, 2, 2)
529     >>> full_indices = np.array([[0, 0, 0], [1, 0, 0]], dtype=int)
530     >>> tt_sub2ind(shape, full_indices)
Expected:
    array([0, 1])
Got:
    array([0, 1], dtype=int64)

C:\Users\TammyKolda\Documents\GIT Repos\pyttb\pyttb\pyttb_utils.py:530: DocTestFailure
============================================== short test summary info ============================================== 
FAILED pyttb/pyttb_utils.py::pyttb.pyttb_utils.tt_ind2sub
FAILED pyttb/pyttb_utils.py::pyttb.pyttb_utils.tt_sub2ind
=========================================== 2 failed, 206 passed in 0.65s =========================================== 
PS C:\Users\TammyKolda\Documents\GIT Repos\pyttb> 

[tgkolda]

@ntjohnson1 I get even more errors when I use #373.

[ntjohnson1]

It looks to me like there was a recent Numpy update, since my PR for #373 passed all the tests.

If it's blocking your progress you should be able to use a slightly older version of numpy (<2.2.1 should be fine if my hypothesis is correct).

[tgkolda]

I'm already using numpy 1.26.4.

[ntjohnson1]

Ahh so you are using an intermediate numpy that's not the latest or oldest. It looks like numpys print format/type promotion is different for just 1.25/1.26 (1.24 and 2.0+ are consistent). This doesn't appear to impact our functionality but doctests look for an exact match or more complicated pattern match. When I ran the full pytest tests on your branch there were additional errors around typos/linting/formatting but they all seemed legitimate.

If you upgrade to numpy 2.0+ or apply this patch you should be all set. It would make sense to merge this in on some branch to reduce our sensitivity to specifics in the printing format.

git diff
diff --git a/pyttb/pyttb_utils.py b/pyttb/pyttb_utils.py
index b7fe010..1f087d3 100644
--- a/pyttb/pyttb_utils.py
+++ b/pyttb/pyttb_utils.py
@@ -470,10 +470,10 @@ def tt_ind2sub(
     Example
     -------
     >>> shape = (2, 2, 2)
-    >>> linear_indices = np.array([0, 1])
-    >>> tt_ind2sub(shape, linear_indices)
-    array([[0, 0, 0],
-           [1, 0, 0]])
+    >>> linear_indices = np.array([0, 1], dtype=int)
+    >>> full_indices = np.array([[0, 0, 0], [1, 0, 0]])
+    >>> np.array_equal(tt_ind2sub(shape, linear_indices), full_indices)
+    True
     """
     if idx.size == 0:
         return np.empty(shape=(0, len(shape)), dtype=int)
@@ -527,8 +527,9 @@ def tt_sub2ind(
     --------
     >>> shape = (2, 2, 2)
     >>> full_indices = np.array([[0, 0, 0], [1, 0, 0]], dtype=int)
-    >>> tt_sub2ind(shape, full_indices)
-    array([0, 1])
+    >>> linear_indices = np.array([0, 1])
+    >>> np.array_equal(tt_sub2ind(shape, full_indices), linear_indices)
+    True

     See Also
     --------

[tgkolda]

Updating to 2.2.1 did the trick. Thanks!