Use torch.cov instead of custom function

tests/test_torch.py

@@ -11,13 +11,6 @@
 def setup_function(fn):
     print("torch version:", torch.__version__, "torchvision version:", torchvision.__version__)
 
-def test_cov():
-    x = np.random.randn(10, 10)
-    cov_np = np.cov(x)
-    cov_t = torchstain.torch.utils.cov(torch.tensor(x))
-
-    np.testing.assert_almost_equal(cov_np, cov_t.numpy())
-
 def test_percentile():
     x = np.random.randn(10, 10)
     p = 20

torchstain/torch/augmentors/macenko.py

@@ -1,6 +1,6 @@
 import torch
 from torchstain.base.augmentors.he_augmentor import HEAugmentor
-from torchstain.torch.utils import cov, percentile
+from torchstain.torch.utils import percentile
 
 """
 Source code ported from: https://github.com/schaugf/HEnorm_python
@@ -66,7 +66,7 @@ def __compute_matrices(self, I, Io, alpha, beta):
         OD, ODhat = self.__convert_rgb2od(I, Io=Io, beta=beta)
 
         # compute eigenvectors
-        _, eigvecs = torch.linalg.eigh(cov(ODhat.T)) 
+        _, eigvecs = torch.linalg.eigh(torch.cov(ODhat.T)) 
         eigvecs = eigvecs[:, [1, 2]]
 
         HE = self.__find_HE(ODhat, eigvecs, alpha)

torchstain/torch/normalizers/macenko.py

@@ -1,6 +1,6 @@
 import torch
 from torchstain.base.normalizers.he_normalizer import HENormalizer
-from torchstain.torch.utils import cov, percentile
+from torchstain.torch.utils import percentile
 
 """
 Source code ported from: https://github.com/schaugf/HEnorm_python
@@ -61,7 +61,7 @@ def __compute_matrices(self, I, Io, alpha, beta):
         OD, ODhat = self.__convert_rgb2od(I, Io=Io, beta=beta)
 
         # compute eigenvectors
-        _, eigvecs = torch.linalg.eigh(cov(ODhat.T)) 
+        _, eigvecs = torch.linalg.eigh(torch.cov(ODhat.T)) 
         eigvecs = eigvecs[:, [1, 2]]
 
         HE = self.__find_HE(ODhat, eigvecs, alpha)

torchstain/torch/normalizers/multitarget.py

@@ -1,5 +1,5 @@
 import torch
-from torchstain.torch.utils import cov, percentile
+from torchstain.torch.utils import percentile
 
 """
 Implementation of the multi-target normalizer from the paper: https://arxiv.org/pdf/2406.02077
@@ -50,7 +50,7 @@ def __compute_matrices_single(self, I, Io, alpha, beta):
         OD, ODhat = self.__convert_rgb2od(I, Io=Io, beta=beta)
 
         # _, eigvecs = torch.symeig(cov(ODhat.T), eigenvectors=True)
-        _, eigvecs = torch.linalg.eigh(cov(ODhat.T), UPLO='U')
+        _, eigvecs = torch.linalg.eigh(torch.cov(ODhat.T), UPLO='U')
         eigvecs = eigvecs[:, [1, 2]]
 
         HE = self.__find_HE(ODhat, eigvecs, alpha)
@@ -77,7 +77,7 @@ def fit(self, Is, Io=240, alpha=1, beta=0.15):
             OD = torch.cat(ODs, dim=0)
             ODhat = torch.cat(ODhats, dim=0)
 
-            eigvecs =  torch.symeig(cov(ODhat.T), eigenvectors=True)[1][:, [1, 2]]
+            eigvecs =  torch.symeig(torch.cov(ODhat.T), eigenvectors=True)[1][:, [1, 2]]
 
             HE =  self.__find_HE(ODhat, eigvecs, alpha)
 
@@ -91,7 +91,7 @@ def fit(self, Is, Io=240, alpha=1, beta=0.15):
                 for I in Is
             ))
 
-            eigvecs = torch.stack([torch.symeig(cov(ODhat.T), eigenvectors=True)[1][:, [1, 2]] for ODhat in ODhats]).mean(dim=0)
+            eigvecs = torch.stack([torch.symeig(torch.cov(ODhat.T), eigenvectors=True)[1][:, [1, 2]] for ODhat in ODhats]).mean(dim=0)
 
             OD = torch.cat(ODs, dim=0)
             ODhat = torch.cat(ODhats, dim=0)

torchstain/torch/utils/__init__.py

@@ -1,4 +1,3 @@
-from torchstain.torch.utils.cov import cov
 from torchstain.torch.utils.percentile import percentile
 from torchstain.torch.utils.stats import *
 from torchstain.torch.utils.split import *