Add zenodo test for remove_stripe_fw

Author: mfep

httomolibgpu/prep/stripe.py

@@ -559,6 +559,7 @@ def remove_stripe_fw(data: cp.ndarray, sigma: float=1, wname: str='sym16', level
             sli_shape = new_sli_shape
         for c in cc:
             mem_stack.free(np.prod(c) * np.float32().itemsize)
+        mem_stack.malloc(np.prod(data) * np.float32().itemsize)
         mem_stack.free(np.prod(sli_shape) * np.float32().itemsize)
         return
 
@@ -585,8 +586,7 @@ def remove_stripe_fw(data: cp.ndarray, sigma: float=1, wname: str='sym16', level
 
     data = sli[:, 0, (nproj_pad - nproj)//2:(nproj_pad + nproj) // 2, :ni]
     data = data.swapaxes(0, 1)
-
-    return data
+    return cp.ascontiguousarray(data)
 
 
 ######## Optimized version for Vo-all ring removal in tomopy########

zenodo-tests/test_prep/test_stripe.py

@@ -5,6 +5,7 @@
 from numpy.testing import assert_allclose
 from httomolibgpu.prep.stripe import (
     remove_stripe_based_sorting,
+    remove_stripe_fw,
     remove_stripe_ti,
     remove_all_stripe,
     raven_filter,
@@ -98,6 +99,50 @@ def test_remove_stripe_ti_i12_dataset4(
     assert output.flags.c_contiguous
 
 
+@pytest.mark.parametrize(
+    "dataset_fixture, sigma_val, level, norm_res_expected",
+    [
+        (
+            "i12_dataset4",
+            0.01,
+            7,
+            52.4856,
+        ),
+        (
+            "i12_dataset4",
+            0.3,
+            5,
+            53.7807,
+        ),
+        (
+            "i12_dataset4",
+            1.0,
+            10,
+            262.2167,
+        ),
+    ],
+    ids=["case_001", "case_003", "case_006"],
+)
+def test_remove_stripe_fw_i12_dataset4(
+    request, dataset_fixture, sigma_val, level, norm_res_expected
+):
+    dataset = request.getfixturevalue(dataset_fixture)
+    data_normalised = normalize(dataset[0], dataset[2], dataset[3], minus_log=True)
+
+    del dataset
+    force_clean_gpu_memory()
+
+    output = remove_stripe_fw(cp.copy(data_normalised), sigma=sigma_val, level=level)
+
+    residual_calc = data_normalised - output
+    norm_res = cp.linalg.norm(residual_calc.flatten())
+
+    assert isclose(norm_res, norm_res_expected, abs_tol=10**-4)
+
+    assert output.dtype == np.float32
+    assert output.flags.c_contiguous
+
+
 @pytest.mark.parametrize(
     "dataset_fixture, snr_val, la_size_val, sm_size_val, norm_res_expected",
     [