fakecbed 0.3.3__tar.gz → 0.3.5__tar.gz

{fakecbed-0.3.3 → fakecbed-0.3.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fakecbed
-Version: 0.3.3
+Version: 0.3.5
 Summary: A Python library for generating quickly images that imitate convergent beam electron diffraction patterns.
 Author-email: Matthew Fitzpatrick <matthew.rc.fitzpatrick@gmail.com>
 Project-URL: Homepage, https://mrfitzpa.github.io/fakecbed

{fakecbed-0.3.3 → fakecbed-0.3.5}/examples/basic_usage.ipynb

@@ -1293,7 +1293,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.11.13"
+   "version": "3.12.8"
   }
  },
  "nbformat": 4,

{fakecbed-0.3.3 → fakecbed-0.3.5}/fakecbed/discretized.py

@@ -1482,6 +1482,11 @@ class CBEDPattern(fancytypes.PreSerializableAndUpdatable):
         illumination_support = self._illumination_support.cpu().detach().clone()
         illumination_support = illumination_support.numpy(force=True)
 
+        if self._disk_supports is None:
+            method_name = ("_calc_disk_supports"
+                           "_and_cache_select_intermediates")
+            method_alias = getattr(self, method_name)
+            self._disk_supports = method_alias(u_x, u_y)
         disk_supports = self._disk_supports.numpy(force=True)
 
         num_disks = self._num_disks
@@ -1701,11 +1706,13 @@ class CBEDPattern(fancytypes.PreSerializableAndUpdatable):
 
         apply_shot_noise = self._apply_shot_noise
         if apply_shot_noise == True:
-            np_rng = np.random.default_rng(self._rng_seed)
-            torch_rng_seed = np_rng.integers(low=0, high=2**32-1).item()
-            torch_rng = torch.Generator(device=u_x.device)
-            torch_rng = torch_rng.manual_seed(torch_rng_seed)
-            image = torch.poisson(image, torch_rng)
+            # ``torch.poisson`` was occasionally causing CUDA errors.
+            rng = np.random.default_rng(self._rng_seed)
+            image_dtype = image.dtype
+            image = image.numpy(force=True)
+            image = rng.poisson(image)
+            image = torch.from_numpy(image)
+            image = image.to(device=self._device, dtype=image_dtype)
 
         image = self._apply_detector_partition_inpainting(input_image=image)
             
@@ -1715,7 +1722,7 @@ class CBEDPattern(fancytypes.PreSerializableAndUpdatable):
             self._illumination_support = method_alias(u_x, u_y)
         illumination_support = self._illumination_support
 
-        image = image*illumination_support        
+        image = image*illumination_support
 
         coords_of_cold_pixels = self._cold_pixels
         L, R, B, T = self._mask_frame

{fakecbed-0.3.3 → fakecbed-0.3.5}/fakecbed/version.py

@@ -17,5 +17,5 @@ __version__: str
 __version_tuple__: VERSION_TUPLE
 version_tuple: VERSION_TUPLE
 
-__version__ = version = '0.3.3'
-__version_tuple__ = version_tuple = (0, 3, 3)
+__version__ = version = '0.3.5'
+__version_tuple__ = version_tuple = (0, 3, 5)

{fakecbed-0.3.3 → fakecbed-0.3.5}/fakecbed.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fakecbed
-Version: 0.3.3
+Version: 0.3.5
 Summary: A Python library for generating quickly images that imitate convergent beam electron diffraction patterns.
 Author-email: Matthew Fitzpatrick <matthew.rc.fitzpatrick@gmail.com>
 Project-URL: Homepage, https://mrfitzpa.github.io/fakecbed

{fakecbed-0.3.3 → fakecbed-0.3.5}/tests/test_discretized.py

@@ -659,6 +659,24 @@ def test_8_of_CBEDPattern():
 
 
 
+def test_9_of_CBEDPattern():
+    cbed_pattern = fakecbed.discretized.CBEDPattern()
+
+    num_pixels_across_pattern = \
+        cbed_pattern.core_attrs["num_pixels_across_pattern"]
+
+    overriding_image = np.zeros(2*(num_pixels_across_pattern,))
+
+    kwargs = {"overriding_image": overriding_image,
+              "skip_validation_and_conversion": False}
+    cbed_pattern.override_image_then_reapply_mask(**kwargs)
+
+    cbed_pattern_signal = cbed_pattern.get_signal(deep_copy=False)
+    
+    return None
+
+
+
 ###########################
 ## Define error messages ##
 ###########################