Added bit (sample-by-sample) normalization (#284)

* Added bit (sample-by-sample) normalization and fixed normalization bug

* improved testing

---------

Co-authored-by: ariellellouch <[email protected]>

dascore/proc/basic.py

@@ -296,7 +296,7 @@ def squeeze(self: PatchType, dim=None) -> PatchType:
 def normalize(
     self: PatchType,
     dim: str,
-    norm: Literal["l1", "l2", "max"] = "l2",
+    norm: Literal["l1", "l2", "max", "bit"] = "l2",
 ) -> PatchType:
     """
     Normalize a patch along a specified dimension.
@@ -311,6 +311,7 @@ def normalize(
             l1 - divide each sample by the l1 of the axis.
             l2 - divide each sample by the l2 of the axis.
             max - divide each sample by the maximum of the absolute value of the axis.
+            bit - sample-by-sample normalization (-1/+1)
     """
     axis = self.dims.index(dim)
     data = self.data
@@ -319,13 +320,23 @@ def normalize(
         norm_values = np.linalg.norm(self.data, axis=axis, ord=order)
     elif norm == "max":
         norm_values = np.max(data, axis=axis)
+    elif norm == "bit":
+        pass
     else:
         msg = (
             f"Norm value of {norm} is not supported. "
-            f"Supported values are {('l1', 'l2', 'max')}"
+            f"Supported values are {('l1', 'l2', 'max', 'bit')}"
         )
         raise ValueError(msg)
-    new_data = data / np.expand_dims(norm_values, axis=axis)
+    if norm == "bit":
+        new_data = np.divide(
+            data, np.abs(data), out=np.zeros_like(data), where=np.abs(data) != 0
+        )
+    else:
+        expanded_norm = np.expand_dims(norm_values, axis=axis)
+        new_data = np.divide(
+            data, expanded_norm, out=np.zeros_like(data), where=expanded_norm != 0
+        )
     return self.new(data=new_data)
 
 

tests/test_proc/test_basic.py

@@ -6,6 +6,7 @@
 import pytest
 
 import dascore as dc
+from dascore import get_example_patch
 from dascore.exceptions import IncompatiblePatchError, UnitError
 from dascore.proc.basic import apply_operator
 from dascore.units import furlongs, get_quantity, m, s
@@ -142,6 +143,27 @@ def test_max(self, random_patch):
         norm = np.abs(np.sum(time_norm.data, axis=axis))
         assert np.allclose(norm, 1)
 
+    def test_bit(self):
+        """Ensure after operation each sample is -1, 1, or 0."""
+        patch = get_example_patch("dispersion_event")
+        bit_norm = patch.normalize("time", norm="bit")
+        assert np.all(np.unique(bit_norm.data) == np.array([-1.0, 0, 1.0]))
+
+    def test_zero_channels(self, random_patch):
+        """Ensure after operation each zero row or vector remains so."""
+        zeroed_data = np.copy(random_patch.data)
+        zeroed_data[0, :] = 0.0
+        zeroed_data[:, 0] = 0.0
+        zeroed_patch = random_patch.new(data=zeroed_data)
+        for norm_type in ["l1", "l2", "max", "bit"]:
+            norm = zeroed_patch.normalize("time", norm=norm_type)
+            assert np.all(norm.data[0, :] == 0.0)
+            assert np.all(norm.data[:, 0] == 0.0)
+        for norm_type in ["l1", "l2", "max", "bit"]:
+            norm = zeroed_patch.normalize("distance", norm=norm_type)
+            assert np.all(norm.data[0, :] == 0.0)
+            assert np.all(norm.data[:, 0] == 0.0)
+
 
 class TestStandarize:
     """Tests for standardization."""