New propagation library created

c3/libraries/propagation.py

@@ -296,7 +296,7 @@ def tf_expm(A, terms):
     return r
 
 
-def tf_expm_dynamic(A, acc=1e-4):
+def tf_expm_dynamic(A, acc=1e-5):
     """
     Matrix exponential by the series method with specified accuracy.
 

c3/utils/tf_utils.py

@@ -208,7 +208,6 @@ def Id_like(A):
 @tf.function
 def tf_kron(A, B):
     """Kronecker product of 2 matrices."""
-    # TODO make kronecker product general to different dimensions
     dims = tf.shape(A) * tf.shape(B)
     tensordot = tf.tensordot(A, B, axes=0)
     reshaped = tf.reshape(tf.transpose(tensordot, perm=[0, 2, 1, 3]), dims)

test/test_exp.py

@@ -0,0 +1,41 @@
+#!/usr/bin/env python
+# coding: utf-8
+
+# In[3]:
+
+
+import numpy as np
+import tensorflow as tf
+import pytest
+
+from numpy.testing import assert_array_almost_equal as almost_equal
+from c3.libraries.constants import Id, X, Y, Z
+from c3.libraries.propagation import tf_expm, tf_expm_dynamic
+
+theta = 2 * np.pi * np.random.rand(1)
+
+
+@pytest.mark.unit
+def test_exponentiation_1() -> None:
+    """Testing that exponentiation methods are almost equal in the numpy sense.
+    Check that, given P = a*X+b*Y+c*Z with a, b, c random normalized numbers,
+    exp(i theta P) = cos(theta)*Id + sin(theta)*P"""
+    a = np.random.rand(1)
+    b = np.random.rand(1)
+    c = np.random.rand(1)
+    norm = np.sqrt(a ** 2 + b ** 2 + c ** 2)
+    # Normalized random coefficients
+    a = a / norm
+    b = b / norm
+    c = c / norm
+
+    P = a * X + b * Y + c * Z
+    theta = 2 * np.pi * np.random.rand(1)
+    rot = theta * P
+    res = np.cos(theta) * Id + 1j * np.sin(theta) * P
+
+    terms = 25
+
+    almost_equal(tf_expm(1j * rot, terms), res)
+    almost_equal(tf_expm_dynamic(1j * rot), res)
+    almost_equal(tf.linalg.expm(1j * rot), res)