Measure threshold for displaced Gaussian states (#615)

* displaced Gaussian states work with measure threshold, (needs thewalrus >=0.16.0)

* require dev version of thewalrus

* update changelog

* update requirements

* updates thewalrus in requirements

Co-authored-by: Nicolas Quesada <zeitus@gmail.com>

.github/CHANGELOG.md

@@ -138,9 +138,13 @@
   broadcasting.
   [(#567)](https://github.com/XanaduAI/strawberryfields/pull/567)
 
+* `measure_threshold` in the `gaussian` backend now supports displaced Gaussian states.
+  [(#615)](https://github.com/XanaduAI/strawberryfields/pull/615)
+
 * Speed improvements to ``gaussian_unitary`` compiler
   [(#603)](https://github.com/XanaduAI/strawberryfields/pull/603)
 
+
 <h3>Breaking Changes</h3>
 
 <h3>Bug fixes</h3>

requirements.txt

@@ -6,7 +6,7 @@ tensorboard>=2.0
 networkx>=2.0
 git+https://github.com/XanaduAI/blackbird@master#egg=quantum-blackbird
 python-dateutil==2.8.0
-thewalrus>=0.15.0
+thewalrus>=0.16.0
 toml
 appdirs
 numba>=0.48.0

strawberryfields/backends/gaussianbackend/backend.py

@@ -248,16 +248,15 @@ def measure_threshold(self, modes, shots=1, select=None, **kwargs):
 
         mu = self.circuit.mean
         cov = self.circuit.scovmatxp()
+        mean = self.circuit.smeanxp()
         # check we are sampling from a gaussian state with zero mean
-        if not allclose(mu, zeros_like(mu)):
-            raise NotImplementedError(
-                "Threshold measurement is only supported for " "Gaussian states with zero mean"
-            )
+
         x_idxs = array(modes)
         p_idxs = x_idxs + len(mu)
         modes_idxs = concatenate([x_idxs, p_idxs])
         reduced_cov = cov[ix_(modes_idxs, modes_idxs)]
-        samples = torontonian_sample_state(reduced_cov, shots)
+        reduced_mean = mean[modes_idxs]
+        samples = torontonian_sample_state(mu=reduced_mean, cov=reduced_cov, samples=shots)
 
         return samples
 

tests/backend/test_threshold_measurement.py

@@ -64,20 +64,21 @@ def test_vacuum_measurements(self, setup_backend, pure):
             meas = backend.measure_threshold([0, 1, 2])[0]
             assert np.all(np.array(meas) == 0)
 
-
-    def test_binary_outcome(self, setup_backend, pure):
+    @pytest.mark.parametrize("alpha", [0, 1, 0.5+0.5j])
+    def test_binary_outcome(self, setup_backend, pure, alpha):
         """Test that the outcomes of a threshold measurement is zero or one."""
         num_modes = 2
         for _ in range(NUM_REPEATS):
             backend = setup_backend(num_modes)
             backend.reset(pure=pure)
 
             r = 0.5
+
             backend.squeeze(r, 0, 0)
+            backend.displacement(abs(alpha), np.angle(alpha), 0)
             backend.beamsplitter(np.pi/4, np.pi, 0, 1)
             meas_modes = [0, 1]
             meas_results = backend.measure_threshold(meas_modes)
 
             for i in range(num_modes):
                 assert meas_results[0][i] == 0 or meas_results[0][i] == 1
-