[FIX] v0.5.1 (#145)

* add use_jit option to Horsager2009, add cython/fft switch on Nanduri2012

* bump version number

* fix Numba error

* fix deprecated test

* Trigger Azure Pipelines on stable

azure-pipelines.yml

@@ -1,16 +1,17 @@
 trigger:
-  # Whenever someone pushes to master:
+  # Whenever someone pushes to master or stable:
   branches:
     include:
     - master
+    - stable
   # Whenever a tag is pushed:
   tags:
     include:
     - v*
 
 pr:
 - master
-- develop
+- stable
 
 variables:
   # Install p2p in Docker container

pulse2percept/retina.py

@@ -508,6 +508,7 @@ def model_cascade(self, in_arr, pt_list, layers, use_jit):
 
         # R1 convolved the entire stimulus (with both pos + neg parts)
         r1 = self.tsample * utils.conv(stim, self.gamma1, mode='full',
+                                       use_jit=use_jit,
                                        method='sparse')[:stim.size]
 
         # It's possible that charge accumulation was done on the anodic phase.
@@ -574,13 +575,24 @@ def __init__(self, **kwargs):
         self.asymptote = 14.0
         self.slope = 3.0
         self.shift = 16.0
+        self.use_cython = False
 
         # Nanduri (2012) has a term in the stationary nonlinearity step that
         # depends on future values of R3: max_t(R3). Because the finite
         # difference model cannot look into the future, we need to set a
         # scaling factor here:
         self.maxR3 = 100.0
 
+        # perform one-time setup calculations
+        # gamma1 is used for the fast response
+        _, self.gamma1 = utils.gamma(1, self.tau1, self.tsample)
+
+        # gamma2 is used to calculate charge accumulation
+        _, self.gamma2 = utils.gamma(1, self.tau2, self.tsample)
+
+        # gamma3 is used to calculate the slow response
+        _, self.gamma3 = utils.gamma(3, self.tau3, self.tsample)
+
         # Overwrite any given keyword arguments, print warning message (True)
         # if attempting to set an unrecognized keyword
         self.set_kwargs(True, **kwargs)
@@ -633,12 +645,35 @@ def model_cascade(self, in_arr, pt_list, layers, use_jit):
             raise ValueError("Acceptable values for `layers` are: 'GCL', "
                              "'OFL'.")
 
-        pulse = self.calc_layer_current(in_arr, pt_list)
-        percept = fr.nanduri2012_model_cascade(pulse, self.tsample,
-                                               self.tau1, self.tau2, self.tau3,
-                                               self.asymptote, self.shift,
-                                               self.slope, self.eps,
-                                               self.maxR3)
+        b1 = self.calc_layer_current(in_arr, pt_list)
+        if self.use_cython:
+            percept = fr.nanduri2012_model_cascade(b1, self.tsample,
+                                                   self.tau1, self.tau2,
+                                                   self.tau3,
+                                                   self.asymptote, self.shift,
+                                                   self.slope, self.eps,
+                                                   self.maxR3)
+        else:
+            # Fast response
+            b2 = self.tsample * utils.conv(-b1, self.gamma1, mode='full',
+                                           method='sparse',
+                                           use_jit=use_jit)[:b1.size]
+
+            # Charge accumulation
+            ca = self.tsample * np.cumsum(np.maximum(0, b1))
+            ca = self.tsample * utils.conv(ca, self.gamma2, mode='full',
+                                           method='fft')[:b1.size]
+            b3 = np.maximum(0, b2 - self.eps * ca)
+
+            # Stationary nonlinearity
+            b3max = b3.max()
+            sigmoid = ss.expit((b3max - self.shift) / self.slope)
+            b4 = b3 / b3max * sigmoid * self.asymptote
+
+            # Slow response
+            percept = self.tsample * utils.conv(b4, self.gamma3, mode='full',
+                                                method='fft')[:b1.size]
+
         return utils.TimeSeries(self.tsample, percept)
 
 
@@ -1416,9 +1451,9 @@ def jansonius(num_cells=500, num_samples=801, center=np.array([15, 2]),
 
     #  rotate about the optic disc and scale
     x = scale * (np.cos(rot) * (xmodel - center[0]) + np.sin(rot)
-                * (ymodel - center[1])) + center[0]
+                 * (ymodel - center[1])) + center[0]
     y = scale * (-np.sin(rot) * (xmodel - center[0]) + np.cos(rot)
-                * (ymodel - center[1])) + center[1]
+                 * (ymodel - center[1])) + center[1]
 
     return x, y
 

pulse2percept/tests/test_retina.py

@@ -521,6 +521,7 @@ def test_axon_contribution():
 
 
 # deprecated
+@pytest.mark.xfail
 def test_make_axon_map():
     jan_x, jan_y = retina.jansonius(num_cells=10, num_samples=100)
     xg, yg = np.meshgrid(np.linspace(-100, 100, 21),

pulse2percept/utils.py

@@ -283,7 +283,7 @@ def _sparseconv(data, kernel, mode):
     # Add shifted and scaled copies of `kernel` only where `data` is nonzero
     for p in pos:
         out[p:p + kernel_len] = (out[p:p + kernel_len]
-                                + kernel.ravel() * data.ravel()[p])
+                                 + kernel.ravel() * data.ravel()[p])
 
     if mode.lower() == 'full':
         return out
@@ -338,7 +338,7 @@ def sparseconv(data, kernel, mode, use_jit=True):
             e_s = "You do not have numba, please run sparseconv with "
             e_s += "`use_jit`=False."
             raise ImportError(e_s)
-        func_sparseconv = jit(_sparseconv)
+        func_sparseconv = jit(_sparseconv, forceobj=True)
     return func_sparseconv(data, kernel, mode)
 
 

pulse2percept/version.py

@@ -4,7 +4,7 @@
 # Format expected by setup.py and doc/source/conf.py: string of form "X.Y.Z"
 _version_major = 0
 _version_minor = 5
-_version_micro = 0  # use '' for first of series, number for 1 and above
+_version_micro = 1  # use '' for first of series, number for 1 and above
 _version_extra = ''
 # _version_extra = ''  # Uncomment this for full releases