Working on the models

model/decoding.py

@@ -14,7 +14,7 @@ from tqdm import trange
 
 import ripser
 
-from .persistence import persistence
+from persistence import persistence
 
 EPSILON = 0.0000000000001
 

model/persistence.py

@@ -16,7 +16,7 @@ import ripser
 from persim import plot_diagrams
 import gudhi
 
-from .decorators import multi_input
+from decorators import multi_input
 
     
 def hausdorff(data1, data2, homdim, coeff):

sym_model.py

@@ -23,7 +23,10 @@ grating_f = sp.cos(k * (x - x0) * sp.cos(theta) + k * (y - y0) * sp.sin(theta) +
 receptive_field = 1 / (2 * sp.pi * sigma * sigma) * sp.exp(-(x ** 2 + y ** 2) / (2 * sigma ** 2)) * sp.cos(
     k * x * sp.cos(theta) + k * y * sp.sin(theta) + phi)
 receptive_field = receptive_field.subs(theta, 0).subs(phi, 0)
-p = sp.cosh(k ** 2 * sigma ** 2 * sp.cos(theta)) * sp.exp(k ** 2 * (1 + sp.cos(theta) ** 2) / 2) * sp.cos(
+# p = sp.cosh(k ** 2 * sigma ** 2 * sp.cos(theta)) * sp.exp(k ** 2 * (1 + sp.cos(theta) ** 2) / 2) * sp.cos(
+#     phi - k * (x0 * sp.cos(theta) + y0 * sp.sin(theta)))
+
+p = sp.cosh(k ** 2 * sigma ** 2 * sp.cos(theta) * 4) * sp.exp(-4 * k ** 2 * sigma ** 2) * sp.cos(
     phi - k * (x0 * sp.cos(theta) + y0 * sp.sin(theta)))
 
 sigma_split = np.arange(0.1, 1, 0.05)
@@ -31,6 +34,10 @@ k_split = np.arange(0.2, 6, 0.2)
 xy_split = np.arange(-1, 1, 0.05)
 
 
+def sigmoid(x):
+    return 1 / (1 + np.exp(-x))
+
+
 @dataclass
 class Cell:
     sigma_val: float = defaults[sigma]
@@ -93,11 +100,17 @@ class Population:
                xy_dist: np.ndarray = np.ones(len(xy_split))):
         return cls(cells=[Cell.random(sigma_dist, k_val, xy_dist) for _ in range(n)])
 
-    def get_response(self, phi_deg: float, theta_deg: float) -> typing.List[float]:
-        return [cell.get_value(theta_deg, phi_deg) for cell in self.cells]
+    def get_response(self, phi_deg: float, theta_deg: float, coef: float = 4, use_sigmoid: bool = True) -> np.ndarray:
+        return (sigmoid if use_sigmoid else (lambda x: x))(np.array([cell.get_value(theta_deg, phi_deg) for cell in self.cells]) * coef)
 
-    def sample_responses(self, n: int) -> np.ndarray:
+    def sample_responses(
+            self, n: int, noise_sigma: float = 0, coef: float = 2, use_sigmoid: bool = True,
+            custom_grid: typing.Optional[np.ndarray] = None
+    ) -> np.ndarray:
         return np.array([
-            self.get_response(phi_deg, theta_deg % 180)
-            for phi_deg, theta_deg in np.random.uniform(0, 360, (n, 2))
-        ])
+            np.array([self.get_response(phi_deg, theta_deg % 180, coef=coef, use_sigmoid=use_sigmoid), np.ones(len(self.cells)) * phi_deg,
+                      np.ones(len(self.cells)) * theta_deg]).swapaxes(0, 1)
+            for phi_deg, theta_deg in (np.random.uniform(0, 360, (n, 2)) if custom_grid is None else custom_grid)
+        ]) + np.random.normal(
+            0, [noise_sigma, 0, 0],
+            (n if custom_grid is None else len(custom_grid), len(self.cells), 3))