Added markdown explanation cells to main mpy notebook. Fixed some sma…

…ll ulab numpy bugs

experimentation/EEG-Analysis.ipynb

@@ -356,8 +356,8 @@
    "outputs": [],
    "source": [
     "def standardise(X):\n",
-    "    axis = np.argmax(X.shape())\n",
-    "    minor_shape = np.min(X.shape())\n",
+    "    axis = np.argmax(X.shape)\n",
+    "    minor_shape = np.min(X.shape)\n",
     "    mu = np.mean(X, axis=axis).reshape((minor_shape, 1))\n",
     "    sigma = np.std(X, axis=axis).reshape((minor_shape, 1))\n",
     "    return (X-mu)/sigma\n",

experimentation/Eigenvalue Algorithms.ipynb

@@ -86,7 +86,7 @@
     "def solve_eig_qr(A, n_eig, lam_iterations=5):\n",
     "    # !! note: eigenvectors can only be found reliably if A is symmetric\n",
     "    Ak = A\n",
-    "    n_eig = min(n_eig, min(A.shape()))\n",
+    "    n_eig = min(n_eig, min(A.shape))\n",
     "\n",
     "    for k in range(lam_iterations):\n",
     "        Qk, Rk = np.linalg.qr(Ak)\n",
@@ -95,7 +95,7 @@
     "    lam = np.diag(Ak) # get eigenvalues\n",
     "    V = []\n",
     "    for l in lam[:n_eig]: # now find `n_eig` eigenvectors\n",
-    "        A_null = (A - np.eye(A.shape()[0])*l).transpose()\n",
+    "        A_null = (A - np.eye(A.shape[0])*l).transpose()\n",
     "        Q, R = np.linalg.qr(A_null) # compute null space of (A-lam*I) to get eigenvector\n",
     "        V.append(Q[:, -1])\n",
     "    return lam, np.array(V).transpose()\n",
@@ -163,7 +163,7 @@
     "        result = {}\n",
     "        Cxx = np.dot(X_test, X_test.transpose()) # precompute data auto correlation matrix\n",
     "        for f in self.stim_freqs:\n",
-    "            Y = harmonic_reference(f, self.fs, np.max(X_test.shape()), Nh=self.Nh, standardise_out=True)\n",
+    "            Y = harmonic_reference(f, self.fs, np.max(X_test.shape), Nh=self.Nh, standardise_out=True)\n",
     "            rho = self.cca_eig(X_test, Y, Cxx=Cxx) # canonical variable matrices. Xc = X^T.W_x\n",
     "            result[f] = rho\n",
     "        return result\n",

micropython/lib/computation.py

@@ -91,7 +91,7 @@ def max_eig(A, iterations, numeric_method="qr"):
         else:
             if numeric_method != "qr":
                 print("Unknown `numeric_method` arg: defaulting to QR solver")
-            lam, v = solve_eig_qr(A, 1, lam_iterations=iterations)
+            lam, v = solve_eig_qr(A, iterations)
             lam = lam[0]  # only need first eigen val (largest returned first)
             v = v[:, 0]  # only first eig vector
 
@@ -104,16 +104,16 @@ def resample(X, factor):
 
 
 def standardise(X):
-    axis = np.argmax(X.shape())
-    minor_shape = np.min(X.shape())
+    axis = np.argmax(X.shape)
+    minor_shape = np.min(X.shape)
     mu = np.mean(X, axis=axis).reshape((minor_shape, 1))
     sigma = np.std(X, axis=axis).reshape((minor_shape, 1))
     return (X - mu) / sigma
 
 
 def cov(X, Y, biased=False):
     assert (
-        X.shape() == Y.shape() and len(X.shape()) == 1
+        X.shape == Y.shape and len(X.shape) == 1
     ), "Expected data vectors of equal length"
     assert len(X) > 1, "At least 2 data points are required"
 
@@ -126,7 +126,7 @@ def cov(X, Y, biased=False):
 
 def corr(X, Y):
     assert (
-        X.shape() == Y.shape() and len(X.shape()) == 1
+        X.shape == Y.shape and len(X.shape) == 1
     ), "Expected data vectors of equal length"
     assert len(X) > 1, "At least 2 data points are required"
 

micropython/lib/decoding.py

@@ -52,7 +52,7 @@ def harmonic_reference(f0, fs, Ns, Nh=1, standardise_out=False):
       Ns              : # of samples in trial
       Nh          : # of harmonics
     Output:
-      y_ref           : Generated reference signals with shape (Nf, Ns, 2*Nh)
+      X           : Generated reference signals with shape (Nf, Ns, 2*Nh)
     """
     X = np.zeros((Nh * 2, Ns))
 

micropython/lib/runner.py

@@ -8,7 +8,7 @@
 from micropython import schedule
 
 
-class BaseRunner:
+class Runner:
     def __init__(self, stimulus_freqs=None) -> None:
         if stimulus_freqs is None:
             self.stim_freqs = config.STIM_FREQS  # assign defaults
@@ -193,7 +193,7 @@ def _setup_logger(self, log_period, logger_type):
             self.logger = None
 
 
-class OnlineRunner(BaseRunner):
+class OnlineRunner(Runner):
     def setup(
         self,
         spi_params=None,