From f4f0cfe0510def48736c4f9f0a4656bf5fb89200 Mon Sep 17 00:00:00 2001
From: David Luevano Alvarado <55825613+luevano@users.noreply.github.com>
Date: Tue, 3 Mar 2020 23:44:51 -0700
Subject: Fix normal ml error

---
 ml_exp/do_ml.py   | 54 ++++++++++++++++++++++++++++++++++--------------------
 ml_exp/kernels.py |  9 ---------
 2 files changed, 34 insertions(+), 29 deletions(-)

diff --git a/ml_exp/do_ml.py b/ml_exp/do_ml.py
index 379d0efd0..57661578a 100644
--- a/ml_exp/do_ml.py
+++ b/ml_exp/do_ml.py
@@ -79,36 +79,50 @@ def simple_ml(descriptors,
         printc(f'\tTest size: {test_size}', 'CYAN')
         printc(f'\tSigma: {test_size}', 'CYAN')
 
-    X_training = descriptors[:training_size]
-    Y_training = energies[:training_size]
-    K_training = gaussian_kernel(X_training,
-                                 X_training,
-                                 sigma,
-                                 use_tf=use_tf)
     if use_tf:
+        X_training = descriptors[:training_size]
+        Y_training = energies[:training_size]
+        K_training = gaussian_kernel(X_training,
+                                     X_training,
+                                     sigma,
+                                     use_tf=use_tf)
+
         # Y_training = tf.expand_dims(Y_training, 1)
         alpha = tf.linalg.cholesky_solve(tf.linalg.cholesky(K_training),
                                          Y_training)
-    else:
-        alpha = LA.cho_solve(LA.cho_factor(K_training),
-                             Y_training)
 
-    X_test = descriptors[-test_size:]
-    Y_test = energies[-test_size:]
-    K_test = gaussian_kernel(X_test,
-                             X_training,
-                             sigma,
-                             use_tf=use_tf)
-    if use_tf:
+        X_test = descriptors[-test_size:]
+        Y_test = energies[-test_size:]
+        K_test = gaussian_kernel(X_test,
+                                 X_training,
+                                 sigma,
+                                 use_tf=use_tf)
+
         # Y_test = tf.expand_dims(Y_test, 1)
         Y_predicted = tf.tensordot(K_test, alpha, 1)
-    else:
-        Y_predicted = np.dot(K_test, alpha)
 
-    print('Ducky')
-    if use_tf:
         mae = tf.reduce_mean(tf.abs(Y_predicted - Y_test))
     else:
+        X_training = descriptors[:training_size]
+        Y_training = energies[:training_size]
+        K_training = gaussian_kernel(X_training,
+                                     X_training,
+                                     sigma,
+                                     use_tf=use_tf)
+
+        # Adding a small value on the diagonal for cho_solve.
+        K_training[np.diag_indices_from(K_training)] += 1e-8
+        alpha = LA.cho_solve(LA.cho_factor(K_training),
+                             Y_training)
+
+        X_test = descriptors[-test_size:]
+        Y_test = energies[-test_size:]
+        K_test = gaussian_kernel(X_test,
+                                 X_training,
+                                 sigma,
+                                 use_tf=use_tf)
+        Y_predicted = np.dot(K_test, alpha)
+
         mae = np.mean(np.abs(Y_predicted - Y_test))
 
     if show_msgs:
diff --git a/ml_exp/kernels.py b/ml_exp/kernels.py
index 26ff0d77b..c203af30e 100644
--- a/ml_exp/kernels.py
+++ b/ml_exp/kernels.py
@@ -73,13 +73,4 @@ def gaussian_kernel(X1,
             else:
                 norm = np.linalg.norm(X2 - X1[i], axis=-1)
             K[i, :] = np.exp(i_sigma * np.square(norm))
-
-        # Old way of calculating the kernel (numba support).
-        """
-        for i, x1 in enumerate(X1):
-            for j, x2 in enumerate(X2):
-                f_norm = np.linalg.norm(x2 - x1)
-                K[i, j] = math.exp(i_sigma * f_norm**2)
-        """
-
     return K
-- 
cgit v1.2.3-70-g09d2