moved MMD and optim notebooks to their folders

docs/notebooks/MMD_misspec_experiment.ipynb → docs/notebooks/OLD_deprecated/MMD_misspec_experiment.ipynb

@@ -2,7 +2,7 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": 2,
    "metadata": {},
    "outputs": [
     {
@@ -11,7 +11,7 @@
        "0"
       ]
      },
-     "execution_count": 1,
+     "execution_count": 2,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -37,6 +37,58 @@
     "\n"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "np.sqrt"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 37,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "w=1.4\n",
+    "sd=3\n",
+    "dg=4\n",
+    "N=400\n",
+    "a = np.random.randn(N,N)/np.sqrt(N)\n",
+    "\n",
+    "covariance_matrix=np.array([\n",
+    "    [dg,0,sd,-w],\n",
+    "    [0,dg,w,sd],\n",
+    "    [sd,w,dg,0],\n",
+    "    [-w,sd,0,dg],\n",
+    "])\n",
+    "N=50\n",
+    "loadings = np.random.randn(N,4)@np.linalg.cholesky(covariance_matrix)/np.sqrt(N)\n",
+    "n1=loadings[:,0]\n",
+    "n2=loadings[:,1]\n",
+    "m1=loadings[:,2]\n",
+    "m2=loadings[:,3]\n",
+    "plt.hist(a.flatten(),bins=100)\n",
+    "\n",
+    "perturbation = np.outer(n1,m1)+np.outer(n2,m2)\n",
+    "a[:N,:N]+=perturbation*2\n",
+    "plt.hist(a.flatten(),bins=100,alpha=0.5)\n",
+    "plt.yscale('log')"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 2,