B_random_decay_mdl.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The autoreload extension is already loaded. To reload it, use:\n",
      "  %reload_ext autoreload\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "from os.path import join as oj\n",
    "import sys, time\n",
    "sys.path.insert(1, oj(sys.path[0], '..'))  # insert parent path\n",
    "import seaborn as sns\n",
    "from sklearn.model_selection import train_test_split\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pmlb\n",
    "from tqdm import tqdm\n",
    "from copy import deepcopy\n",
    "import pickle as pkl\n",
    "import pandas as pd\n",
    "import data\n",
    "import numpy.linalg as npl \n",
    "import numpy.random as npr\n",
    "from scipy.optimize import minimize\n",
    "# from process_results import process_results, aggregate_results, best_ridge\n",
    "from scipy.stats import ortho_group\n",
    "from sklearn.linear_model import LinearRegression, RidgeCV\n",
    "import scipy.linalg as scl\n",
    "import pickle\n",
    "%matplotlib inline\n",
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "save_dir = 'results'\n",
    "os.makedirs(save_dir, exist_ok=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib as mpl\n",
    "label_size = 20\n",
    "mpl.rcParams['xtick.labelsize'] = label_size \n",
    "mpl.rcParams['ytick.labelsize'] = label_size \n",
    "mpl.rcParams['axes.labelsize'] = label_size\n",
    "mpl.rcParams['axes.titlesize'] = 14\n",
    "mpl.rcParams['figure.titlesize'] = 10\n",
    "# mpl.rcParams['lines.markersize'] = 8\n",
    "mpl.rcParams['grid.linewidth'] = 2.5\n",
    "mpl.rcParams['legend.fontsize'] = 20\n",
    "# pylab.rcParams['xtick.major.pad']=5\n",
    "# pylab.rcParams['ytick.major.pad']=5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "cs=sns.color_palette()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## MDL COMP FOR different eigenvalues decay of Matrices"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.8711756154949768"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.sum(np.log(1+f**2 *eigenvals[:min(n, d)] * (w**2/noise_std**2)[:min(n, d)])) / n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "noise_std = 1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.14172606273661373"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.sum(np.log(1+f**2 *eigenvals[:min(n, d)] * (w**2/noise_std**2)[:min(n, d)])) / n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/3 [00:00<?, ?it/s]\n",
      "0it [00:00, ?it/s]\u001b[A\n",
      "3it [00:00, 22.87it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 100 0 20 200\n",
      "0 100 1 20 200\n",
      "0 400 0 20 200\n",
      "0 400 1 20 200\n",
      "0 100 0 100 200\n",
      "0 100 1 100 200\n",
      "0 400 0 100 200\n",
      "0 400 1 100 200\n",
      "0 100 0 200 200\n",
      "0 100 1 200 200\n",
      "0 400 0 200 200\n",
      "0 400 1 200 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "4it [00:00, 10.72it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 100 0 400 200\n",
      "0 100 1 400 200\n",
      "0 400 0 400 200\n",
      "0 400 1 400 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "5it [00:01,  2.19it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 100 0 1000 200\n",
      "0 100 1 1000 200\n",
      "0 400 0 1000 200\n",
      "0 400 1 1000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "6it [00:09,  2.53s/it]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 100 0 2000 200\n",
      "0 100 1 2000 200\n",
      "0 400 0 2000 200\n",
      "0 400 1 2000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "0it [00:00, ?it/s]\u001b[A\n",
      "3it [00:00, 22.50it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 100 0 20 200\n",
      "0 100 1 20 200\n",
      "0 400 0 20 200\n",
      "0 400 1 20 200\n",
      "0 100 0 100 200\n",
      "0 100 1 100 200\n",
      "0 400 0 100 200\n",
      "0 400 1 100 200\n",
      "0 100 0 200 200\n",
      "0 100 1 200 200\n",
      "0 400 0 200 200\n",
      "0 400 1 200 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "4it [00:00, 10.42it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 100 0 400 200\n",
      "0 100 1 400 200\n",
      "0 400 0 400 200\n",
      "0 400 1 400 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "5it [00:01,  2.11it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 100 0 1000 200\n",
      "0 100 1 1000 200\n",
      "0 400 0 1000 200\n",
      "0 400 1 1000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      " 33%|███▎      | 1/3 [00:18<00:37, 18.98s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 100 0 2000 200\n",
      "0 100 1 2000 200\n",
      "0 400 0 2000 200\n",
      "0 400 1 2000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "0it [00:00, ?it/s]\u001b[A\n",
      "3it [00:00, 21.10it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 100 0 20 200\n",
      "1 100 1 20 200\n",
      "1 400 0 20 200\n",
      "1 400 1 20 200\n",
      "1 100 0 100 200\n",
      "1 100 1 100 200\n",
      "1 400 0 100 200\n",
      "1 400 1 100 200\n",
      "1 100 0 200 200\n",
      "1 100 1 200 200\n",
      "1 400 0 200 200\n",
      "1 400 1 200 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "4it [00:00,  9.76it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 100 0 400 200\n",
      "1 100 1 400 200\n",
      "1 400 0 400 200\n",
      "1 400 1 400 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "5it [00:01,  1.97it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 100 0 1000 200\n",
      "1 100 1 1000 200\n",
      "1 400 0 1000 200\n",
      "1 400 1 1000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "6it [00:09,  2.58s/it]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 100 0 2000 200\n",
      "1 100 1 2000 200\n",
      "1 400 0 2000 200\n",
      "1 400 1 2000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "0it [00:00, ?it/s]\u001b[A\n",
      "3it [00:00, 29.38it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 100 0 20 200\n",
      "1 100 1 20 200\n",
      "1 400 0 20 200\n",
      "1 400 1 20 200\n",
      "1 100 0 100 200\n",
      "1 100 1 100 200\n",
      "1 400 0 100 200\n",
      "1 400 1 100 200\n",
      "1 100 0 200 200\n",
      "1 100 1 200 200\n",
      "1 400 0 200 200\n",
      "1 400 1 200 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "4it [00:00, 14.51it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 100 0 400 200\n",
      "1 100 1 400 200\n",
      "1 400 0 400 200\n",
      "1 400 1 400 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "5it [00:01,  3.16it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 100 0 1000 200\n",
      "1 100 1 1000 200\n",
      "1 400 0 1000 200\n",
      "1 400 1 1000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      " 67%|██████▋   | 2/3 [00:35<00:18, 18.35s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 100 0 2000 200\n",
      "1 100 1 2000 200\n",
      "1 400 0 2000 200\n",
      "1 400 1 2000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "0it [00:00, ?it/s]\u001b[A\n",
      "3it [00:00, 27.36it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 100 0 20 200\n",
      "2 100 1 20 200\n",
      "2 400 0 20 200\n",
      "2 400 1 20 200\n",
      "2 100 0 100 200\n",
      "2 100 1 100 200\n",
      "2 400 0 100 200\n",
      "2 400 1 100 200\n",
      "2 100 0 200 200\n",
      "2 100 1 200 200\n",
      "2 400 0 200 200\n",
      "2 400 1 200 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "4it [00:00, 14.21it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 100 0 400 200\n",
      "2 100 1 400 200\n",
      "2 400 0 400 200\n",
      "2 400 1 400 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "5it [00:01,  3.06it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 100 0 1000 200\n",
      "2 100 1 1000 200\n",
      "2 400 0 1000 200\n",
      "2 400 1 1000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "6it [00:31,  9.24s/it]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 100 0 2000 200\n",
      "2 100 1 2000 200\n",
      "2 400 0 2000 200\n",
      "2 400 1 2000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "0it [00:00, ?it/s]\u001b[A\n",
      "3it [00:00, 28.66it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 100 0 20 200\n",
      "2 100 1 20 200\n",
      "2 400 0 20 200\n",
      "2 400 1 20 200\n",
      "2 100 0 100 200\n",
      "2 100 1 100 200\n",
      "2 400 0 100 200\n",
      "2 400 1 100 200\n",
      "2 100 0 200 200\n",
      "2 100 1 200 200\n",
      "2 400 0 200 200\n",
      "2 400 1 200 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "4it [00:00, 13.79it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 100 0 400 200\n",
      "2 100 1 400 200\n",
      "2 400 0 400 200\n",
      "2 400 1 400 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "5it [00:01,  2.94it/s]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 100 0 1000 200\n",
      "2 100 1 1000 200\n",
      "2 400 0 1000 200\n",
      "2 400 1 1000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "100%|██████████| 3/3 [01:15<00:00, 24.67s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 100 0 2000 200\n",
      "2 100 1 2000 200\n",
      "2 400 0 2000 200\n",
      "2 400 1 2000 200\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "save_data = True\n",
    "\n",
    "npr.seed(121123)\n",
    "\n",
    "\n",
    "noise_std = 0.1\n",
    "d0s = np.array([100, 400], dtype=int)\n",
    "ds = np.array([20, 100, 200, 400, 1000, 2000])\n",
    "ns = np.array([200])\n",
    "epows = [0., 0.5]\n",
    "\n",
    "num_norms = 2\n",
    "num_trials = 3\n",
    "rmt_mdl_comp = np.zeros((num_trials, num_norms, len(ds), len(ns), len(d0s), len(epows)))\n",
    "rmt_mdl_comp_2 = np.zeros((num_trials, num_norms, len(ds), len(ns), len(d0s), len(epows)))\n",
    "rmt_mdl_comp_3 = np.zeros((num_trials, num_norms, len(ds), len(ns), len(d0s), len(epows)))\n",
    "\n",
    "\n",
    "new_rmt_mdl_comp = np.zeros((num_trials,num_norms, len(ds), len(ns), len(d0s), len(epows)))\n",
    "new_rmt_mdl_comp_2 = np.zeros((num_trials,num_norms, len(ds), len(ns), len(d0s), len(epows)))\n",
    "new_rmt_mdl_comp_3 = np.zeros((num_trials,num_norms, len(ds), len(ns), len(d0s), len(epows)))\n",
    "\n",
    "dmax = max(ds)\n",
    "\n",
    "beta_stars = np.zeros((len(d0s), dmax))\n",
    "# initialize beta_Star fixed\n",
    "for l, d0 in enumerate(d0s):\n",
    "    beta_fixed = npr.randn(d0)\n",
    "    beta_fixed /= npl.norm(beta_fixed)\n",
    "    beta_stars[l, :d0] = beta_fixed\n",
    "\n",
    "\n",
    "for m in tqdm(range(num_trials)):\n",
    "    \n",
    "    # change X_train\n",
    "    for p, epow in enumerate(epows):\n",
    "        \n",
    "        eigs = np.diag(np.arange(1, dmax+1)**(-epow))\n",
    "        X_train = npr.randn(max(ns), dmax) @ eigs\n",
    "\n",
    "        y_train_stars = np.zeros((len(d0s), ns[0]))\n",
    "\n",
    "        for ltemp, d0 in enumerate(d0s):\n",
    "            y_train_stars[ltemp, :] = X_train[:, :d0] @ beta_stars[ltemp, :d0]\n",
    "\n",
    "        for j, d in tqdm(enumerate(ds)):\n",
    "            for k, n in enumerate(ns):\n",
    "                X_train_d = X_train[:n, :d]\n",
    "                eigenvals, eigenvecs = scl.eigh(X_train_d.T @ X_train_d)\n",
    "                eigenvals = eigenvals[::-1]\n",
    "                V, s, Uh = scl.svd(X_train_d, full_matrices=False)\n",
    "\n",
    "                for l, d0 in enumerate(d0s):\n",
    "\n",
    "                    w = eigenvecs.T @ beta_stars[l, :d]\n",
    "                    w = w[::-1] # arrange in order\n",
    "\n",
    "                    v = V.T @ y_train_stars[l, :]\n",
    "\n",
    "\n",
    "                    for i in range(num_norms):\n",
    "                        print(m, d0, i, d, n)\n",
    "                        f = (i+1)**2\n",
    "\n",
    "                        rmt_mdl_comp[m, i, j, k, l, p] = np.sum(np.log(1+f**2 *eigenvals[:min(n, d)] * (w**2/noise_std**2)[:min(n, d)])) / n\n",
    "                        rmt_mdl_comp_2[m, i, j, k, l, p]  = rmt_mdl_comp[m, i, j, k, l, p] + np.sum(np.log((noise_std**2 / (f**2 * w**2))[:min(n, d)])) / n\n",
    "                        rmt_mdl_comp_3[m, i, j, k, l, p] = rmt_mdl_comp_2[m, i, j, k, l, p] - np.log(0.005) * min(n, d)/n\n",
    "\n",
    "\n",
    "                        new_rmt_mdl_comp[m, i, j, k, l, p] = np.sum(np.log(1+f**2 * v**2/noise_std**2)) / n\n",
    "                        new_rmt_mdl_comp_2[m, i, j, k, l, p] = new_rmt_mdl_comp[m, i, j, k, l, p] + np.sum(np.log( noise_std**2 / (f**2 * v**2))) / n\n",
    "                        new_rmt_mdl_comp_3[m, i, j, k, l, p] = new_rmt_mdl_comp_2[m, i, j, k, l, p] - np.log(0.005) *min(n, d)/n "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "if save_data:\n",
    "    results = {'ns': ns, 'd0s': d0s,  'ds' : ds,\n",
    "               'epows': epows,\n",
    "               'num_trials': num_trials, \n",
    "               'num_norms': num_norms,\n",
    "               'noise_var': noise_std**2,\n",
    "                'rmt_mdl_comp': rmt_mdl_comp,\n",
    "               'rmt_mdl_comp_2':rmt_mdl_comp_2,\n",
    "               'rmt_mdl_comp_3' : rmt_mdl_comp_3,\n",
    "               'new_rmt_mdl_comp': new_rmt_mdl_comp,\n",
    "               'new_rmt_mdl_comp_2': new_rmt_mdl_comp_2,\n",
    "               'new_rmt_mdl_comp_3': new_rmt_mdl_comp_3\n",
    "              }\n",
    "    file = open(oj('intermediate', \n",
    "                   'decay_mdl_n_%d_noise_%d.txt'%(ns[0], noise_std)), 'wb')\n",
    "    pickle.dump(results, file)\n",
    "    file.close()\n",
    "    \n",
    "# results = pickle.load(open(oj('decay_mdl_n_200.txt', 'intermediate'), 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "save_fig = True\n",
    "for ff, (file, noise_std) in enumerate(zip(['decay_mdl_n_200_noise_1.txt', 'decay_mdl_n_200_noise_0.txt'], [1, .1])):\n",
    "    results = pickle.load(open(oj('intermediate', file), 'rb'))\n",
    "    \n",
    "    num_norms = results[\"num_norms\"]\n",
    "    d0s = results[\"d0s\"]\n",
    "    ns = results[\"ns\"]\n",
    "    ds = results[\"ds\"]\n",
    "    rmt_mdl_comp = results[\"rmt_mdl_comp\"]\n",
    "    rmt_mdl_comp_2 = results[\"rmt_mdl_comp_2\"]\n",
    "    rmt_mdl_comp_3 = results[\"rmt_mdl_comp_3\"]\n",
    "\n",
    "\n",
    "    lss = ['-', '--', '-.', ':']\n",
    "    mss = ['s', 'd', 'o', '^']\n",
    "    n = ns[0]\n",
    "    gammas = ds/n\n",
    "\n",
    "    for i in range(1):\n",
    "#         plt.figure(figsize=[8, 6])\n",
    "\n",
    "#         plt.loglog(gammas, rmt_mdl_comp.mean(0)[i, :, 0, 0, 0],   color=cs[0], linestyle=lss[0], linewidth=4.,  marker=mss[0],  markersize=10, label=r'$d_\\star=%d, \\alpha=%.1f$'%(d0s[0], epows[0]))\n",
    "#         plt.loglog(gammas, rmt_mdl_comp.mean(0)[i, :, 0, 0, 1],   color=cs[1], linestyle=lss[1], linewidth=4.,  marker=mss[1],  markersize=10, label=r'$d_\\star=%d, \\alpha=%.1f$'%(d0s[0], epows[1]))\n",
    "#         plt.loglog(gammas, rmt_mdl_comp.mean(0)[i, :, 0, 1, 0],   color=cs[2], linestyle=lss[2], linewidth=4.,  marker=mss[2],  markersize=10, label=r'$d_\\star=%d, \\alpha=%.1f$'%(d0s[1], epows[0]))\n",
    "#         plt.loglog(gammas, rmt_mdl_comp.mean(0)[i, :, 0, 1, 1],   color=cs[3], linestyle=lss[3], linewidth=4.,  marker=mss[3],  markersize=10, label=r'$d_\\star=%d, \\alpha=%.1f$'%(d0s[1], epows[1]))\n",
    "\n",
    "#         plt.grid(True, alpha=0.5, linewidth=1.)\n",
    "#         plt.legend(frameon=False, fontsize=20) \n",
    "#         if ff==1:\n",
    "#             plt.ylim([5e-2, 5e0])\n",
    "#         plt.title(r'Behavior of MDL-COMP ($n=200$)', fontsize=20)\n",
    "#         plt.xlabel('d/n')\n",
    "#         plt.tight_layout()\n",
    "#         if save_fig:\n",
    "#             plt.savefig('fig3/decay_mdl_comp_norm_%d_n_%d_noise_std_%d.pdf'%((i+1)**2, n, noise_std))\n",
    "#         plt.show()\n",
    "\n",
    "        plt.figure(figsize=[8, 6])\n",
    "\n",
    "        plt.loglog(gammas, rmt_mdl_comp_2.mean(0)[i, :, 0, 0, 0],   color=cs[0], linestyle=lss[0], linewidth=4.,  marker=mss[0],  markersize=10, label=r'$d_\\star=%d, \\alpha=%.1f$'%(d0s[0], epows[0]))\n",
    "        plt.loglog(gammas, rmt_mdl_comp_2.mean(0)[i, :, 0, 0, 1],   color=cs[1], linestyle=lss[1], linewidth=4.,  marker=mss[1],  markersize=10, label=r'$d_\\star=%d, \\alpha=%.1f$'%(d0s[0], epows[1]))\n",
    "        plt.loglog(gammas, rmt_mdl_comp_2.mean(0)[i, :, 0, 1, 0],   color=cs[2], linestyle=lss[2], linewidth=4.,  marker=mss[2],  markersize=10, label=r'$d_\\star=%d, \\alpha=%.1f$'%(d0s[1], epows[0]))\n",
    "        plt.loglog(gammas, rmt_mdl_comp_2.mean(0)[i, :, 0, 1, 1],   color=cs[3], linestyle=lss[3], linewidth=4.,  marker=mss[3],  markersize=10, label=r'$d_\\star=%d, \\alpha=%.1f$'%(d0s[1], epows[1]))\n",
    "\n",
    "        plt.grid(True, alpha=0.5, linewidth=1.)\n",
    "        plt.legend(frameon=False, fontsize=20) \n",
    "        plt.xlabel('d/n')\n",
    "    #     plt.ylim([5e-2, 5e0])\n",
    "        plt.title(r'Behavior of MDL-COMP for different designs', fontsize=20)\n",
    "        plt.tight_layout()\n",
    "        if save_fig:\n",
    "            plt.savefig('fig3/decay_new_mdl_comp_norm_%d_n_%d_noise_std_%d.pdf'%((i+1)**2, n, noise_std))\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "def loc_calc_thetahat(l):\n",
    "    inv = scl.pinv(X_train_d.T @ X_train_d + l * np.eye(d))\n",
    "#     inv = eigenvecs @ npl.pinv(np.eye(d) + np.diag(eigenvals)) @ eigenvecs.T\n",
    "    return inv @ X_train_d.T @ y_train_n\n",
    "\n",
    "def loc_mdl1_loss(l):\n",
    "    thetahat = loc_calc_thetahat(l)\n",
    "    mse_norm = npl.norm(y_train_n - X_train_d @ thetahat)**2 / (2 * var)\n",
    "    theta_norm = npl.norm(thetahat)**2 / (2 * var)\n",
    "    eigensum = 0.5 * np.sum(np.log((eigenvals + l) / l))\n",
    "    return (mse_norm + theta_norm + eigensum) / n\n",
    "\n",
    "\n",
    "def loc_mdl1_loss_ret_all(l):\n",
    "    thetahat = loc_calc_thetahat(l)\n",
    "    mse_norm = npl.norm(y_train_n - X_train_d @ thetahat)**2 / (2 * var)\n",
    "    theta_norm = npl.norm(thetahat)**2 / (2 * var)\n",
    "    eigensum = 0.5 * np.sum(np.log((eigenvals + l) / l))\n",
    "    return mse_norm/n , theta_norm/n,  eigensum / n\n",
    "\n",
    "def g(a, b):\n",
    "    val = np.sqrt(a*(1+np.sqrt(b))**2+1)\n",
    "    val -= np.sqrt(a*(1-np.sqrt(b))**2+1)\n",
    "    val = val**2\n",
    "    return(val/4)\n",
    "\n",
    "def thm3(gammas, snr):\n",
    "    delta = g(snr, gammas)\n",
    "    bnd = gammas * np.log(1+snr-delta)\n",
    "    bnd += np.log(1+gammas * snr - delta)\n",
    "    bnd -= delta/snr\n",
    "    return(bnd)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "ename": "KeyError",
     "evalue": "'eigsum'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyError\u001b[0m                                  Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-58-7d83202ea20c>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0md0s\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mresults\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"d0s\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0mrmt_mdl_comp\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mresults\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"rmt_mdl_comp\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 10\u001b[0;31m \u001b[0meigsum\u001b[0m  \u001b[0;34m=\u001b[0m \u001b[0mresults\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"eigsum\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     11\u001b[0m \u001b[0mn\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mns\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m \u001b[0mgammas\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mds\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0mn\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyError\u001b[0m: 'eigsum'"
     ]
    }
   ],
   "source": [
    "results = pickle.load(open('prac_mdl_n_200_noise_0.txt', 'rb'))\n",
    "\n",
    "# results = pickle.load(open('prac_mdl_n_%d_noise_%d.txt'%(ns[0], noise_std), 'rb'))\n",
    "save_fig = False\n",
    "\n",
    "ds = results[\"ds\"]\n",
    "ns = results[\"ns\"]\n",
    "d0s = results[\"d0s\"]\n",
    "rmt_mdl_comp = results[\"rmt_mdl_comp\"]\n",
    "eigsum  = results[\"eigsum\"]\n",
    "n = ns[0]\n",
    "gammas = ds/n\n",
    "\n",
    "lss = ['-', '--', '-.', ':']\n",
    "mss = ['s', 'd', 'o', '^']\n",
    "\n",
    "for i in range(1):\n",
    "    for l, d0 in enumerate(d0s):\n",
    "        plt.figure(figsize=[8, 6])\n",
    "\n",
    "        plt.loglog(gammas, rmt_mdl_comp.mean(0)[i, :, 0, l, 0],   color=cs[0], linestyle=lss[0], linewidth=4.,  marker=mss[0],  markersize=10, label='MDL-COMP (Thm1)')\n",
    "        plt.loglog(gammas, thm3(gammas, np.minimum(d0,ds)/ds),  color=cs[1], linestyle=lss[1], linewidth=4.,  marker=mss[1], markersize=10, label='MDL-COMP (Thm2)')\n",
    "        plt.loglog(gammas, eigsum.mean(0)[i, :, 0, l, 0], color=cs[2], linestyle=lss[2], linewidth=4.,  marker=mss[2], markersize=10, label= 'Prac-MDL-COMP (Eigensum)')\n",
    "        plt.grid(True, alpha=0.5, linewidth=1.)\n",
    "#         plt.ylim([1e-3, 8e-1])\n",
    "        plt.legend(frameon=False, fontsize=20) \n",
    "        plt.title(r'Different MDL-COMP ($n=%d, d_\\star=%d$)'%(ns[0], d0), fontsize=20)\n",
    "        plt.xlabel('d/n')\n",
    "        plt.tight_layout()\n",
    "        if save_fig:\n",
    "            plt.savefig('fig3/prac_mdl_comp_norm_%d_n_%d_noise_%d_d0_%d.pdf'%((i+1)**2, n, noise_std, d0))\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}