LS_Reals.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "outside-generator",
   "metadata": {},
   "outputs": [],
   "source": [
    "from __future__ import print_function\n",
    "import argparse\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "import torch.utils.data as data\n",
    "from torchvision import datasets, transforms\n",
    "import numpy as np\n",
    "import torchvision\n",
    "from resnet import *\n",
    "import cvxpy as cp\n",
    "from sklearn.metrics import precision_recall_fscore_support\n",
    "from sklearn.neural_network import MLPClassifier, MLPRegressor\n",
    "import os\n",
    "import copy\n",
    "from scipy import stats\n",
    "from scipy.spatial.distance import cdist\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.gaussian_process import GaussianProcessRegressor\n",
    "from sklearn.gaussian_process.kernels import RBF, WhiteKernel,Matern,DotProduct ,PairwiseKernel\n",
    "import seaborn as sns\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "selected-touch",
   "metadata": {},
   "outputs": [],
   "source": [
    "K_rbf = RBF(length_scale=.9, length_scale_bounds=(1e-2, 1e3))\n",
    "k_dot = DotProduct(sigma_0=1.0, sigma_0_bounds=1e-05)\n",
    "\n",
    "K_matern = Matern(length_scale=40.1, length_scale_bounds=1e-05, nu=6.5)\n",
    "\n",
    "kernel = 1.0 * K_rbf \\\n",
    "    + WhiteKernel(noise_level=0.01, noise_level_bounds=(1e-10, 1e+1))\n",
    "\n",
    "K_RBF = K_rbf\n",
    "\n",
    "\n",
    "def regressor(n,m,DpX, DpY):\n",
    "    gp = GaussianProcessRegressor(kernel=kernel,alpha=0.0).fit(DpX[:, np.newaxis], DpY)\n",
    "    loss_P = np.linalg.norm(gp.predict(DpX[:, np.newaxis])-DpY)\n",
    "    loss_Q = np.linalg.norm(gp.predict(DqX[:, np.newaxis])-DqY)\n",
    "    print('gp loss:',loss_P,loss_Q)\n",
    "    return gp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "interstate-appraisal",
   "metadata": {},
   "outputs": [],
   "source": [
    "def invcdf(x,a):\n",
    "    return (-(1-a)+np.sqrt((1-a)**2+4*a*x))/(2*a)\n",
    "\n",
    "def weight(x,a,b):\n",
    "    return (2*a*x+1-a)/(2*b*x+1-b)\n",
    "\n",
    "def sampling(n,m,a,b,sigma):\n",
    "    DpY = np.array([invcdf(np.random.uniform(),a) for x in range(n)])\n",
    "    DqY = np.array([invcdf(np.random.uniform(),b) for x in range(m)])\n",
    "    DpX = np.random.normal(0,sigma,n)+DpY\n",
    "    DqX = np.random.normal(0,sigma,m)+DqY\n",
    "    return DpX, DpY, DqX, DqY\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "id": "champion-fitting",
   "metadata": {},
   "outputs": [],
   "source": [
    "def w_comp(n,m,a,b,DpX,DqX,gp,DpY,DqY):\n",
    "    rho = 1e-6\n",
    "    pred_p = gp.predict(DpX[:, np.newaxis])\n",
    "    pred_q = gp.predict(DqX[:, np.newaxis])\n",
    "    Kp_u = K_RBF.__call__(pred_p[:, np.newaxis],pred_p[:, np.newaxis])/(n*n)\n",
    "    u, s, vh = np.linalg.svd(Kp_u, full_matrices=True)\n",
    "#     print('svd:',s)\n",
    "    Kpp = K_RBF.__call__(DpY[:, np.newaxis],DpY[:, np.newaxis])\n",
    "    Kpq_u = K_RBF.__call__(pred_p[:, np.newaxis],pred_p[:, np.newaxis])/(n*n)\n",
    "    +K_RBF.__call__(pred_q[:, np.newaxis],pred_q[:, np.newaxis])/(m*m)\n",
    "    - K_RBF.__call__(pred_p[:, np.newaxis],pred_q[:, np.newaxis])/(n*m)\n",
    "    - K_RBF.__call__(pred_q[:, np.newaxis],pred_p[:, np.newaxis])/(n*m)\n",
    "    Kqp_u = K_RBF.__call__(pred_q[:, np.newaxis],pred_p[:, np.newaxis])/(n*m)\n",
    "    Kpp_u = K_RBF.__call__(pred_p[:, np.newaxis],pred_p[:, np.newaxis])/(n*n)\n",
    "\n",
    "    rank = 100\n",
    "    u, s, vh = np.linalg.svd(np.matmul(np.matmul(Kpp.T,Kp_u),Kpp) +rho * Kpp, full_matrices=True)\n",
    "    rhs = (np.sum(np.matmul(Kqp_u,Kpp),axis=0) - np.sum(np.matmul(Kpp_u , Kpp),axis=0))\n",
    "    aa = np.matmul(np.matmul(np.matmul(u[:,0:rank], np.diag(s[0:rank]**(-1))), vh[0:rank,:]),rhs)\n",
    "    \n",
    "    \n",
    "    x = np.linspace(.0, 1.0, num=100)\n",
    "    iw = np.array([(2*b*y/100+1-b)/(2*a*y/100+1-a) for y in range(100)])\n",
    "    loss = np.sqrt(np.sum((np.matmul(aa,K_RBF.__call__(DpY[:, np.newaxis],x[:, np.newaxis]))+1-iw)**2)/100)\n",
    "    \n",
    "    return aa, loss \n",
    "\n",
    "\n",
    "\n",
    "def eval_alpha(x,alpha,DpY):\n",
    "    return alpha.T*K_RBF.__call__(DpY[:, np.newaxis],x[:, np.newaxis])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ancient-jacksonville",
   "metadata": {},
   "outputs": [],
   "source": [
    "m_max = 10000\n",
    "max_ite = 10\n",
    "max_epoch = 10\n",
    "stor_alpha = np.zeros((max_epoch,max_ite,m_max))\n",
    "stor_y = np.zeros((max_epoch,max_ite,m_max))\n",
    "stor_loss = np.zeros((max_epoch,max_ite))\n",
    "\n",
    "nc_vec = np.zeros((max_epoch))\n",
    "for epoch in range(max_epoch):\n",
    "    print('epoch: ',epoch)\n",
    "    for ite in range(max_ite):\n",
    "        n = 1000 * (epoch+1)\n",
    "        nc_vec[epoch] = n\n",
    "        m = 1000 * (epoch+1)\n",
    "        #samples\n",
    "        sigma = 0.01\n",
    "        a = .5\n",
    "        b = -.5\n",
    "        DpX, DpY, DqX, DqY=sampling(n,m,a,b,sigma)\n",
    "        gp = regressor(n,m, DpX, DpY)\n",
    "\n",
    "        alpha, loss = w_comp(n,m,a,b,DpX,DqX,gp,DpY,DqY)\n",
    "        stor_alpha[epoch,ite,0:m] = alpha\n",
    "        stor_y[epoch,ite,0:n] = DpY\n",
    "        stor_loss[epoch,ite] = loss\n",
    "        print(loss)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "id": "administrative-overall",
   "metadata": {},
   "outputs": [],
   "source": [
    "iw = np.array([(2*b*y/100+1-b)/(2*a*y/100+1-a) for y in range(100)])\n",
    "nn = np.sqrt(np.sum(iw**2)/100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "id": "viral-opinion",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "y_mean = np.mean(stor_loss/nn,axis = 1)\n",
    "\n",
    "y_std = np.std(stor_loss/nn,axis = 1)\n",
    "\n",
    "\n",
    "\n",
    "sns.set_theme()\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "\n",
    "\n",
    "ax.plot(nc_vec.astype(int),y_mean)\n",
    "ax.set_xticks(nc_vec)\n",
    "ax.fill_between(nc_vec.astype(int), (y_mean-y_std/2), (y_mean+y_std/2), color='b', alpha=.4)\n",
    "\n",
    "ax.set_xlabel('Number of Samples')\n",
    "ax.set_ylabel('Relative Error')\n",
    "\n",
    "sns.set_style(style='whitegrid')\n",
    "sns.set_style('ticks')\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "tropical-french",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.8.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}