how-many-samples.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import corner"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generate fake data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_covariance_matrix(n_dim, min_correlation, max_correlation):\n",
    "    \"\"\" Generate a random covariance matrix with the specified dimensions and correlation range.\n",
    "    \"\"\"\n",
    "    # Initialize a matrix of zeros\n",
    "    A = np.zeros((n_dim, n_dim))\n",
    "\n",
    "    # Populate the matrix with random correlations within the specified range\n",
    "    for i in range(n_dim):\n",
    "        for j in range(i, n_dim):\n",
    "            if i == j:\n",
    "                A[i, j] = 1  # Variance of 1 along the diagonal\n",
    "            else:\n",
    "                A[i, j] = A[j, i] = np.random.uniform(min_correlation, max_correlation)\n",
    "\n",
    "    # Ensure the matrix is positive definite\n",
    "    A = A.T.dot(A)\n",
    "    return A"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_dim = 14\n",
    "min_correlation = -1\n",
    "max_correlation = 1\n",
    "\n",
    "A_cov = random_covariance_matrix(n_dim, min_correlation, max_correlation)\n",
    "A_mean = np.random.uniform(-2, 2, n_dim)  # Random mean offset\n",
    "\n",
    "# Train and val sets\n",
    "X_samples = np.random.multivariate_normal(A_mean, A_cov, 100000)\n",
    "X_samples_val = np.random.multivariate_normal(A_mean, A_cov, 1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {