{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from pathlib import Path\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import scipy\n",
    "from glob import glob"
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "Prior code from mva:"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "outputs": [],
   "source": [
    "def read_xy(path: str):\n",
    "    df = pd.read_csv(path, skipinitialspace=True)\n",
    "    df['N'] = df.index + 1\n",
    "    df['r'] = (df.x ** 2 + df.y ** 2) ** 0.5\n",
    "    df['cr'] = df.r.cummax()\n",
    "    df['fd'] = np.log(df.N) / np.log(df.cr)\n",
    "\n",
    "    return df\n",
    "\n",
    "\n",
    "def read_all(load_dir: str):\n",
    "    paths = glob(f'{load_dir}/*.csv')\n",
    "    return [read_xy(path) for path in paths]\n",
    "\n",
    "\n",
    "# We expect the format of sp data to be:\n",
    "# ROOT/{probability}/{RUN_ID}.csv\n",
    "\n",
    "def read_sp_dir(load_dir: str):\n",
    "    probability = float(Path(load_dir).name)\n",
    "    return probability, read_all(load_dir)\n",
    "\n",
    "\n",
    "def read_sp_full(root: str):\n",
    "    raw_data = [read_sp_dir(load_dir) for load_dir in glob(f'{root}/*')]\n",
    "    fd_data = [fd_of_dfs(dfs) for (p, dfs) in raw_data]\n",
    "    probabilities = [p for (p, dfs) in raw_data]\n",
    "    fd_means, fd_stds = list(zip(*fd_data))\n",
    "    out = pd.DataFrame(zip(probabilities, fd_means, fd_stds), columns=['p', 'fd_mean', 'fd_std'])\n",
    "    out.sort_values(by='p', inplace=True)\n",
    "\n",
    "    return out"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "def convergent_tail_index(series, tol):\n",
    "    diffs = np.abs(np.ediff1d(series))\n",
    "    for i in range(0, len(diffs)):\n",
    "        if np.max(diffs[i:]) <= tol:\n",
    "            return i\n",
    "\n",
    "    # No convergence found\n",
    "    return None\n",
    "\n",
    "\n",
    "def mean_of_tail(series, tol=0.05):\n",
    "    tail_index = convergent_tail_index(series, tol)\n",
    "    return np.mean(series[tail_index:])\n",
    "\n",
    "\n",
    "def fd_of_dfs(dfs):\n",
    "    fds = [mean_of_tail(df.fd, 0.001) for df in dfs]\n",
    "    fds_clean = [f for f in fds if f < np.inf]\n",
    "    return np.mean(fds_clean), np.std(fds_clean)"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "c_sticking_probability = read_sp_full(\"../data/stick-probability\")"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "rust_sticking_probability = read_sp_full(\"../data/rust-stick-probability\")"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "plt.plot(c_sticking_probability.p, c_sticking_probability.fd_mean)\n",
    "plt.plot(rust_sticking_probability.p, rust_sticking_probability.fd_mean)"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "plt.errorbar(c_sticking_probability.p, c_sticking_probability.fd_mean, c_sticking_probability.fd_std, fmt=\"o\")\n",
    "plt.errorbar(rust_sticking_probability.p, rust_sticking_probability.fd_mean, rust_sticking_probability.fd_std, fmt=\"o\")"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Hmmm even with a different RNG I still get a dip around 0.6, I think theres something hiding there"
   ],
   "metadata": {
    "collapsed": false
   }
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}