notebook/demo.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ac7a3091-0874-4ebc-b915-54e4b83ae36f",
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "0b4c2d88-737c-4ae5-9842-f0df6bdd2cb7",
   "metadata": {},
   "outputs": [],
   "source": [
    "from pathlib import Path\n",
    "import sys\n",
    "\n",
    "sys.path.append(str(Path(\"..\").resolve()))\n",
    "from lib3.brownian_motion import ParamBrownianMotion,BrownianMotion\n",
    "from simulation4 import N_seed, x0s, sigmas, get_simulator"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "73168675-5ee3-476e-aef2-eef06c9ca4fd",
   "metadata": {},
   "source": [
    "事前にsimulation4.pyを実行しておく"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4002b3ba-ef6e-43c7-909b-7ce809b37cbe",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "03c7988d-a82b-4ff3-8418-e5e6533afcb9",
   "metadata": {},
   "source": [
    "## 実験結果を検索 (mlflow.search_runs)\n",
    "pandasのDataFrameとして得られるので、その後の加工が楽。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "20d3ec57-f668-4523-8b75-de2b4f3a43a4",
   "metadata": {},
   "outputs": [],
   "source": [
    "import mlflow\n",
    "import simulation4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "13f835b0-b866-4cfc-b67c-5bb21c512809",
   "metadata": {},
   "outputs": [],
   "source": [
    "cache_dir = str(Path(simulation4.__file__).parent.joinpath(\"mlruns\"))\n",
    "mlflow.set_tracking_uri(cache_dir)\n",
    "mlflow_client = mlflow.tracking.MlflowClient(\n",
    "    tracking_uri=cache_dir\n",
    ")\n",
    "exp = mlflow_client.get_experiment_by_name(\"sim4\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5a120402-30a5-45eb-8be0-7488b18c1eb2",
   "metadata": {},
   "source": [
    "### 正常にシミュレーションが終了した結果を全て取得する"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "c8ef7196-7a6a-477f-b0f7-7b81d7197333",
   "metadata": {},
   "outputs": [],
   "source": [
    "query = \"attributes.status = 'FINISHED'\"\n",
    "df_result = mlflow.search_runs(\n",
    "    experiment_ids=[exp.experiment_id],\n",
    "    filter_string=query,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "c4ce877d-765d-439b-b677-7babcbb12a9c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>run_id</th>\n",
       "      <th>experiment_id</th>\n",
       "      <th>status</th>\n",
       "      <th>artifact_uri</th>\n",
       "      <th>start_time</th>\n",
       "      <th>end_time</th>\n",
       "      <th>metrics.state</th>\n",
       "      <th>params.save_full_traj</th>\n",
       "      <th>params.record_per</th>\n",
       "      <th>params.param_bm.seed</th>\n",
       "      <th>params.total_step</th>\n",
       "      <th>params.param_bm.initial_state</th>\n",
       "      <th>params.param_bm.sigma</th>\n",
       "      <th>tags.mlflow.source.type</th>\n",
       "      <th>tags.mlflow.source.name</th>\n",
       "      <th>tags.mlflow.user</th>\n",
       "      <th>tags.mlflow.source.git.commit</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>a00470b61cde450d9836b6f95dc9ca1d</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/a00470b61cde4...</td>\n",
       "      <td>2021-10-15 08:16:21.285000+00:00</td>\n",
       "      <td>2021-10-15 08:16:22.111000+00:00</td>\n",
       "      <td>-0.630327</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>4</td>\n",
       "      <td>500</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>bccf724fcf70463a86a1d89dab17d94d</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/bccf724fcf704...</td>\n",
       "      <td>2021-10-15 08:16:20.152000+00:00</td>\n",
       "      <td>2021-10-15 08:16:20.890000+00:00</td>\n",
       "      <td>-0.815163</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>4</td>\n",
       "      <td>500</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>a579174b220d4b099c22ce268180a1fb</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/a579174b220d4...</td>\n",
       "      <td>2021-10-15 08:16:18.922000+00:00</td>\n",
       "      <td>2021-10-15 08:16:19.780000+00:00</td>\n",
       "      <td>1.369673</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>4</td>\n",
       "      <td>500</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>38c6e52d7c584795afc0ca747ebaffd3</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/38c6e52d7c584...</td>\n",
       "      <td>2021-10-15 08:16:17.561000+00:00</td>\n",
       "      <td>2021-10-15 08:16:18.480000+00:00</td>\n",
       "      <td>1.184837</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>4</td>\n",
       "      <td>500</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>b4296fb013e349dd9f92ea47bbedf81a</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/b4296fb013e34...</td>\n",
       "      <td>2021-10-15 08:16:16.399000+00:00</td>\n",
       "      <td>2021-10-15 08:16:17.190000+00:00</td>\n",
       "      <td>4.499493</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "      <td>500</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>583c6d554d6a4d96a3183469022e69e0</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/583c6d554d6a4...</td>\n",
       "      <td>2021-10-15 08:16:15.361000+00:00</td>\n",
       "      <td>2021-10-15 08:16:16.074000+00:00</td>\n",
       "      <td>1.749746</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "      <td>500</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2d3571f7042a41dda8fb2c4b1fadfe05</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/2d3571f7042a4...</td>\n",
       "      <td>2021-10-15 08:16:14.245000+00:00</td>\n",
       "      <td>2021-10-15 08:16:15.072000+00:00</td>\n",
       "      <td>6.499493</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "      <td>500</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>9f6bfa2a81fa4012b81d8869ff511276</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/9f6bfa2a81fa4...</td>\n",
       "      <td>2021-10-15 08:16:13.105000+00:00</td>\n",
       "      <td>2021-10-15 08:16:13.946000+00:00</td>\n",
       "      <td>3.749746</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "      <td>500</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>d6bc9961c6be4369a42d6beaaebadb7b</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/d6bc9961c6be4...</td>\n",
       "      <td>2021-10-15 08:16:11.465000+00:00</td>\n",
       "      <td>2021-10-15 08:16:12.834000+00:00</td>\n",
       "      <td>-6.341827</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>2</td>\n",
       "      <td>500</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>aaa35b73c37d4844b314942ba59db07f</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/aaa35b73c37d4...</td>\n",
       "      <td>2021-10-15 08:16:09.251000+00:00</td>\n",
       "      <td>2021-10-15 08:16:10.730000+00:00</td>\n",
       "      <td>-3.670913</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>2</td>\n",
       "      <td>500</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>ff13baae46724743af54d5c7e697a9af</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/ff13baae46724...</td>\n",
       "      <td>2021-10-15 08:16:07.761000+00:00</td>\n",
       "      <td>2021-10-15 08:16:08.898000+00:00</td>\n",
       "      <td>-4.341827</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>2</td>\n",
       "      <td>500</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>72f591e3a245401cbd54a31649b4a5ad</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/72f591e3a2454...</td>\n",
       "      <td>2021-10-15 08:16:05.770000+00:00</td>\n",
       "      <td>2021-10-15 08:16:07.334000+00:00</td>\n",
       "      <td>-1.670913</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>2</td>\n",
       "      <td>500</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>a0b4e9494e524b0a8dd271ce77227ebf</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/a0b4e9494e524...</td>\n",
       "      <td>2021-10-15 08:16:04.629000+00:00</td>\n",
       "      <td>2021-10-15 08:16:05.522000+00:00</td>\n",
       "      <td>-4.573832</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>500</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>8ab4e33186974750b7e1cf85256e1c1a</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/8ab4e33186974...</td>\n",
       "      <td>2021-10-15 08:16:03.439000+00:00</td>\n",
       "      <td>2021-10-15 08:16:04.390000+00:00</td>\n",
       "      <td>-2.786916</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>500</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>4b69507e8a7b471082ce36e53e6e65a8</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/4b69507e8a7b4...</td>\n",
       "      <td>2021-10-15 08:16:02.510000+00:00</td>\n",
       "      <td>2021-10-15 08:16:03.256000+00:00</td>\n",
       "      <td>-2.573832</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>500</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>60a966bca911405dbd63903d844e0fc0</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/60a966bca9114...</td>\n",
       "      <td>2021-10-15 08:16:01.609000+00:00</td>\n",
       "      <td>2021-10-15 08:16:02.350000+00:00</td>\n",
       "      <td>-0.786916</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>500</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>1b972798fc3a4685a4237158c2924374</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/1b972798fc3a4...</td>\n",
       "      <td>2021-10-15 08:16:00.441000+00:00</td>\n",
       "      <td>2021-10-15 08:16:01.472000+00:00</td>\n",
       "      <td>-3.688990</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>500</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>d143527f9099445da38d3747a1768034</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/d143527f90994...</td>\n",
       "      <td>2021-10-15 08:15:59.643000+00:00</td>\n",
       "      <td>2021-10-15 08:16:00.324000+00:00</td>\n",
       "      <td>-2.344495</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>500</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>884ef2999cb94f01b378165d1d5e3de7</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/884ef2999cb94...</td>\n",
       "      <td>2021-10-15 08:15:58.648000+00:00</td>\n",
       "      <td>2021-10-15 08:15:59.580000+00:00</td>\n",
       "      <td>-1.688990</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>500</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>1c9f9b2e8afa4dd4a1d568a922218eaa</td>\n",
       "      <td>4</td>\n",
       "      <td>FINISHED</td>\n",
       "      <td>/workspaces/mlflow-demo/mlruns/4/1c9f9b2e8afa4...</td>\n",
       "      <td>2021-10-15 08:15:57.577000+00:00</td>\n",
       "      <td>2021-10-15 08:15:58.599000+00:00</td>\n",
       "      <td>-0.344495</td>\n",
       "      <td>True</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>500</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.1</td>\n",
       "      <td>LOCAL</td>\n",
       "      <td>simulation4.py</td>\n",
       "      <td>vscode</td>\n",
       "      <td>eab15a1ae74715fae02cc0618e06c456b7dd6b83</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                              run_id experiment_id    status  \\\n",
       "0   a00470b61cde450d9836b6f95dc9ca1d             4  FINISHED   \n",
       "1   bccf724fcf70463a86a1d89dab17d94d             4  FINISHED   \n",
       "2   a579174b220d4b099c22ce268180a1fb             4  FINISHED   \n",
       "3   38c6e52d7c584795afc0ca747ebaffd3             4  FINISHED   \n",
       "4   b4296fb013e349dd9f92ea47bbedf81a             4  FINISHED   \n",
       "5   583c6d554d6a4d96a3183469022e69e0             4  FINISHED   \n",
       "6   2d3571f7042a41dda8fb2c4b1fadfe05             4  FINISHED   \n",
       "7   9f6bfa2a81fa4012b81d8869ff511276             4  FINISHED   \n",
       "8   d6bc9961c6be4369a42d6beaaebadb7b             4  FINISHED   \n",
       "9   aaa35b73c37d4844b314942ba59db07f             4  FINISHED   \n",
       "10  ff13baae46724743af54d5c7e697a9af             4  FINISHED   \n",
       "11  72f591e3a245401cbd54a31649b4a5ad             4  FINISHED   \n",
       "12  a0b4e9494e524b0a8dd271ce77227ebf             4  FINISHED   \n",
       "13  8ab4e33186974750b7e1cf85256e1c1a             4  FINISHED   \n",
       "14  4b69507e8a7b471082ce36e53e6e65a8             4  FINISHED   \n",
       "15  60a966bca911405dbd63903d844e0fc0             4  FINISHED   \n",
       "16  1b972798fc3a4685a4237158c2924374             4  FINISHED   \n",
       "17  d143527f9099445da38d3747a1768034             4  FINISHED   \n",
       "18  884ef2999cb94f01b378165d1d5e3de7             4  FINISHED   \n",
       "19  1c9f9b2e8afa4dd4a1d568a922218eaa             4  FINISHED   \n",
       "\n",
       "                                         artifact_uri  \\\n",
       "0   /workspaces/mlflow-demo/mlruns/4/a00470b61cde4...   \n",
       "1   /workspaces/mlflow-demo/mlruns/4/bccf724fcf704...   \n",
       "2   /workspaces/mlflow-demo/mlruns/4/a579174b220d4...   \n",
       "3   /workspaces/mlflow-demo/mlruns/4/38c6e52d7c584...   \n",
       "4   /workspaces/mlflow-demo/mlruns/4/b4296fb013e34...   \n",
       "5   /workspaces/mlflow-demo/mlruns/4/583c6d554d6a4...   \n",
       "6   /workspaces/mlflow-demo/mlruns/4/2d3571f7042a4...   \n",
       "7   /workspaces/mlflow-demo/mlruns/4/9f6bfa2a81fa4...   \n",
       "8   /workspaces/mlflow-demo/mlruns/4/d6bc9961c6be4...   \n",
       "9   /workspaces/mlflow-demo/mlruns/4/aaa35b73c37d4...   \n",
       "10  /workspaces/mlflow-demo/mlruns/4/ff13baae46724...   \n",
       "11  /workspaces/mlflow-demo/mlruns/4/72f591e3a2454...   \n",
       "12  /workspaces/mlflow-demo/mlruns/4/a0b4e9494e524...   \n",
       "13  /workspaces/mlflow-demo/mlruns/4/8ab4e33186974...   \n",
       "14  /workspaces/mlflow-demo/mlruns/4/4b69507e8a7b4...   \n",
       "15  /workspaces/mlflow-demo/mlruns/4/60a966bca9114...   \n",
       "16  /workspaces/mlflow-demo/mlruns/4/1b972798fc3a4...   \n",
       "17  /workspaces/mlflow-demo/mlruns/4/d143527f90994...   \n",
       "18  /workspaces/mlflow-demo/mlruns/4/884ef2999cb94...   \n",
       "19  /workspaces/mlflow-demo/mlruns/4/1c9f9b2e8afa4...   \n",
       "\n",
       "                         start_time                         end_time  \\\n",
       "0  2021-10-15 08:16:21.285000+00:00 2021-10-15 08:16:22.111000+00:00   \n",
       "1  2021-10-15 08:16:20.152000+00:00 2021-10-15 08:16:20.890000+00:00   \n",
       "2  2021-10-15 08:16:18.922000+00:00 2021-10-15 08:16:19.780000+00:00   \n",
       "3  2021-10-15 08:16:17.561000+00:00 2021-10-15 08:16:18.480000+00:00   \n",
       "4  2021-10-15 08:16:16.399000+00:00 2021-10-15 08:16:17.190000+00:00   \n",
       "5  2021-10-15 08:16:15.361000+00:00 2021-10-15 08:16:16.074000+00:00   \n",
       "6  2021-10-15 08:16:14.245000+00:00 2021-10-15 08:16:15.072000+00:00   \n",
       "7  2021-10-15 08:16:13.105000+00:00 2021-10-15 08:16:13.946000+00:00   \n",
       "8  2021-10-15 08:16:11.465000+00:00 2021-10-15 08:16:12.834000+00:00   \n",
       "9  2021-10-15 08:16:09.251000+00:00 2021-10-15 08:16:10.730000+00:00   \n",
       "10 2021-10-15 08:16:07.761000+00:00 2021-10-15 08:16:08.898000+00:00   \n",
       "11 2021-10-15 08:16:05.770000+00:00 2021-10-15 08:16:07.334000+00:00   \n",
       "12 2021-10-15 08:16:04.629000+00:00 2021-10-15 08:16:05.522000+00:00   \n",
       "13 2021-10-15 08:16:03.439000+00:00 2021-10-15 08:16:04.390000+00:00   \n",
       "14 2021-10-15 08:16:02.510000+00:00 2021-10-15 08:16:03.256000+00:00   \n",
       "15 2021-10-15 08:16:01.609000+00:00 2021-10-15 08:16:02.350000+00:00   \n",
       "16 2021-10-15 08:16:00.441000+00:00 2021-10-15 08:16:01.472000+00:00   \n",
       "17 2021-10-15 08:15:59.643000+00:00 2021-10-15 08:16:00.324000+00:00   \n",
       "18 2021-10-15 08:15:58.648000+00:00 2021-10-15 08:15:59.580000+00:00   \n",
       "19 2021-10-15 08:15:57.577000+00:00 2021-10-15 08:15:58.599000+00:00   \n",
       "\n",
       "    metrics.state params.save_full_traj params.record_per  \\\n",
       "0       -0.630327                  True                10   \n",
       "1       -0.815163                  True                10   \n",
       "2        1.369673                  True                10   \n",
       "3        1.184837                  True                10   \n",
       "4        4.499493                  True                10   \n",
       "5        1.749746                  True                10   \n",
       "6        6.499493                  True                10   \n",
       "7        3.749746                  True                10   \n",
       "8       -6.341827                  True                10   \n",
       "9       -3.670913                  True                10   \n",
       "10      -4.341827                  True                10   \n",
       "11      -1.670913                  True                10   \n",
       "12      -4.573832                  True                10   \n",
       "13      -2.786916                  True                10   \n",
       "14      -2.573832                  True                10   \n",
       "15      -0.786916                  True                10   \n",
       "16      -3.688990                  True                10   \n",
       "17      -2.344495                  True                10   \n",
       "18      -1.688990                  True                10   \n",
       "19      -0.344495                  True                10   \n",
       "\n",
       "   params.param_bm.seed params.total_step params.param_bm.initial_state  \\\n",
       "0                     4               500                          -1.0   \n",
       "1                     4               500                          -1.0   \n",
       "2                     4               500                           1.0   \n",
       "3                     4               500                           1.0   \n",
       "4                     3               500                          -1.0   \n",
       "5                     3               500                          -1.0   \n",
       "6                     3               500                           1.0   \n",
       "7                     3               500                           1.0   \n",
       "8                     2               500                          -1.0   \n",
       "9                     2               500                          -1.0   \n",
       "10                    2               500                           1.0   \n",
       "11                    2               500                           1.0   \n",
       "12                    1               500                          -1.0   \n",
       "13                    1               500                          -1.0   \n",
       "14                    1               500                           1.0   \n",
       "15                    1               500                           1.0   \n",
       "16                    0               500                          -1.0   \n",
       "17                    0               500                          -1.0   \n",
       "18                    0               500                           1.0   \n",
       "19                    0               500                           1.0   \n",
       "\n",
       "   params.param_bm.sigma tags.mlflow.source.type tags.mlflow.source.name  \\\n",
       "0                    0.2                   LOCAL          simulation4.py   \n",
       "1                    0.1                   LOCAL          simulation4.py   \n",
       "2                    0.2                   LOCAL          simulation4.py   \n",
       "3                    0.1                   LOCAL          simulation4.py   \n",
       "4                    0.2                   LOCAL          simulation4.py   \n",
       "5                    0.1                   LOCAL          simulation4.py   \n",
       "6                    0.2                   LOCAL          simulation4.py   \n",
       "7                    0.1                   LOCAL          simulation4.py   \n",
       "8                    0.2                   LOCAL          simulation4.py   \n",
       "9                    0.1                   LOCAL          simulation4.py   \n",
       "10                   0.2                   LOCAL          simulation4.py   \n",
       "11                   0.1                   LOCAL          simulation4.py   \n",
       "12                   0.2                   LOCAL          simulation4.py   \n",
       "13                   0.1                   LOCAL          simulation4.py   \n",
       "14                   0.2                   LOCAL          simulation4.py   \n",
       "15                   0.1                   LOCAL          simulation4.py   \n",
       "16                   0.2                   LOCAL          simulation4.py   \n",
       "17                   0.1                   LOCAL          simulation4.py   \n",
       "18                   0.2                   LOCAL          simulation4.py   \n",
       "19                   0.1                   LOCAL          simulation4.py   \n",
       "\n",
       "   tags.mlflow.user             tags.mlflow.source.git.commit  \n",
       "0            vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "1            vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "2            vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "3            vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "4            vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "5            vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "6            vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "7            vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "8            vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "9            vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "10           vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "11           vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "12           vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "13           vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "14           vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "15           vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "16           vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "17           vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "18           vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  \n",
       "19           vscode  eab15a1ae74715fae02cc0618e06c456b7dd6b83  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_result"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d191b103-b974-4412-8208-5489573a2a9a",
   "metadata": {},
   "source": [
    "ただし時系列データをmetricとして記録させている場合は、このDataFrameには時系列全体の情報は入っておらず、最終時刻の値しか入っていない。時系列全体を取得する方法は後述。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ff5d1cc0-7328-436e-bf26-7c61f2ccf088",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "d9af8fb0-9061-4562-b8e8-50c5da0faeb8",
   "metadata": {},
   "source": [
    "### sigmaとinitial_stateでグループ分けして最終状態の統計量を取得"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b30c70ee-6e9a-4bf3-b5a1-cf11664bba4c",
   "metadata": {},
   "source": [
    "パラメータが数値ではなく文字列として格納されているので、数字に直しておくと便利"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "60144e05-1825-4986-be6b-b529bed10f9b",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_result[\"sigma\"] = df_result[\"params.param_bm.sigma\"].apply(float)\n",
    "df_result[\"x0\"] = df_result[\"params.param_bm.initial_state\"].apply(float)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "f7e46d8c-8bb2-41fa-a697-843e0dd75a5e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sigma</th>\n",
       "      <th>x0</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">0.1</th>\n",
       "      <th>-1.0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>-1.573548</td>\n",
       "      <td>2.126323</td>\n",
       "      <td>-3.670913</td>\n",
       "      <td>-2.786916</td>\n",
       "      <td>-2.344495</td>\n",
       "      <td>-0.815163</td>\n",
       "      <td>1.749746</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>0.426452</td>\n",
       "      <td>2.126323</td>\n",
       "      <td>-1.670913</td>\n",
       "      <td>-0.786916</td>\n",
       "      <td>-0.344495</td>\n",
       "      <td>1.184837</td>\n",
       "      <td>3.749746</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">0.2</th>\n",
       "      <th>-1.0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>-2.147097</td>\n",
       "      <td>4.252646</td>\n",
       "      <td>-6.341827</td>\n",
       "      <td>-4.573832</td>\n",
       "      <td>-3.688990</td>\n",
       "      <td>-0.630327</td>\n",
       "      <td>4.499493</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>-0.147097</td>\n",
       "      <td>4.252646</td>\n",
       "      <td>-4.341827</td>\n",
       "      <td>-2.573832</td>\n",
       "      <td>-1.688990</td>\n",
       "      <td>1.369673</td>\n",
       "      <td>6.499493</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            count      mean       std       min       25%       50%       75%  \\\n",
       "sigma x0                                                                        \n",
       "0.1   -1.0    5.0 -1.573548  2.126323 -3.670913 -2.786916 -2.344495 -0.815163   \n",
       "       1.0    5.0  0.426452  2.126323 -1.670913 -0.786916 -0.344495  1.184837   \n",
       "0.2   -1.0    5.0 -2.147097  4.252646 -6.341827 -4.573832 -3.688990 -0.630327   \n",
       "       1.0    5.0 -0.147097  4.252646 -4.341827 -2.573832 -1.688990  1.369673   \n",
       "\n",
       "                 max  \n",
       "sigma x0              \n",
       "0.1   -1.0  1.749746  \n",
       "       1.0  3.749746  \n",
       "0.2   -1.0  4.499493  \n",
       "       1.0  6.499493  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_result.groupby([\"sigma\", \"x0\"])[\"metrics.state\"].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "ee236605-d737-4ed7-b9e8-84e9860c8443",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>x0</th>\n",
       "      <th>-1.0</th>\n",
       "      <th>1.0</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sigma</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0.1</th>\n",
       "      <td>-1.573548</td>\n",
       "      <td>0.426452</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.2</th>\n",
       "      <td>-2.147097</td>\n",
       "      <td>-0.147097</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "x0         -1.0       1.0\n",
       "sigma                    \n",
       "0.1   -1.573548  0.426452\n",
       "0.2   -2.147097 -0.147097"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_result.pivot_table(index=\"sigma\", columns=\"x0\", values=\"metrics.state\", aggfunc=np.mean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ec0c135f-9a03-48cb-aa18-9011096a7e99",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='x0', ylabel='sigma'>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.heatmap(df_result.pivot_table(\n",
    "    index=\"sigma\", columns=\"x0\", values=\"metrics.state\",\n",
    "    aggfunc=np.mean\n",
    "))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c1389251-e1e8-4f96-b4e2-a5a5c207fd00",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "a8499013-94a5-469b-a44f-ac7e0e83976d",
   "metadata": {},
   "source": [
    "### 時系列全体を取得する\n",
    "`run_id`を指定してget_metric_histroy()を呼ぶ必要がある。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "911f6096-a722-41cf-8e13-910f6df28982",
   "metadata": {},
   "outputs": [],
   "source": [
    "history = mlflow_client.get_metric_history(df_result.loc[0, \"run_id\"], \"state\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "5b7f9718-df96-46ba-a8c7-468d3ff07941",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<Metric: key='state', step=0, timestamp=1634285781487, value=-1.0>,\n",
       " <Metric: key='state', step=9, timestamp=1634285781502, value=-1.3982970775032995>,\n",
       " <Metric: key='state', step=19, timestamp=1634285781517, value=-1.123613722671141>,\n",
       " <Metric: key='state', step=29, timestamp=1634285781524, value=-1.852119143219679>,\n",
       " <Metric: key='state', step=39, timestamp=1634285781535, value=-1.9458795783072127>,\n",
       " <Metric: key='state', step=49, timestamp=1634285781545, value=-1.9294883382857977>,\n",
       " <Metric: key='state', step=59, timestamp=1634285781554, value=-2.02788229242719>,\n",
       " <Metric: key='state', step=69, timestamp=1634285781564, value=-1.7838786995375902>,\n",
       " <Metric: key='state', step=79, timestamp=1634285781574, value=-2.053931820406514>,\n",
       " <Metric: key='state', step=89, timestamp=1634285781589, value=-2.3679593624603448>,\n",
       " <Metric: key='state', step=99, timestamp=1634285781603, value=-2.2660534863895614>,\n",
       " <Metric: key='state', step=109, timestamp=1634285781611, value=-1.7231916355977204>,\n",
       " <Metric: key='state', step=119, timestamp=1634285781619, value=-1.7765696876822925>,\n",
       " <Metric: key='state', step=129, timestamp=1634285781627, value=0.08774785612724555>,\n",
       " <Metric: key='state', step=139, timestamp=1634285781636, value=-0.02478633549975645>,\n",
       " <Metric: key='state', step=149, timestamp=1634285781677, value=0.08850939126386698>,\n",
       " <Metric: key='state', step=159, timestamp=1634285781695, value=0.0614162311349771>,\n",
       " <Metric: key='state', step=169, timestamp=1634285781713, value=0.13722125357138487>,\n",
       " <Metric: key='state', step=179, timestamp=1634285781722, value=0.9204246178819764>,\n",
       " <Metric: key='state', step=189, timestamp=1634285781732, value=0.9692840217246916>,\n",
       " <Metric: key='state', step=199, timestamp=1634285781746, value=0.916429641206412>,\n",
       " <Metric: key='state', step=209, timestamp=1634285781757, value=0.11793162118213996>,\n",
       " <Metric: key='state', step=219, timestamp=1634285781777, value=0.957317216280837>,\n",
       " <Metric: key='state', step=229, timestamp=1634285781786, value=1.431614246659716>,\n",
       " <Metric: key='state', step=239, timestamp=1634285781793, value=1.048711917941299>,\n",
       " <Metric: key='state', step=249, timestamp=1634285781809, value=1.822324586594352>,\n",
       " <Metric: key='state', step=259, timestamp=1634285781818, value=3.3243348693080876>,\n",
       " <Metric: key='state', step=269, timestamp=1634285781827, value=3.739692898972357>,\n",
       " <Metric: key='state', step=279, timestamp=1634285781837, value=3.739896874685517>,\n",
       " <Metric: key='state', step=289, timestamp=1634285781846, value=3.884492277708427>,\n",
       " <Metric: key='state', step=299, timestamp=1634285781852, value=5.119437619363439>,\n",
       " <Metric: key='state', step=309, timestamp=1634285781861, value=5.019322487474905>,\n",
       " <Metric: key='state', step=319, timestamp=1634285781874, value=4.101147579904474>,\n",
       " <Metric: key='state', step=329, timestamp=1634285781885, value=4.398865391041291>,\n",
       " <Metric: key='state', step=339, timestamp=1634285781894, value=4.142346713240965>,\n",
       " <Metric: key='state', step=349, timestamp=1634285781900, value=5.491616862460497>,\n",
       " <Metric: key='state', step=359, timestamp=1634285781920, value=5.256801994228996>,\n",
       " <Metric: key='state', step=369, timestamp=1634285781928, value=5.237316322748587>,\n",
       " <Metric: key='state', step=379, timestamp=1634285781936, value=3.5117452514745193>,\n",
       " <Metric: key='state', step=389, timestamp=1634285781946, value=1.8122130468323014>,\n",
       " <Metric: key='state', step=399, timestamp=1634285781955, value=1.7308211516186887>,\n",
       " <Metric: key='state', step=409, timestamp=1634285781963, value=1.303489335137691>,\n",
       " <Metric: key='state', step=419, timestamp=1634285781972, value=1.6698361013130183>,\n",
       " <Metric: key='state', step=429, timestamp=1634285781988, value=0.9875629765134888>,\n",
       " <Metric: key='state', step=439, timestamp=1634285781997, value=1.2993181260863327>,\n",
       " <Metric: key='state', step=449, timestamp=1634285782005, value=-0.32478878135046657>,\n",
       " <Metric: key='state', step=459, timestamp=1634285782017, value=-1.6977631268014806>,\n",
       " <Metric: key='state', step=469, timestamp=1634285782030, value=-1.2171847979791337>,\n",
       " <Metric: key='state', step=479, timestamp=1634285782039, value=0.06829385226924223>,\n",
       " <Metric: key='state', step=489, timestamp=1634285782051, value=-0.27906732149917024>,\n",
       " <Metric: key='state', step=499, timestamp=1634285782059, value=-0.6303268686666377>]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "37b7335e-f1b6-49cf-9319-19df7d239324",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "fa0a9d05-cb33-40b3-9580-23f4558ea20b",
   "metadata": {},
   "source": [
    "## 実験結果の取得 (Simulatorクラスを用いる場合)\n",
    "Simulatorクラスにget_metric_histroy()とget_state_trajectory()を実装している。これらを用いれば簡便に時系列情報を取得できる"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8afc8f99-a471-468f-92f1-13ab9412a2d8",
   "metadata": {},
   "source": [
    "### mlflowでの時系列の記録の二通りの方法\n",
    "* `metric_history`は`record_per`(=10)ステップおきにmlflowではmetricとして記録したもの\n",
    "\n",
    "* `state_trajectory`は全てのステップの情報を格納したnumpy配列をmlflowのartifactとして記録したもの\n",
    "\n",
    "metricとして記録するとmlflowのwebインターフェースから見れるが、あまりデータ点が多いと読み込むのが大変になるので間引いたものを用いている。artifactはただのバイナリファイルとして保存されるので、webインターフェースからは中身を見れないがデータが大きくても問題ない。\n",
    "\n",
    "参考: https://mlflow.org/docs/latest/tracking.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "788249c0-a1a6-4563-b3f2-d4424366dcaa",
   "metadata": {},
   "outputs": [],
   "source": [
    "sim = get_simulator(seed=0, x0=x0s[0], sigma=sigmas[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "5b100e5b-6b65-4495-a4f5-9ed9d9fbd3ef",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'run_id': '1c9f9b2e8afa4dd4a1d568a922218eaa',\n",
       " 'experiment_id': '4',\n",
       " 'status': 'FINISHED',\n",
       " 'artifact_uri': '/workspaces/mlflow-demo/mlruns/4/1c9f9b2e8afa4dd4a1d568a922218eaa/artifacts',\n",
       " 'start_time': Timestamp('2021-10-15 08:15:57.577000+0000', tz='UTC'),\n",
       " 'end_time': Timestamp('2021-10-15 08:15:58.599000+0000', tz='UTC'),\n",
       " 'metrics': {'state': -0.34449492590095615},\n",
       " 'params': {'save_full_traj': 'True',\n",
       "  'record_per': '10',\n",
       "  'param_bm': {'seed': '0', 'initial_state': '1.0', 'sigma': '0.1'},\n",
       "  'total_step': '500'},\n",
       " 'tags': {'mlflow': {'source': {'type': 'LOCAL',\n",
       "    'name': 'simulation4.py',\n",
       "    'git': {'commit': 'eab15a1ae74715fae02cc0618e06c456b7dd6b83'}},\n",
       "   'user': 'vscode'}}}"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sim.result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "9ee4e8a6-83a1-4aac-8c07-a8db5d71169f",
   "metadata": {},
   "outputs": [],
   "source": [
    "metric_history = sim.get_metric_history()\n",
    "state_trajectory = sim.get_state_trajectory()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "992e7a3f-0575-4039-b1e2-9368ea14ba2d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f1de1da8e80>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(state_trajectory, label=\"state_trajectory\")\n",
    "plt.plot([metric.step for metric in metric_history],\n",
    "         [metric.value for metric in metric_history], label=\"metric_histroy\")\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "50b8d571-98bb-4bd3-8855-b25b4e5a78a4",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}