notebooks/Scaling-PPO-in-AC-Environment.ipynb

{
    "cells": [
        {
            "cell_type": "code",
            "execution_count": 2,
            "metadata": {},
            "outputs": [],
            "source": [
                "import wandb\n",
                "import numpy as np\n",
                "import plotly.graph_objects as go\n",
                "import plotly.io as pio\n",
                "pio.renderers.default = \"svg\"\n",
                "\n",
                "api = wandb.Api()\n",
                "runs = api.runs('shehper/AC-Solver-PPO')"
            ]
        },
        {
            "cell_type": "markdown",
            "metadata": {},
            "source": [
                "### Performance vs Interactions"
            ]
        },
        {
            "cell_type": "code",
            "execution_count": 3,
            "metadata": {},
            "outputs": [],
            "source": [
                "class Performance_vs_Interactions:\n",
                "    \"\"\"\n",
                "    A class to obtain performance metrics (number of solved examples or mean return) as functions of total number of environment interactions \n",
                "    across various experiments.\n",
                "\n",
                "    Attributes:\n",
                "        states_type (str): Specifies the type of states to include, either \"solved\" or \"all\". This pertains to the initial state distribution of AC Environment.\n",
                "        mean_returns_data (dict): Stores the mean returns data of runs. Keys are ids of runs, and values are mean returns. \n",
                "        num_solved_data (dict): Stores the number of solved instances data of runs.\n",
                "        optimal_lr_data (dict): Stores the optimal learning rates data of runs. It saves optimal learning rate as a function of number of environment interactions.\n",
                "\n",
                "    Methods:\n",
                "        relevant_runs: Returns a list of runs that meet the criteria relevant for this analysis.\n",
                "        _selection_criteria(run): Evaluates a run to determine if it meets all the specified conditions.\n",
                "        get_num_solved_data: Computes and returns the average number of solved instances for each run configuration.\n",
                "        get_mean_returns_data: Computes and returns the average normalized returns for each run configuration.\n",
                "        get_data(metric): Returns the aggregated data based on the specified metric ('num_solved' or 'mean_returns').\n",
                "        get_optimal_lr(metric): Computes and returns the optimal learning rate as a function of total number of environment interactions\n",
                "                                for maximizing the specified metric.\n",
                "        get_perf_vs_E(metric): Returns a tuple of sorted environment steps and the corresponding performance data.\n",
                "    \"\"\"\n",
                "    def __init__(self, states_type):\n",
                "        self.states_type=states_type\n",
                "        self.mean_returns_data = None\n",
                "        self.num_solved_data = None\n",
                "        self.optimal_lr_data = None\n",
                "    \n",
                "    @property\n",
                "    def relevant_runs(self):\n",
                "        relevant_runs = [run for run in runs if self._selection_criteria(run)]\n",
                "        print(f\"{len(relevant_runs)} found for the task!\")\n",
                "        return relevant_runs\n",
                "    \n",
                "    def _selection_criteria(self, run):\n",
                "        cfg = run.config\n",
                "        state = run.state\n",
                "        allowed_states = (\"finished\") # could also be (\"finished\", \"running\")\n",
                "        if all([cfg[\"states_type\"] == self.states_type,\n",
                "                cfg[\"anneal_lr\"] == False,\n",
                "                cfg[\"max_env_steps\"] == 200,\n",
                "                cfg[\"num_steps\"] == 200,\n",
                "                cfg[\"nodes_counts\"] == [512, 512],\n",
                "                cfg[\"total_timesteps\"] <= 8e7,\n",
                "                state in allowed_states]):\n",
                "            return True \n",
                "        else:\n",
                "            return False\n",
                "        \n",
                "    def get_num_solved_data(self):\n",
                "        self.num_solved_data = {}\n",
                "\n",
                "        for run in self.relevant_runs:\n",
                "            cfg = run.config\n",
                "            T = cfg[\"total_timesteps\"]\n",
                "            lr = cfg[\"learning_rate\"]\n",
                "            n_solved = run.summary[\"charts/solved\"]\n",
                "\n",
                "            if (T, lr) not in self.num_solved_data.keys():\n",
                "                self.num_solved_data[(T, lr)] = []\n",
                "            self.num_solved_data[(T, lr)].append(n_solved)\n",
                "\n",
                "        for k, v in self.num_solved_data.items():\n",
                "            self.num_solved_data[k] = np.mean(v)\n",
                "\n",
                "        return self.num_solved_data\n",
                "    \n",
                "    def get_mean_returns_data(self):\n",
                "        self.mean_returns_data = {}\n",
                "\n",
                "        for run in self.relevant_runs:\n",
                "            cfg = run.config\n",
                "            T = cfg[\"total_timesteps\"]\n",
                "            lr = cfg[\"learning_rate\"]\n",
                "            returns = run.summary[\"charts/normalized_returns_mean\"]\n",
                "\n",
                "            if (T, lr) not in self.mean_returns_data.keys():\n",
                "                self.mean_returns_data[(T, lr)] = []\n",
                "            self.mean_returns_data[(T, lr)].append(returns)\n",
                "\n",
                "        for k, v in self.mean_returns_data.items():\n",
                "            self.mean_returns_data[k] = np.mean(v)\n",
                "\n",
                "        return self.mean_returns_data\n",
                "\n",
                "    def get_data(self, metric=\"num_solved\"):\n",
                "        if metric == \"num_solved\":\n",
                "            return self.get_num_solved_data()\n",
                "        elif metric == \"mean_returns\":\n",
                "            return self.get_mean_returns_data()\n",
                "        else:\n",
                "            raise(\"metric must be num_solved or mean_returns\")\n",
                "\n",
                "    def get_optimal_lr(self, metric=\"num_solved\"):\n",
                "\n",
                "        self.optimal_lr_data = {}\n",
                "\n",
                "        if metric == \"num_solved\":\n",
                "            data = self.num_solved_data\n",
                "        elif metric == \"mean_returns\":\n",
                "            data = self.mean_returns_data\n",
                "\n",
                "        if data is None:\n",
                "            data = self.get_data(metric=metric)\n",
                "\n",
                "        for (T, lr), num_solved in data.items():\n",
                "            if T not in self.optimal_lr_data.keys():\n",
                "                self.optimal_lr_data[T] = (lr, num_solved)\n",
                "            else:\n",
                "                _, previous_best_num_solved = self.optimal_lr_data[T]\n",
                "                if num_solved > previous_best_num_solved:\n",
                "                    self.optimal_lr_data[T] = (lr, num_solved)\n",
                "\n",
                "        return self.optimal_lr_data\n",
                "\n",
                "    def get_perf_vs_E(self, metric=\"num_solved\"):\n",
                "        if self.optimal_lr_data is None:\n",
                "            self.optimal_lr_data = self.get_optimal_lr(metric=metric)\n",
                "\n",
                "        Es = list(self.optimal_lr_data.keys())\n",
                "        Es.sort()\n",
                "        perfs = list(self.optimal_lr_data[k][1] for k in Es)\n",
                "\n",
                "        return (Es, perfs)"
            ]
        },
        {
            "cell_type": "code",
            "execution_count": 4,
            "metadata": {},
            "outputs": [
                {
                    "name": "stdout",
                    "output_type": "stream",
                    "text": [
                        "123 found for the task!\n",
                        "142 found for the task!\n"
                    ]
                },
                {
                    "data": {
                        "image/svg+xml": [
                            "<svg class=\"main-svg\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\" width=\"1200\" height=\"900\" style=\"\" viewBox=\"0 0 1200 900\"><rect x=\"0\" y=\"0\" width=\"1200\" height=\"900\" style=\"fill: rgb(255, 255, 255); fill-opacity: 1;\"/><defs id=\"defs-813bdb\"><g class=\"clips\"><clipPath id=\"clip813bdbxyplot\" class=\"plotclip\"><rect width=\"900\" height=\"720\"/></clipPath><clipPath class=\"axesclip\" id=\"clip813bdbx\"><rect x=\"84\" y=\"0\" width=\"900\" height=\"900\"/></clipPath><clipPath class=\"axesclip\" id=\"clip813bdby\"><rect x=\"0\" y=\"100\" width=\"1200\" height=\"720\"/></clipPath><clipPath class=\"axesclip\" id=\"clip813bdbxy\"><rect x=\"84\" y=\"100\" width=\"900\" height=\"720\"/></clipPath></g><g class=\"gradients\"/><g class=\"patterns\"/></defs><g class=\"bglayer\"><rect class=\"bg\" x=\"84\" y=\"100\" width=\"900\" height=\"720\" style=\"fill: rgb(229, 236, 246); fill-opacity: 1; stroke-width: 0;\"/></g><g class=\"layer-below\"><g class=\"imagelayer\"/><g class=\"shapelayer\"/></g><g class=\"cartesianlayer\"><g class=\"subplot xy\"><g class=\"layer-subplot\"><g class=\"shapelayer\"/><g class=\"imagelayer\"/></g><g class=\"minor-gridlayer\"><g class=\"x\"/><g class=\"y\"/></g><g class=\"gridlayer\"><g class=\"x\"><path class=\"xgrid crisp\" transform=\"translate(91.88,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(113.45,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(132.75,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(259.69,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(333.94,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(386.63,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(427.49,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(460.88,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(489.11,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(513.5699999999999,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(535.14,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(554.4300000000001,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(681.37,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(755.63,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(808.31,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(849.18,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(882.57,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(910.8,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(935.25,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(956.82,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(976.12,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/></g><g class=\"y\"><path class=\"ygrid crisp\" transform=\"translate(0,763.43)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,693.29)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,638.89)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,594.44)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,556.85)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,524.3)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,495.58)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,469.89)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,300.9)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,202.04000000000002)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,131.9)\" d=\"M84,0h900\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/></g></g><g class=\"zerolinelayer\"/><path class=\"xlines-below\"/><path class=\"ylines-below\"/><g class=\"overlines-below\"/><g class=\"xaxislayer-below\"/><g class=\"yaxislayer-below\"/><g class=\"overaxes-below\"/><g class=\"plot\" transform=\"translate(84,100)\" clip-path=\"url(#clip813bdbxyplot)\"><g class=\"scatterlayer mlayer\"><g class=\"trace scatter tracea5268d\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M48.75,599.46L216.55,448.86L343.49,341.54L470.43,198.88L724.31,54.23L851.25,39.75\" style=\"vector-effect: none; fill: none; stroke: rgb(99, 110, 250); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(48.75,599.46)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(216.55,448.86)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(343.49,341.54)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(470.43,198.88)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(724.31,54.23)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(851.25,39.75)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/></g><g class=\"text\"/></g><g class=\"trace scatter trace3ac276\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M48.75,680.25L216.55,398.3L343.49,375.65L470.43,314.19L724.31,83.53L851.25,74.59\" style=\"vector-effect: none; fill: none; stroke: rgb(239, 85, 59); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(48.75,680.25)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(216.55,398.3)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(343.49,375.65)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(470.43,314.19)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(724.31,83.53)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(851.25,74.59)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/></g><g class=\"text\"/></g></g></g><g class=\"overplot\"/><path class=\"xlines-above crisp\" d=\"M0,0\" style=\"fill: none;\"/><path class=\"ylines-above crisp\" d=\"M0,0\" style=\"fill: none;\"/><g class=\"overlines-above\"/><g class=\"xaxislayer-above\"><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" transform=\"translate(91.88,0)\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\">8</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(113.45,0)\">9</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"4.5\" y=\"848\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(132.75,0)\">1M</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(259.69,0)\">2</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(333.94,0)\">3</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(386.63,0)\">4</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(427.49,0)\">5</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(460.88,0)\">6</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(489.11,0)\">7</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(513.5699999999999,0)\">8</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(535.14,0)\">9</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"4.5\" y=\"848\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(554.4300000000001,0)\">10M</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(681.37,0)\">2</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(755.63,0)\">3</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(808.31,0)\">4</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(849.18,0)\">5</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(882.57,0)\">6</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(910.8,0)\">7</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(935.25,0)\">8</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"834.5\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(956.82,0)\">9</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"4.5\" y=\"848\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(976.12,0)\">100M</text></g></g><g class=\"yaxislayer-above\"><g class=\"ytick\"><text text-anchor=\"end\" x=\"83\" y=\"4.725\" transform=\"translate(0,763.43)\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\">3</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"83\" y=\"4.725\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,693.29)\">4</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"83\" y=\"4.725\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,638.89)\">5</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"83\" y=\"4.725\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,594.44)\">6</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"83\" y=\"4.725\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,556.85)\">7</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"83\" y=\"4.725\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,524.3)\">8</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"83\" y=\"4.725\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,495.58)\">9</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"78.5\" y=\"3.3\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,469.89)\">100</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"83\" y=\"4.725\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,300.9)\">2</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"83\" y=\"4.725\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,202.04000000000002)\">3</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"83\" y=\"4.725\" style=\"font-family: Helvetica; font-size: 13.5px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,131.9)\">4</text></g></g><g class=\"overaxes-above\"/></g></g><g class=\"polarlayer\"/><g class=\"smithlayer\"/><g class=\"ternarylayer\"/><g class=\"geolayer\"/><g class=\"funnelarealayer\"/><g class=\"pielayer\"/><g class=\"iciclelayer\"/><g class=\"treemaplayer\"/><g class=\"sunburstlayer\"/><g class=\"glimages\"/><defs id=\"topdefs-813bdb\"><g class=\"clips\"/><clipPath id=\"legend813bdb\"><rect width=\"186\" height=\"94\" x=\"0\" y=\"0\"/></clipPath></defs><g class=\"layer-above\"><g class=\"imagelayer\"/><g class=\"shapelayer\"/></g><g class=\"infolayer\"><g class=\"legend\" pointer-events=\"all\" transform=\"translate(1002,100)\"><rect class=\"bg\" shape-rendering=\"crispEdges\" style=\"stroke: rgb(68, 68, 68); stroke-opacity: 1; fill: rgb(255, 255, 255); fill-opacity: 1; stroke-width: 0px;\" width=\"186\" height=\"94\" x=\"0\" y=\"0\"/><g class=\"scrollbox\" transform=\"\" clip-path=\"url(#legend813bdb)\"><g class=\"groups\"><g class=\"traces\" transform=\"translate(0,26)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"11.7\" style=\"font-family: Arial; font-size: 30px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">GS-solved</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(99, 110, 250); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-21\" width=\"180.0625\" height=\"42\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g><g class=\"traces\" transform=\"translate(0,68)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"11.7\" style=\"font-family: Arial; font-size: 30px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">All</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(239, 85, 59); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M3,0A3,3 0 1,1 0,-3A3,3 0 0,1 3,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-21\" width=\"180.0625\" height=\"42\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g></g></g><rect class=\"scrollbar\" rx=\"20\" ry=\"3\" width=\"0\" height=\"0\" style=\"fill: rgb(128, 139, 164); fill-opacity: 1;\" x=\"0\" y=\"0\"/></g><g class=\"g-gtitle\"><text class=\"gtitle\" x=\"60\" y=\"50\" text-anchor=\"start\" dy=\"0em\" style=\"font-family: Arial; font-size: 40px; fill: rgb(0, 0, 0); opacity: 1; font-weight: normal; white-space: pre;\">Scaling of performance with environment interactions</text></g><g class=\"g-xtitle\"><text class=\"xtitle\" x=\"534\" y=\"887.5\" text-anchor=\"middle\" style=\"font-family: Helvetica; font-size: 30px; fill: rgb(0, 0, 0); opacity: 1; font-weight: normal; white-space: pre;\">Number of environment interactions</text></g><g class=\"g-ytitle\" transform=\"translate(14.546875,0)\"><text class=\"ytitle\" transform=\"rotate(-90,12.953125,460)\" x=\"12.953125\" y=\"460\" text-anchor=\"middle\" style=\"font-family: Helvetica; font-size: 30px; fill: rgb(0, 0, 0); opacity: 1; font-weight: normal; white-space: pre;\">Number of presentations solved by PPO</text></g></g></svg>"
                        ]
                    },
                    "metadata": {},
                    "output_type": "display_data"
                }
            ],
            "source": [
                "states_types = (\"solved\", \"all\")\n",
                "\n",
                "data_perf_vs_E = {}\n",
                "for state_type in states_types:\n",
                "    initate_class = Performance_vs_Interactions(states_type=state_type)\n",
                "    data_perf_vs_E[state_type] = initate_class.get_perf_vs_E()\n",
                "\n",
                "#### ----------- #####\n",
                "\n",
                "fig = go.Figure()\n",
                "\n",
                "def get_legend(state_type):\n",
                "    return \"GS-solved\" if state_type == \"solved\" else \"All\"\n",
                "\n",
                "for state_type in states_types:\n",
                "    x = data_perf_vs_E[state_type][0]\n",
                "    y = data_perf_vs_E[state_type][1]\n",
                "    fig.add_trace(go.Scatter(\n",
                "            x=x, y=y, mode='markers+lines', name=f\"{get_legend(state_type)}\"\n",
                "        ))\n",
                "    \n",
                "fig.update_layout(\n",
                "            title='Scaling of performance with environment interactions',\n",
                "            title_font=dict(size=40, family='Arial', color='black'),\n",
                "            xaxis=dict(type='log', \n",
                "                       title='Number of environment interactions',\n",
                "                       title_font=dict(size=30, family='Helvetica', color='black'),\n",
                "                       tickfont=dict(size=18, family='Helvetica', color='black')),\n",
                "            yaxis=dict(type='log', \n",
                "                       title='Number of presentations solved by PPO',\n",
                "                       title_font=dict(size=30, family='Helvetica', color='black'),\n",
                "                       tickfont=dict(size=18, family='Helvetica', color='black')),\n",
                "            legend=dict(\n",
                "            font=dict(size=30, family='Arial', color='black'),\n",
                "            ),\n",
                "            width=1200,\n",
                "            height=900\n",
                "        )\n",
                "# Save to PDF\n",
                "# fig.write_image(\"images/scaling_env.pdf\")\n",
                "fig.show()"
            ]
        },
        {
            "cell_type": "markdown",
            "metadata": {},
            "source": [
                "### Horizon Length vs Interactions"
            ]
        },
        {
            "cell_type": "code",
            "execution_count": 5,
            "metadata": {},
            "outputs": [],
            "source": [
                "class Horizon_vs_Interactions:\n",
                "    \"\"\" \n",
                "    A class for analyzing the relationship between training horizon (timesteps) and interactions.\n",
                "\n",
                "    Attributes:\n",
                "    -----------\n",
                "    states_type (str): The type of states used in the runs. This is a key parameter for filtering relevant runs.\n",
                "    \n",
                "    timesteps (float): The total number of environment interactions for each run. This must be either 4e7 or 1e7. \n",
                "        An assertion is made to ensure the correct value is used.\n",
                "    \n",
                "    Methods:\n",
                "    --------\n",
                "    relevant_runs():\n",
                "        Retrieves the runs relevant to this experimental setup. Requires us to specify states_type and timesteps.\n",
                "    \n",
                "    _selection_criteria(run):\n",
                "        Determines if a given run meets the criteria for inclusion based on its configuration and state.\n",
                "    \"\"\"\n",
                "\n",
                "    def __init__(self, states_type, timesteps):\n",
                "        self.states_type=states_type\n",
                "        self.timesteps=timesteps\n",
                "        assert self.timesteps in [4e7, 1e7], f\"expect timesteps to be 4e7 or 1e7, found {timesteps}\"\n",
                "    \n",
                "    @property\n",
                "    def relevant_runs(self):\n",
                "        \"\"\"Get the runs that are relevant for this class. \n",
                "        Returns dict: \n",
                "        key: (total_timesteps, learning_rate)\n",
                "        value: a list of 3 runs with different seeds\n",
                "        \"\"\"\n",
                "        relevant_runs = {}\n",
                "        total_relevant_runs = 0\n",
                "        for run in runs:\n",
                "            if self._selection_criteria(run):\n",
                "                total_relevant_runs += 1\n",
                "                max_env_steps = run.config[\"max_env_steps\"]\n",
                "                learning_rate = run.config[\"learning_rate\"]\n",
                "\n",
                "                if (max_env_steps, learning_rate) not in relevant_runs.keys():\n",
                "                    relevant_runs[(max_env_steps, learning_rate)] = []\n",
                "\n",
                "                relevant_runs[(max_env_steps, learning_rate)].append(run)\n",
                "\n",
                "        print(f\"{total_relevant_runs} runs found for the task!\")\n",
                "        return relevant_runs\n",
                "    \n",
                "    def _selection_criteria(self, run):\n",
                "        cfg = run.config\n",
                "        state = run.state\n",
                "        allowed_states = (\"finished\") # could also be (\"finished\", \"running\")\n",
                "        if all([cfg[\"num_steps\"] == 512,\n",
                "                cfg[\"states_type\"] == self.states_type,\n",
                "                cfg[\"anneal_lr\"] == False,\n",
                "                cfg[\"nodes_counts\"] == [512, 512],\n",
                "                cfg[\"total_timesteps\"] == self.timesteps,\n",
                "                state in allowed_states]):\n",
                "            return True \n",
                "        else:\n",
                "            return False\n"
            ]
        },
        {
            "cell_type": "code",
            "execution_count": 6,
            "metadata": {},
            "outputs": [],
            "source": [
                "def get_timesteps_vs_performance(run_history, n_solved):\n",
                "    \"\"\"\n",
                "    Finds the first timestep at which the performance in a run history meets or exceeds specified thresholds.\n",
                "\n",
                "    Parameters:\n",
                "    -----------\n",
                "    run_history : A list or Numpy Array representing the performance metric recorded at each timestep during a run.\n",
                "        \n",
                "    n_solved : int or list of int\n",
                "        The performance threshold(s) to check for. Can be a single integer or a list of integers. \n",
                "        The function will return the first timestep where each threshold is met or exceeded.\n",
                "\n",
                "    Returns:\n",
                "    --------\n",
                "    dict\n",
                "        A dictionary where the keys are the thresholds from `n_solved`, and the values are the corresponding\n",
                "        timesteps at which each threshold was first met or exceeded.\n",
                "    \"\"\"\n",
                "\n",
                "    if isinstance(n_solved, int):\n",
                "        n_solved = [n_solved]    \n",
                "    \n",
                "    n_solved.sort()\n",
                "\n",
                "    curr = 0\n",
                "    out = {}\n",
                "\n",
                "    for t, t_data in enumerate(run_history):\n",
                "        if t_data >= n_solved[curr]:\n",
                "            out[n_solved[curr]] = t\n",
                "            curr += 1\n",
                "            if curr == len(n_solved):\n",
                "                break\n",
                "\n",
                "    assert curr == len(n_solved), \"curr has length less than n_solved\"\n",
                "\n",
                "    assert len(out) == len(n_solved)\n",
                "\n",
                "    return out"
            ]
        },
        {
            "cell_type": "code",
            "execution_count": 7,
            "metadata": {},
            "outputs": [],
            "source": [
                "def analyze_horizon_vs_interactions(states_type, n_solved, timesteps=4e7):\n",
                "    \"\"\"\n",
                "    Analyzes the relationship between training horizon (timesteps) and interactions in a reinforcement learning setup.\n",
                "\n",
                "    Parameters:\n",
                "    -----------\n",
                "    state_types (tuple or list of str): Types of states to analyze, e.g., (\"solved\", \"all\").\n",
                "    n_solved (list of int): Performance thresholds at which we evaluate timesteps.\n",
                "    timesteps (float, optional (default=4e7)): The total number of timesteps for the analysis. Must be either 4e7 or 1e7.\n",
                "\n",
                "    Returns:\n",
                "    --------\n",
                "    new_outs (dict):\n",
                "        A nested dictionary where the outer keys are performance thresholds (n_solved).\n",
                "        The values are dictionaries that map horizons to the first timestep at which each performance threshold is met.\n",
                "    \"\"\"\n",
                "\n",
                "    horizon_vs_interactions = Horizon_vs_Interactions(states_type=states_type,\n",
                "                    timesteps=4e7)\n",
                "    relevant_runs = horizon_vs_interactions.relevant_runs\n",
                "\n",
                "    horizons = set()\n",
                "    lrs = set()\n",
                "\n",
                "    for (horizon, lr) in relevant_runs.keys():\n",
                "        horizons.add(horizon)\n",
                "        lrs.add(lr)\n",
                "\n",
                "    optimal_lrs = {}\n",
                "    for horizon in horizons:\n",
                "        optimal_lrs[horizon] = 0 \n",
                "        prev_best_performance = -1\n",
                "        for lr in lrs:\n",
                "            \n",
                "            curr_runs = relevant_runs[(horizon, lr)]\n",
                "            assert len(curr_runs) == 3\n",
                "            mean_performance = np.mean([run.summary[\"charts/solved\"] for run in curr_runs])\n",
                "            if mean_performance > prev_best_performance:\n",
                "                prev_best_performance = mean_performance\n",
                "                optimal_lrs[horizon] = lr\n",
                "        \n",
                "    best_runs = {(horizon, lr): relevant_runs[(horizon, lr)] for (horizon, lr) in optimal_lrs.items()}\n",
                "    best_runs_mean_history = {(horizon, lr): [] for (horizon, lr) in best_runs.keys()}\n",
                "\n",
                "    for (horizon, lr) in best_runs_mean_history:\n",
                "        history = []\n",
                "        for run in relevant_runs[(horizon, lr)]:\n",
                "            run_history = [row[\"charts/solved\"]for row in run.scan_history()]\n",
                "            history.append(run_history)\n",
                "        \n",
                "        mean_history = np.array(history).mean(axis=0)\n",
                "\n",
                "        best_runs_mean_history[(horizon, lr)] = mean_history\n",
                "\n",
                "    outs = {}\n",
                "    for (horizon, lr) in best_runs_mean_history:\n",
                "        run_history = best_runs_mean_history[(horizon, lr)]\n",
                "\n",
                "        n_solved_to_steps_dict = get_timesteps_vs_performance(run_history, n_solved)\n",
                "        outs[horizon] = n_solved_to_steps_dict\n",
                "\n",
                "    # k: {k': v}\n",
                "    # for each k', we make a scatterplot of {k: v}\n",
                "    new_outs = {}\n",
                "    for (horizon, n_solved_to_steps_dict) in outs.items():\n",
                "        for n_solved, steps in n_solved_to_steps_dict.items():\n",
                "            if n_solved not in new_outs.keys():\n",
                "                new_outs[n_solved] = {}\n",
                "            new_outs[n_solved][horizon] = steps\n",
                "\n",
                "    return new_outs"
            ]
        },
        {
            "cell_type": "code",
            "execution_count": 8,
            "metadata": {},
            "outputs": [
                {
                    "name": "stdout",
                    "output_type": "stream",
                    "text": [
                        "90 runs found for the task!\n"
                    ]
                }
            ],
            "source": [
                "states_type = \"all\"  # Define the types of states to analyze\n",
                "n_solved = [16, 32, 64, 96, 128, 160, 192, 224, 256]  \n",
                "new_outs = analyze_horizon_vs_interactions(states_type, n_solved)"
            ]
        },
        {
            "cell_type": "code",
            "execution_count": 9,
            "metadata": {},
            "outputs": [
                {
                    "data": {
                        "image/svg+xml": [
                            "<svg class=\"main-svg\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\" width=\"1200\" height=\"900\" style=\"\" viewBox=\"0 0 1200 900\"><rect x=\"0\" y=\"0\" width=\"1200\" height=\"900\" style=\"fill: rgb(255, 255, 255); fill-opacity: 1;\"/><defs id=\"defs-9d774f\"><g class=\"clips\"><clipPath id=\"clip9d774fxyplot\" class=\"plotclip\"><rect width=\"870\" height=\"720\"/></clipPath><clipPath class=\"axesclip\" id=\"clip9d774fx\"><rect x=\"89\" y=\"0\" width=\"870\" height=\"900\"/></clipPath><clipPath class=\"axesclip\" id=\"clip9d774fy\"><rect x=\"0\" y=\"100\" width=\"1200\" height=\"720\"/></clipPath><clipPath class=\"axesclip\" id=\"clip9d774fxy\"><rect x=\"89\" y=\"100\" width=\"870\" height=\"720\"/></clipPath></g><g class=\"gradients\"/><g class=\"patterns\"/></defs><g class=\"bglayer\"><rect class=\"bg\" x=\"89\" y=\"100\" width=\"870\" height=\"720\" style=\"fill: rgb(229, 236, 246); fill-opacity: 1; stroke-width: 0;\"/></g><g class=\"layer-below\"><g class=\"imagelayer\"/><g class=\"shapelayer\"/></g><g class=\"cartesianlayer\"><g class=\"subplot xy\"><g class=\"layer-subplot\"><g class=\"shapelayer\"/><g class=\"imagelayer\"/></g><g class=\"minor-gridlayer\"><g class=\"x\"/><g class=\"y\"/></g><g class=\"gridlayer\"><g class=\"x\"><path class=\"xgrid crisp\" transform=\"translate(201.70999999999998,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(373.14,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(544.5699999999999,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(716,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"xgrid crisp\" transform=\"translate(887.43,0)\" d=\"M0,100v720\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/></g><g class=\"y\"><path class=\"ygrid crisp\" transform=\"translate(0,617.26)\" d=\"M89,0h870\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,452.76)\" d=\"M89,0h870\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,288.26)\" d=\"M89,0h870\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/><path class=\"ygrid crisp\" transform=\"translate(0,123.76)\" d=\"M89,0h870\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 1px;\"/></g></g><g class=\"zerolinelayer\"><path class=\"yzl zl crisp\" transform=\"translate(0,781.76)\" d=\"M89,0h870\" style=\"stroke: rgb(255, 255, 255); stroke-opacity: 1; stroke-width: 2px;\"/></g><path class=\"xlines-below\"/><path class=\"ylines-below\"/><g class=\"overlines-below\"/><g class=\"xaxislayer-below\"/><g class=\"yaxislayer-below\"/><g class=\"overaxes-below\"/><g class=\"plot\" transform=\"translate(89,100)\" clip-path=\"url(#clip9d774fxyplot)\"><g class=\"scatterlayer mlayer\"><g class=\"trace scatter trace53bd1e\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M51,676.5L160.71,672.22L380.14,664.98L599.57,651.5L819,642.28\" style=\"vector-effect: none; fill: none; stroke: rgb(99, 110, 250); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(51,676.5)\" d=\"M6,0A6,6 0 1,1 0,-6A6,6 0 0,1 6,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(160.71,672.22)\" d=\"M6,0A6,6 0 1,1 0,-6A6,6 0 0,1 6,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(380.14,664.98)\" d=\"M6,0A6,6 0 1,1 0,-6A6,6 0 0,1 6,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(599.57,651.5)\" d=\"M6,0A6,6 0 1,1 0,-6A6,6 0 0,1 6,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(819,642.28)\" d=\"M6,0A6,6 0 1,1 0,-6A6,6 0 0,1 6,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/></g><g class=\"text\"/></g><g class=\"trace scatter trace84dca0\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M51,674.2L160.71,669.59L380.14,651.17L599.57,638.66L819,624.85\" style=\"vector-effect: none; fill: none; stroke: rgb(239, 85, 59); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(51,674.2)\" d=\"M6,6H-6V-6H6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(160.71,669.59)\" d=\"M6,6H-6V-6H6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(380.14,651.17)\" d=\"M6,6H-6V-6H6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(599.57,638.66)\" d=\"M6,6H-6V-6H6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(819,624.85)\" d=\"M6,6H-6V-6H6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/></g><g class=\"text\"/></g><g class=\"trace scatter trace64a105\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M51,659.39L160.71,641.95L380.14,610.7L599.57,589.31L819,539.96\" style=\"vector-effect: none; fill: none; stroke: rgb(0, 204, 150); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(51,659.39)\" d=\"M7.8,0L0,7.8L-7.8,0L0,-7.8Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(0, 204, 150); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(160.71,641.95)\" d=\"M7.8,0L0,7.8L-7.8,0L0,-7.8Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(0, 204, 150); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(380.14,610.7)\" d=\"M7.8,0L0,7.8L-7.8,0L0,-7.8Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(0, 204, 150); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(599.57,589.31)\" d=\"M7.8,0L0,7.8L-7.8,0L0,-7.8Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(0, 204, 150); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(819,539.96)\" d=\"M7.8,0L0,7.8L-7.8,0L0,-7.8Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(0, 204, 150); fill-opacity: 1;\"/></g><g class=\"text\"/></g><g class=\"trace scatter trace84b577\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M51,599.18L160.71,577.14L380.14,506.41L599.57,466.93L819,391.58\" style=\"vector-effect: none; fill: none; stroke: rgb(171, 99, 250); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(51,599.18)\" d=\"M7.2,2.4H2.4V7.2H-2.4V2.4H-7.2V-2.4H-2.4V-7.2H2.4V-2.4H7.2Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(171, 99, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(160.71,577.14)\" d=\"M7.2,2.4H2.4V7.2H-2.4V2.4H-7.2V-2.4H-2.4V-7.2H2.4V-2.4H7.2Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(171, 99, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(380.14,506.41)\" d=\"M7.2,2.4H2.4V7.2H-2.4V2.4H-7.2V-2.4H-2.4V-7.2H2.4V-2.4H7.2Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(171, 99, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(599.57,466.93)\" d=\"M7.2,2.4H2.4V7.2H-2.4V2.4H-7.2V-2.4H-2.4V-7.2H2.4V-2.4H7.2Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(171, 99, 250); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(819,391.58)\" d=\"M7.2,2.4H2.4V7.2H-2.4V2.4H-7.2V-2.4H-2.4V-7.2H2.4V-2.4H7.2Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(171, 99, 250); fill-opacity: 1;\"/></g><g class=\"text\"/></g><g class=\"trace scatter traceff0c37\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M51,585.37L160.71,510.02L380.14,412.64L599.57,336.31L819,270.18\" style=\"vector-effect: none; fill: none; stroke: rgb(255, 161, 90); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(51,585.37)\" d=\"M0,3.39l3.39,3.39l3.39,-3.39l-3.39,-3.39l3.39,-3.39l-3.39,-3.39l-3.39,3.39l-3.39,-3.39l-3.39,3.39l3.39,3.39l-3.39,3.39l3.39,3.39Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 161, 90); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(160.71,510.02)\" d=\"M0,3.39l3.39,3.39l3.39,-3.39l-3.39,-3.39l3.39,-3.39l-3.39,-3.39l-3.39,3.39l-3.39,-3.39l-3.39,3.39l3.39,3.39l-3.39,3.39l3.39,3.39Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 161, 90); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(380.14,412.64)\" d=\"M0,3.39l3.39,3.39l3.39,-3.39l-3.39,-3.39l3.39,-3.39l-3.39,-3.39l-3.39,3.39l-3.39,-3.39l-3.39,3.39l3.39,3.39l-3.39,3.39l3.39,3.39Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 161, 90); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(599.57,336.31)\" d=\"M0,3.39l3.39,3.39l3.39,-3.39l-3.39,-3.39l3.39,-3.39l-3.39,-3.39l-3.39,3.39l-3.39,-3.39l-3.39,3.39l3.39,3.39l-3.39,3.39l3.39,3.39Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 161, 90); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(819,270.18)\" d=\"M0,3.39l3.39,3.39l3.39,-3.39l-3.39,-3.39l3.39,-3.39l-3.39,-3.39l-3.39,3.39l-3.39,-3.39l-3.39,3.39l3.39,3.39l-3.39,3.39l3.39,3.39Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 161, 90); fill-opacity: 1;\"/></g><g class=\"text\"/></g><g class=\"trace scatter traced5aabb\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M51,565.96L160.71,484.69L380.14,378.42L599.57,264.59L819,215.57\" style=\"vector-effect: none; fill: none; stroke: rgb(25, 211, 243); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(51,565.96)\" d=\"M-6.93,3H6.93L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(25, 211, 243); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(160.71,484.69)\" d=\"M-6.93,3H6.93L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(25, 211, 243); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(380.14,378.42)\" d=\"M-6.93,3H6.93L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(25, 211, 243); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(599.57,264.59)\" d=\"M-6.93,3H6.93L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(25, 211, 243); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(819,215.57)\" d=\"M-6.93,3H6.93L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(25, 211, 243); fill-opacity: 1;\"/></g><g class=\"text\"/></g><g class=\"trace scatter traced1821f\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M51,525.49L160.71,453.44L380.14,350.13L599.57,195.17L819,169.84\" style=\"vector-effect: none; fill: none; stroke: rgb(255, 102, 146); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(51,525.49)\" d=\"M-6.93,-3H6.93L0,6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 102, 146); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(160.71,453.44)\" d=\"M-6.93,-3H6.93L0,6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 102, 146); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(380.14,350.13)\" d=\"M-6.93,-3H6.93L0,6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 102, 146); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(599.57,195.17)\" d=\"M-6.93,-3H6.93L0,6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 102, 146); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(819,169.84)\" d=\"M-6.93,-3H6.93L0,6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 102, 146); fill-opacity: 1;\"/></g><g class=\"text\"/></g><g class=\"trace scatter trace9548c4\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M51,460.02L160.71,423.5L380.14,309.99L599.57,131.34L819,132.66\" style=\"vector-effect: none; fill: none; stroke: rgb(182, 232, 128); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(51,460.02)\" d=\"M5.71,-1.85L3.53,4.85H-3.53L-5.71,-1.85L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(182, 232, 128); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(160.71,423.5)\" d=\"M5.71,-1.85L3.53,4.85H-3.53L-5.71,-1.85L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(182, 232, 128); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(380.14,309.99)\" d=\"M5.71,-1.85L3.53,4.85H-3.53L-5.71,-1.85L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(182, 232, 128); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(599.57,131.34)\" d=\"M5.71,-1.85L3.53,4.85H-3.53L-5.71,-1.85L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(182, 232, 128); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(819,132.66)\" d=\"M5.71,-1.85L3.53,4.85H-3.53L-5.71,-1.85L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(182, 232, 128); fill-opacity: 1;\"/></g><g class=\"text\"/></g><g class=\"trace scatter traceb3f70b\" style=\"stroke-miterlimit: 2; opacity: 1;\"><g class=\"fills\"/><g class=\"errorbars\"/><g class=\"lines\"><path class=\"js-line\" d=\"M51,417.25L160.71,332.69L380.14,233.33L599.57,43.5L819,66.86\" style=\"vector-effect: none; fill: none; stroke: rgb(255, 151, 255); stroke-opacity: 1; stroke-width: 2px; opacity: 1;\"/></g><g class=\"points\"><path class=\"point\" transform=\"translate(51,417.25)\" d=\"M5.2,-3V3L0,6L-5.2,3V-3L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 151, 255); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(160.71,332.69)\" d=\"M5.2,-3V3L0,6L-5.2,3V-3L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 151, 255); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(380.14,233.33)\" d=\"M5.2,-3V3L0,6L-5.2,3V-3L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 151, 255); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(599.57,43.5)\" d=\"M5.2,-3V3L0,6L-5.2,3V-3L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 151, 255); fill-opacity: 1;\"/><path class=\"point\" transform=\"translate(819,66.86)\" d=\"M5.2,-3V3L0,6L-5.2,3V-3L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 151, 255); fill-opacity: 1;\"/></g><g class=\"text\"/></g></g></g><g class=\"overplot\"/><path class=\"xlines-above crisp\" d=\"M0,0\" style=\"fill: none;\"/><path class=\"ylines-above crisp\" d=\"M0,0\" style=\"fill: none;\"/><g class=\"overlines-above\"/><g class=\"xaxislayer-above\"><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"839\" transform=\"translate(201.70999999999998,0)\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\">50</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"839\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(373.14,0)\">100</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"839\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(544.5699999999999,0)\">150</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"839\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(716,0)\">200</text></g><g class=\"xtick\"><text text-anchor=\"middle\" x=\"0\" y=\"839\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(887.43,0)\">250</text></g></g><g class=\"yaxislayer-above\"><g class=\"ytick\"><text text-anchor=\"end\" x=\"88\" y=\"6.3\" transform=\"translate(0,781.76)\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\">0</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"88\" y=\"6.3\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,617.26)\">500</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"88\" y=\"6.3\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,452.76)\">1000</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"88\" y=\"6.3\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,288.26)\">1500</text></g><g class=\"ytick\"><text text-anchor=\"end\" x=\"88\" y=\"6.3\" style=\"font-family: Helvetica; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre; opacity: 1;\" transform=\"translate(0,123.76)\">2000</text></g></g><g class=\"overaxes-above\"/></g></g><g class=\"polarlayer\"/><g class=\"smithlayer\"/><g class=\"ternarylayer\"/><g class=\"geolayer\"/><g class=\"funnelarealayer\"/><g class=\"pielayer\"/><g class=\"iciclelayer\"/><g class=\"treemaplayer\"/><g class=\"sunburstlayer\"/><g class=\"glimages\"/><defs id=\"topdefs-9d774f\"><g class=\"clips\"/><clipPath id=\"legend9d774f\"><rect width=\"212\" height=\"277\" x=\"0\" y=\"0\"/></clipPath></defs><g class=\"layer-above\"><g class=\"imagelayer\"/><g class=\"shapelayer\"/></g><g class=\"infolayer\"><g class=\"legend\" pointer-events=\"all\" transform=\"translate(976.4,100)\"><rect class=\"bg\" shape-rendering=\"crispEdges\" style=\"stroke: rgb(68, 68, 68); stroke-opacity: 1; fill: rgb(255, 255, 255); fill-opacity: 1; stroke-width: 0px;\" width=\"212\" height=\"277\" x=\"0\" y=\"0\"/><g class=\"scrollbox\" transform=\"\" clip-path=\"url(#legend9d774f)\"><text class=\"legendtitletext\" text-anchor=\"start\" x=\"2\" y=\"28.6\" style=\"font-family: Arial; font-size: 22px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">Presentations Solved</text><g class=\"groups\"><g class=\"traces\" transform=\"translate(0,46.800000000000004)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"7.0200000000000005\" style=\"font-family: Arial; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">16</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(99, 110, 250); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M6,0A6,6 0 1,1 0,-6A6,6 0 0,1 6,0Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(99, 110, 250); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-13.200000000000001\" width=\"70.046875\" height=\"26.400000000000002\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g><g class=\"traces\" transform=\"translate(0,73.2)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"7.0200000000000005\" style=\"font-family: Arial; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">32</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(239, 85, 59); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M6,6H-6V-6H6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(239, 85, 59); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-13.200000000000001\" width=\"70.046875\" height=\"26.400000000000002\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g><g class=\"traces\" transform=\"translate(0,99.60000000000001)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"7.0200000000000005\" style=\"font-family: Arial; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">64</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(0, 204, 150); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M7.8,0L0,7.8L-7.8,0L0,-7.8Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(0, 204, 150); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-13.200000000000001\" width=\"70.046875\" height=\"26.400000000000002\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g><g class=\"traces\" transform=\"translate(0,126.00000000000001)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"7.0200000000000005\" style=\"font-family: Arial; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">96</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(171, 99, 250); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M7.2,2.4H2.4V7.2H-2.4V2.4H-7.2V-2.4H-2.4V-7.2H2.4V-2.4H7.2Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(171, 99, 250); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-13.200000000000001\" width=\"70.046875\" height=\"26.400000000000002\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g><g class=\"traces\" transform=\"translate(0,152.4)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"7.0200000000000005\" style=\"font-family: Arial; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">128</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(255, 161, 90); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M0,3.39l3.39,3.39l3.39,-3.39l-3.39,-3.39l3.39,-3.39l-3.39,-3.39l-3.39,3.39l-3.39,-3.39l-3.39,3.39l3.39,3.39l-3.39,3.39l3.39,3.39Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 161, 90); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-13.200000000000001\" width=\"70.046875\" height=\"26.400000000000002\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g><g class=\"traces\" transform=\"translate(0,178.79999999999998)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"7.0200000000000005\" style=\"font-family: Arial; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">160</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(25, 211, 243); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M-6.93,3H6.93L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(25, 211, 243); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-13.200000000000001\" width=\"70.046875\" height=\"26.400000000000002\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g><g class=\"traces\" transform=\"translate(0,205.2)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"7.0200000000000005\" style=\"font-family: Arial; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">192</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(255, 102, 146); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M-6.93,-3H6.93L0,6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 102, 146); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-13.200000000000001\" width=\"70.046875\" height=\"26.400000000000002\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g><g class=\"traces\" transform=\"translate(0,231.6)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"7.0200000000000005\" style=\"font-family: Arial; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">224</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(182, 232, 128); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M5.71,-1.85L3.53,4.85H-3.53L-5.71,-1.85L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(182, 232, 128); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-13.200000000000001\" width=\"70.046875\" height=\"26.400000000000002\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g><g class=\"traces\" transform=\"translate(0,258)\" style=\"opacity: 1;\"><text class=\"legendtext\" text-anchor=\"start\" x=\"40\" y=\"7.0200000000000005\" style=\"font-family: Arial; font-size: 18px; fill: rgb(0, 0, 0); fill-opacity: 1; white-space: pre;\">256</text><g class=\"layers\" style=\"opacity: 1;\"><g class=\"legendfill\"/><g class=\"legendlines\"><path class=\"js-line\" d=\"M5,0h30\" style=\"fill: none; stroke: rgb(255, 151, 255); stroke-opacity: 1; stroke-width: 2px;\"/></g><g class=\"legendsymbols\"><g class=\"legendpoints\"><path class=\"scatterpts\" transform=\"translate(20,0)\" d=\"M5.2,-3V3L0,6L-5.2,3V-3L0,-6Z\" style=\"opacity: 1; stroke-width: 0px; fill: rgb(255, 151, 255); fill-opacity: 1;\"/></g></g></g><rect class=\"legendtoggle\" x=\"0\" y=\"-13.200000000000001\" width=\"70.046875\" height=\"26.400000000000002\" style=\"fill: rgb(0, 0, 0); fill-opacity: 0;\"/></g></g></g><rect class=\"scrollbar\" rx=\"20\" ry=\"3\" width=\"0\" height=\"0\" style=\"fill: rgb(128, 139, 164); fill-opacity: 1;\" x=\"0\" y=\"0\"/></g><g class=\"g-gtitle\"><text class=\"gtitle\" x=\"60\" y=\"50\" text-anchor=\"start\" dy=\"0em\" style=\"font-family: Arial; font-size: 30px; fill: rgb(0, 0, 0); opacity: 1; font-weight: normal; white-space: pre;\">Scaling of environment interactions with horizon length at fixed performance levels</text></g><g class=\"g-xtitle\"><text class=\"xtitle\" x=\"524\" y=\"878.5\" text-anchor=\"middle\" style=\"font-family: Helvetica; font-size: 30px; fill: rgb(0, 0, 0); opacity: 1; font-weight: normal; white-space: pre;\">Horizon Length</text></g><g class=\"g-ytitle\" transform=\"translate(15.046875,0)\"><text class=\"ytitle\" transform=\"rotate(-90,12.453125,460)\" x=\"12.453125\" y=\"460\" text-anchor=\"middle\" style=\"font-family: Helvetica; font-size: 30px; fill: rgb(0, 0, 0); opacity: 1; font-weight: normal; white-space: pre;\">Number of Environment Interactions</text></g></g></svg>"
                        ]
                    },
                    "metadata": {},
                    "output_type": "display_data"
                }
            ],
            "source": [
                "import plotly.graph_objects as go\n",
                "\n",
                "# Create a scatter plot for each (k, v) pair in the dictionary using Plotly\n",
                "fig = go.Figure()\n",
                "\n",
                "marker_shapes = ['circle', 'square', 'diamond', 'cross', 'x', 'triangle-up', 'triangle-down', 'pentagon', 'hexagon']\n",
                "\n",
                "for idx, (k, v) in enumerate(new_outs.items()):\n",
                "    if k < 300:\n",
                "        x = list([h for h in v.keys() if h >= 32])\n",
                "        x.sort()\n",
                "        y = [v[kp] for kp in x]\n",
                "        fig.add_trace(go.Scatter(x=x, y=y, mode='lines+markers', name=f'{k}',\n",
                "                                marker=dict(size=12, symbol=marker_shapes[idx % len(marker_shapes)])))\n",
                "\n",
                "# Add legend and labels\n",
                "fig.update_layout(\n",
                "    title='Scaling of environment interactions with horizon length at fixed performance levels',\n",
                "    title_font=dict(size=30, family='Arial', color='black'),\n",
                "    xaxis=dict(#type='log', \n",
                "                title='Horizon Length',\n",
                "                title_font=dict(size=30, family='Helvetica', color='black'),\n",
                "                tickfont=dict(size=18, family='Helvetica', color='black')),\n",
                "    yaxis=dict(#type='log', \n",
                "                title='Number of Environment Interactions',\n",
                "                title_font=dict(size=30, family='Helvetica', color='black'),\n",
                "                tickfont=dict(size=18, family='Helvetica', color='black')),\n",
                "    legend=dict(\n",
                "    font=dict(size=18, family='Arial', color='black'),\n",
                "    ),\n",
                "    legend_title='Presentations Solved',\n",
                "    # template='plotly_white',\n",
                "    width=1200,\n",
                "    height=900\n",
                ")\n",
                "\n",
                "# fig.write_image(\"images/env_vs_horizon.pdf\")\n",
                "fig.show()"
            ]
        },
        {
            "cell_type": "code",
            "execution_count": null,
            "metadata": {},
            "outputs": [],
            "source": []
        }
    ],
    "metadata": {
        "kernelspec": {
            "display_name": "ac_solver-zphkDIyV-py3.9",
            "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": "3.9.16"
        }
    },
    "nbformat": 4,
    "nbformat_minor": 2
}