{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "f1dce0b3-e06f-49ae-9c17-a52a0d0e4ce3",
   "metadata": {},
   "source": [
    "# Restart and resolution change\n",
    "\n",
    "Fluidsim provides 2 command-line tools to restart a simulation (`fluidsim-restart`) \n",
    "and create a new file with a modified resolution (`fluidsim-modif-resolution`). \n",
    "Let's present these very useful tools!\n",
    "\n",
    "First, one can get help by invoking these tools with the `-h` option:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "859a2448-8e87-49b7-816a-b620c6009496",
   "metadata": {},
   "outputs": [],
   "source": [
    "!fluidsim-restart -h"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "45998ecb-188b-419b-b5a0-f59f01b7e102",
   "metadata": {},
   "outputs": [],
   "source": [
    "!fluidsim-modif-resolution -h"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "10b0ba37-bacc-42d5-b667-caa2d6e5599e",
   "metadata": {
    "tags": [
     "hide-cell"
    ]
   },
   "outputs": [],
   "source": [
    "from pathlib import Path\n",
    "import shutil\n",
    "\n",
    "path_fls = \"/tmp/sim_data\"\n",
    "path_base = Path(path_fls) / \"doc_aniso\"\n",
    "shutil.rmtree(path_base, ignore_errors=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "110dde77-c159-4c1f-ba06-f67290b1a994",
   "metadata": {
    "tags": [
     "hide-cell"
    ]
   },
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "os.environ[\"FLUIDSIM_PATH\"] = path_fls\n",
    "os.environ[\"TRANSONIC_BACKEND\"] = \"python\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3634c512-df0a-4716-bbfb-472353d55dd3",
   "metadata": {},
   "source": [
    "Then, let's use these tools. First we need a simulation directory. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a14010ba-fe29-4dc6-926e-a59a25b6a616",
   "metadata": {
    "tags": [
     "remove-stderr"
    ]
   },
   "outputs": [],
   "source": [
    "from fluidsim.util.scripts.turb_trandom_anisotropic import main\n",
    "\n",
    "params, sim = main(\n",
    "    args=(\n",
    "        '--sub-directory \"doc_aniso\" -nz 12 --ratio-nh-nz 2 --t_end 1 '\n",
    "        '--modify-params \"params.output.periods_print.print_stdout = 0.25; '\n",
    "        'params.output.periods_save.spect_energy_budg = 0\"'\n",
    "    )\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "465b6db5-6ec2-43e4-a47b-51c6d4b06ab5",
   "metadata": {},
   "source": [
    "We then define a variable with the path of the directory containing the results of the simulation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7bc0d08a-1882-416a-b341-b68e16215c28",
   "metadata": {},
   "outputs": [],
   "source": [
    "sim.output.path_run"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e035f9b9-35d7-4f8a-9a52-899fd47741ce",
   "metadata": {},
   "outputs": [],
   "source": [
    "from pprint import pprint\n",
    "\n",
    "\n",
    "def ls_path_run(glob=\"*\"):\n",
    "    pprint(sorted(p.name for p in Path(sim.output.path_run).glob(glob)))\n",
    "\n",
    "\n",
    "ls_path_run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d79d54ea-3db7-497a-be92-476cf6dad549",
   "metadata": {},
   "outputs": [],
   "source": [
    "sim.output.spatial_means.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e3948e8d-cf69-4e62-9521-312e45445ca5",
   "metadata": {},
   "source": [
    "Ah, we see that `t_end = 1` was not enough to obtain a statistically steady state. We need to run the simulation longer. Let's restart it to run until `t_end = 2`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8387162e-7c03-49de-9286-bf5a8a0c4b7a",
   "metadata": {
    "tags": [
     "remove-stderr"
    ]
   },
   "outputs": [],
   "source": [
    "!fluidsim-restart {sim.output.path_run} --t_end 2.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "21a6bcff-30ce-42a0-a257-637bc1b0c38c",
   "metadata": {},
   "outputs": [],
   "source": [
    "ls_path_run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f52cbc23-53ee-42cf-afad-d52a82da273a",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "sim.output.spatial_means.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e4f15fab-8a8a-409f-b470-3fb8807bae6c",
   "metadata": {},
   "source": [
    "Finally, we are not far from a statistically steady state. We can relaunch the simulation to save results about this state. Here we show how to restart a simulation with the spatiotemporal spectra output activated:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "452a2a30-4358-484f-b666-857c85038e8e",
   "metadata": {},
   "outputs": [],
   "source": [
    "!fluidsim-restart {sim.output.path_run} --t_end 3.0 --modify-params \"params.output.periods_save.spatiotemporal_spectra = 2 * pi / (4 * params.N)\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "26919f71-b361-419a-8298-7bb7b1be88d6",
   "metadata": {},
   "outputs": [],
   "source": [
    "ls_path_run()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "11ad70f1-60e0-4b53-9e99-7279ec4a2e9b",
   "metadata": {},
   "source": [
    "`spatiotemporal` is a directory containing the data that can be used to compute spatiotemporal spectra:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f3811502-d957-46bc-908f-aabd24eafd51",
   "metadata": {},
   "outputs": [],
   "source": [
    "ls_path_run(\"spatiotemporal/*\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a409d832-21b1-4703-ad6b-962a8d95beca",
   "metadata": {},
   "outputs": [],
   "source": [
    "sim.output.spatial_means.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "37aeec81-21e5-44fe-aef4-2f4d93efd34b",
   "metadata": {},
   "source": [
    "We want to start a larger simulation from the last state of this small simulation. We first need to create a larger file:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4a2cc4d2-ba33-4805-b60b-54688d98c077",
   "metadata": {},
   "outputs": [],
   "source": [
    "!fluidsim-modif-resolution {sim.output.path_run} 5/4"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "927ffff4-25b1-41b3-bd4d-72f50eb935eb",
   "metadata": {},
   "source": [
    "We can now launch the last simulation from the file just created."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b9b1f74f-d0da-4579-9788-8ecbe6245361",
   "metadata": {},
   "outputs": [],
   "source": [
    "path_new_state = next(Path(sim.output.path_run).glob(\"State_phys*\"))\n",
    "str(path_new_state)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d2f368ae-52b3-482e-8708-188ee7022dad",
   "metadata": {},
   "outputs": [],
   "source": [
    "!fluidsim-restart {path_new_state} --t_end 3.1 --modify-params \"params.NEW_DIR_RESULTS = True; params.nu_2 /= 2; params.output.periods_print.print_stdout = 0.05; params.output.periods_save.spatial_means = 0.01;\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e203da2c-8f2c-418e-b2b1-57850e9418f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "[p.name for p in path_base.glob(\"*\")]"
   ]
  }
 ],
 "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": "python3",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.6"
  },
  "metadata": {
   "execution": {
    "allow_errors": true,
    "timeout": 150
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}