example_notebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SFINCS - grpc4bmi example\n",
    "\n",
    "This notebook shows how the SFINCS BMI server built in this package can be used\n",
    "with grpc4bmi. It assumes you have built the docker container with\n",
    "\n",
    "```\n",
    "cd src\n",
    "docker build -t sfincs-bmiserver .\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Requirements\n",
    "\n",
    "We use grpc4bmi to communicate with the BMI model inside a container."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# pip installa grpc4bmi"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Case specification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets use the example parameter set in https://github.com/Deltares/hydromt_sfincs/tree/main/examples/sfincs_compound"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# !git clone https://github.com/Deltares/hydromt_sfincs/"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Example above returns 0 for each time method, which is incorrect so we will use the alternate parameter set below"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# !cp -r ~/dcache/parameter-sets/sfincs_humber ./\n",
    "# !cp -r ~/ewatercycle/sfincs_humber ./"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pathlib import Path\n",
    "parameter_set_dir = Path(\"sfincs_humber\").absolute()\n",
    "parameter_set_dir\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313510\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([313509,    123,    124, ..., 803648, 803649, 803650], dtype=int32)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# see the sfincs internal data structure:\n",
    "# note: the first value is the number of points (`np` in fortran code) and the length of our other arrays\n",
    "indices = np.fromfile('sfincs_humber/sfincs.ind', dtype='int32')\n",
    "print(len(indices))  \n",
    "indices"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Instantiate the model\n",
    "\n",
    "This starts the model container and connects the client to it.\n",
    "We can already get the component name"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Option 1: Build and run the container and connect to it\n",
    "\n",
    "# from grpc4bmi.bmi_client_docker import BmiClientDocker\n",
    "# model = BmiClientDocker(image='sfincs-bmiserver_tim', image_port=50051, work_dir=str(parameter_set_dir), delay=1)\n",
    "\n",
    "# Option 2: Connect to the devcontainer directly, easy for development\n",
    "# see https://github.com/eWaterCycle/sfincs-bmi-server#development-container\n",
    "from grpc4bmi.bmi_grpc_client import BmiClient\n",
    "import grpc\n",
    "model = BmiClient(grpc.insecure_channel(\"localhost:50051\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Sfincs hydrodynamic model (C)'"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.get_component_name()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Interact with the model\n",
    "\n",
    "We can initialize the model and see that it had effect: time variables seem to be set from the input file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.initialize(str(parameter_set_dir / 'sfincs.inp'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.update()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'get_current_time': 950400.0,\n",
       " 'get_start_time': 950400.0,\n",
       " 'get_end_time': 950400.0,\n",
       " 'get_time_units': 's',\n",
       " 'get_time_step': 9.999999974752427e-07}"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "{\n",
    "    'get_current_time': model.get_current_time(), \n",
    "    'get_start_time':model.get_start_time(),\n",
    "    'get_end_time': model.get_end_time(), \n",
    "    'get_time_units': model.get_time_units(), \n",
    "    'get_time_step': model.get_time_step(),\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'get_input_item_count': 6,\n",
       " 'get_output_item_count': 10,\n",
       " 'get_input_var_names': ('zs', 'zb', 'qsrc_1', 'qsrc_2', 'tsrc', 'zst_bnd'),\n",
       " 'get_output_var_names': ('z_xz',\n",
       "  'z_yz',\n",
       "  'zs',\n",
       "  'zb',\n",
       "  'qsrc_1',\n",
       "  'qsrc_2',\n",
       "  'xsrc',\n",
       "  'ysrc',\n",
       "  'tsrc',\n",
       "  'zst_bnd'),\n",
       " 'get_var_grid': 0,\n",
       " 'get_var_itemsize': 4,\n",
       " 'get_var_nbytes': 1254036,\n",
       " 'get_var_type': 'float32',\n",
       " 'get_var_units': 'm above reference level'}"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Var methods\n",
    "{\n",
    "    'get_input_item_count': model.get_input_item_count(),\n",
    "    'get_output_item_count': model.get_output_item_count(),\n",
    "    'get_input_var_names': model.get_input_var_names(),\n",
    "    'get_output_var_names': model.get_output_var_names(),\n",
    "    'get_var_grid': model.get_var_grid('zs'),\n",
    "    'get_var_itemsize': model.get_var_itemsize('zs'),\n",
    "    'get_var_nbytes': model.get_var_nbytes('zs'),\n",
    "    'get_var_type': model.get_var_type('zs'),\n",
    "    'get_var_units': model.get_var_units('zs'),\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'get_var_grid': 1,\n",
       " 'get_var_itemsize': 4,\n",
       " 'get_var_nbytes': 100,\n",
       " 'get_var_type': 'float32',\n",
       " 'get_var_units': None}"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "{\n",
    "    'get_var_grid': model.get_var_grid('qsrc_1'),\n",
    "    'get_var_itemsize': model.get_var_itemsize('qsrc_1'),\n",
    "    'get_var_nbytes': model.get_var_nbytes('qsrc_1'),\n",
    "    'get_var_type': model.get_var_type('qsrc_1'),\n",
    "    'get_var_units': model.get_var_units('qsrc_1'),\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model.get_var_location('zs')\n",
    "# Throws grpc4bmi erorr, maybe because enum in C and Python not matching"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model.get_grid_shape(1, np.zeros((1), dtype=int))\n",
    "# not defined in bmi for unstructured grid"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'get_grid_rank': 2, 'get_grid_size': 313509, 'get_grid_type': 'unstructured'}"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Grid methods\n",
    "grid_id = 0\n",
    "{\n",
    "    'get_grid_rank': model.get_grid_rank(grid_id),\n",
    "    'get_grid_size': model.get_grid_size(grid_id),\n",
    "    'get_grid_type': model.get_grid_type(grid_id),\n",
    "    # 'get_grid_shape': model.get_grid_shape(grid_id, np.empty(model.get_grid_rank(grid_id), dtype=int))\n",
    "}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "313509"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "size = model.get_grid_size(0)\n",
    "size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.2735e+02, 1.3934e+02, 2.3900e+01, 1.5710e+01, 7.8300e+00,\n",
       "       4.1100e+00, 3.6500e+00, 9.3000e-01, 1.3300e+00, 3.9000e-01,\n",
       "       4.6000e-01, 3.2000e-01, 2.5000e-01, 1.4000e-01, 1.6000e-01,\n",
       "       1.3000e-01, 2.0000e-01, 2.4000e-01, 2.5000e-01, 2.3000e-01,\n",
       "       1.3000e-01, 1.6000e-01, 1.2000e-01, 1.9000e-01, 8.0000e-02],\n",
       "      dtype=float32)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.get_value('qsrc_1', np.zeros(25, dtype=np.float32))\n",
    "# model.set_value('qsrc_1', 1.5*np.ones(25, dtype=np.float32))\n",
    "model.get_value('qsrc_1', np.zeros(25, dtype=np.float32))\n",
    "model.get_value('qsrc_1', np.zeros(25, dtype=np.float32))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([9.641531 , 9.0527935, 8.568492 , ..., 0.       , 0.       ,\n",
       "       0.       ], dtype=float32)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.get_value('zs', np.zeros(size, dtype=np.float32))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.2740e+02, 1.3748e+02, 2.3550e+01, 1.5490e+01, 7.7400e+00,\n",
       "       4.1200e+00, 3.6100e+00, 9.2000e-01, 1.3100e+00, 3.9000e-01,\n",
       "       4.6000e-01, 3.2000e-01, 2.5000e-01, 1.4000e-01, 1.6000e-01,\n",
       "       1.3000e-01, 2.0000e-01, 2.4000e-01, 2.4000e-01, 2.3000e-01,\n",
       "       1.2000e-01, 1.6000e-01, 1.1000e-01, 1.8000e-01, 8.0000e-02],\n",
       "      dtype=float32)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.get_value('qsrc_2', np.zeros(25, dtype=np.float32))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 2., 10.], dtype=float32)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.get_value('tsrc', np.zeros(model.get_grid_size(model.get_var_grid('tsrc')), dtype=np.float32))\n",
    "model.set_value('tsrc', np.array([2, 10], dtype=np.float32))\n",
    "model.get_value('tsrc', np.zeros(model.get_grid_size(model.get_var_grid('tsrc')), dtype=np.float32))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model.get_value('xsrc', np.zeros(25, dtype=np.float32))\n",
    "# model.get_grid_x(0, np.empty(size, dtype=np.float32))\n",
    "# Gives segmentation fault which kills container"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Check consistency of return values\n",
    "attempt_1 = model.get_value('z_xz', np.zeros(size, dtype=np.float32))\n",
    "attempt_2 = model.get_value('z_xz', np.zeros(size, dtype=np.float32))\n",
    "np.testing.assert_allclose(attempt_1, attempt_2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "float32 4 100 25\n"
     ]
    }
   ],
   "source": [
    "print(\n",
    "    model.get_var_type('xsrc'),\n",
    "    item_size := model.get_var_itemsize('xsrc'),\n",
    "    total_size := model.get_var_nbytes('xsrc'),\n",
    "    total_size // item_size,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([1.89096e+05, 8.22210e+04, 2.49120e+04, 1.15470e+04, 4.16400e+03,\n",
       "        1.15700e+03, 2.55000e+02, 9.30000e+01, 5.60000e+01, 8.00000e+00]),\n",
       " array([ 0.        ,  4.55390024,  9.10780048, 13.66170025, 18.21560097,\n",
       "        22.76950073, 27.3234005 , 31.87730026, 36.43120193, 40.98509979,\n",
       "        45.53900146]),\n",
       " <BarContainer object of 10 artists>)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.hist(model.get_value('zs', np.zeros(size, dtype=np.float32)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# model.get_value_at_indices('z_xz', np.array([0, 1, 2], dtype='int32'), np.zeros(3, dtype=np.float32))\n",
    "# segfault"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "950410.499130249"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.get_current_time()\n",
    "model.update_until(10.)\n",
    "model.get_current_time()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "950410.499130249\n",
      "950415.7486953735\n",
      "950420.998260498\n",
      "950426.2478256226\n",
      "950431.4973907471\n",
      "950436.7469558716\n",
      "950441.9965209961\n",
      "950447.2460861206\n",
      "950452.4956512451\n",
      "950457.7452163696\n"
     ]
    }
   ],
   "source": [
    "for i in range(10):\n",
    "    print(model.get_current_time())\n",
    "    model.update()\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([1.50933e+05, 1.05755e+05, 3.64440e+04, 1.33810e+04, 5.07600e+03,\n",
       "        1.46100e+03, 2.92000e+02, 9.90000e+01, 5.90000e+01, 9.00000e+00]),\n",
       " array([-1.27353752,  3.40771627,  8.08897018, 12.77022457, 17.45147896,\n",
       "        22.13273239, 26.81398582, 31.49523926, 36.1764946 , 40.85774612,\n",
       "        45.53900146]),\n",
       " <BarContainer object of 10 artists>)"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model.get_value('zs', np.zeros(size))\n",
    "plt.hist(model.get_value('zs', np.zeros(size, dtype=np.float32)))"
   ]
  }
 ],
 "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": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}