{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "feaca835",
   "metadata": {},
   "source": [
    "# Synthetic Trajectory Generation with Nomad\n",
    "\n",
    "This notebook demonstrates how to generate realistic synthetic human mobility trajectories."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "c283fa66",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-08-13T02:19:28.574446Z",
     "iopub.status.busy": "2026-08-13T02:19:28.574446Z",
     "iopub.status.idle": "2026-08-13T02:19:30.744183Z",
     "shell.execute_reply": "2026-08-13T02:19:30.743183Z"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import time\n",
    "from pathlib import Path\n",
    "from joblib import Parallel, delayed\n",
    "\n",
    "import nomad.data as data_folder\n",
    "from nomad.city_gen import City\n",
    "from nomad.traj_gen import Agent, Population\n",
    "from nomad.stop_detection.viz import plot_pings, plot_time_barcode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "dbf680cf",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-08-13T02:19:30.746184Z",
     "iopub.status.busy": "2026-08-13T02:19:30.745183Z",
     "iopub.status.idle": "2026-08-13T02:19:31.444570Z",
     "shell.execute_reply": "2026-08-13T02:19:31.444570Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "City: Garden City\n",
      "Dimensions: (22, 22)\n",
      "Buildings: 106\n"
     ]
    }
   ],
   "source": [
    "data_dir = Path(data_folder.__file__).parent\n",
    "city = City.from_geopackage(data_dir / \"garden-city.gpkg\")\n",
    "city._build_hub_network(hub_size=16)\n",
    "city.compute_gravity(exponent=2.0)\n",
    "city.compute_shortest_paths(callable_only=True)\n",
    "\n",
    "print(f\"City: {city.name}\")\n",
    "print(f\"Dimensions: {city.dimensions}\")\n",
    "print(f\"Buildings: {len(city.buildings_gdf)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9534f6e1",
   "metadata": {},
   "source": [
    "## Part 1: Effect of Sampling Parameters on Sparsity\n",
    "\n",
    "Generate 3 agents with 2-day trajectories, varying beta_duration and beta_start \n",
    "to show their effect on sparsity (q = observed points / ground truth points)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "df0a983d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-08-13T02:19:31.447572Z",
     "iopub.status.busy": "2026-08-13T02:19:31.446571Z",
     "iopub.status.idle": "2026-08-13T02:19:32.581034Z",
     "shell.execute_reply": "2026-08-13T02:19:32.581034Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Agent 0: q=0.051, beta_start=100, beta_dur=60\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Agent 1: q=0.031, beta_start=250, beta_dur=150\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Agent 2: q=0.135, beta_start=400, beta_dur=240\n"
     ]
    }
   ],
   "source": [
    "population = Population(city)\n",
    "population.generate_agents(N=3, seed=42, name_count=2)\n",
    "\n",
    "# Vary beta_duration and beta_start to target different sparsity levels\n",
    "sampling_params = [\n",
    "    {'beta_ping': 5, 'beta_start': 100, 'beta_durations': 60},   \n",
    "    {'beta_ping': 5, 'beta_start': 250, 'beta_durations': 150},  \n",
    "    {'beta_ping': 5, 'beta_start': 400, 'beta_durations': 240}   \n",
    "]\n",
    "\n",
    "# Generate 2-day trajectories for quick visualization\n",
    "for i, (agent_id, agent) in enumerate(population.roster.items()):\n",
    "    agent.generate_trajectory(\n",
    "        datetime=pd.Timestamp(\"2024-01-01T07:00-04:00\"),\n",
    "        end_time=pd.Timestamp(\"2024-01-03T07:00-04:00\"),\n",
    "        seed=i\n",
    "    )\n",
    "\n",
    "    agent.set_beta_params(sampling_params[i])\n",
    "    agent.sample_trajectory(\n",
    "        replace_sparse_traj=True,\n",
    "        seed=i\n",
    "    )\n",
    "    \n",
    "    q = len(agent.sparse_traj) / len(agent.trajectory)\n",
    "    print(f\"Agent {i}: q={q:.3f}, beta_start={sampling_params[i]['beta_start']}, \"\n",
    "          f\"beta_dur={sampling_params[i]['beta_durations']}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "2b39c915",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-08-13T02:19:32.583037Z",
     "iopub.status.busy": "2026-08-13T02:19:32.583037Z",
     "iopub.status.idle": "2026-08-13T02:19:34.735046Z",
     "shell.execute_reply": "2026-08-13T02:19:34.735046Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\franc\\AppData\\Local\\Temp\\ipykernel_22336\\3481434713.py:24: UserWarning: FigureCanvasAgg is non-interactive, and thus cannot be shown\n",
      "  plt.show()\n"
     ]
    }
   ],
   "source": [
    "fig, axes = plt.subplots(2, 3, figsize=(15, 10), \n",
    "                         gridspec_kw={'height_ratios': [10, 1]})\n",
    "\n",
    "for i, (agent_id, agent) in enumerate(population.roster.items()):\n",
    "    ax_map = axes[0, i]\n",
    "    ax_barcode = axes[1, i]\n",
    "    \n",
    "    city.plot_city(ax=ax_map, doors=False, address=False)\n",
    "    \n",
    "    traj = agent.sparse_traj\n",
    "    plot_pings(traj, ax=ax_map, s=15, point_color='red', \n",
    "               x='x', y='y', timestamp='timestamp')\n",
    "    \n",
    "    plot_time_barcode(traj['timestamp'], ax=ax_barcode, set_xlim=True)\n",
    "    \n",
    "    q = len(traj) / len(agent.trajectory)\n",
    "    ax_map.set_title(f\"Agent {i}: {len(traj)} obs (q={q:.2f})\\n\"\n",
    "                     f\"beta_start={sampling_params[i]['beta_start']}, \"\n",
    "                     f\"beta_dur={sampling_params[i]['beta_durations']}\")\n",
    "    ax_map.set_axis_off()\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.savefig('data/trajectories_visualization.png', dpi=150, bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4738e919",
   "metadata": {},
   "source": [
    "## Part 2: Parallel Generation at Scale\n",
    "\n",
    "Generate trajectories for 15 users using parallelization."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "dae35ed1",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-08-13T02:19:34.737046Z",
     "iopub.status.busy": "2026-08-13T02:19:34.737046Z",
     "iopub.status.idle": "2026-08-13T02:19:34.750292Z",
     "shell.execute_reply": "2026-08-13T02:19:34.750292Z"
    }
   },
   "outputs": [],
   "source": [
    "def generate_agent_trajectory(args):\n",
    "    \"\"\"Worker function for parallel generation.\"\"\"\n",
    "    identifier, home, work, seed = args\n",
    "    data_dir = Path(data_folder.__file__).parent\n",
    "    city = City.from_geopackage(data_dir / \"garden-city.gpkg\")\n",
    "    city._build_hub_network(hub_size=16)\n",
    "    city.compute_gravity(exponent=2.0)\n",
    "    city.compute_shortest_paths(callable_only=True)\n",
    "    agent = Agent(identifier=identifier, city=city, home=home, workplace=work)\n",
    "    \n",
    "    agent.generate_trajectory(\n",
    "        datetime=pd.Timestamp(\"2024-01-01T07:00-04:00\"),\n",
    "        end_time=pd.Timestamp(\"2024-01-08T07:00-04:00\"),\n",
    "        seed=seed\n",
    "    )\n",
    "    agent.set_beta_params(beta_start=None, beta_durations=None, beta_ping=5)\n",
    "    agent.sample_trajectory(\n",
    "        replace_sparse_traj=True,\n",
    "        seed=seed\n",
    "    )\n",
    "    sparse_df = agent.sparse_traj.copy()\n",
    "    sparse_df['user_id'] = identifier\n",
    "    sparse_df['home'] = home\n",
    "    sparse_df['workplace'] = work\n",
    "    return sparse_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "21bd0af5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-08-13T02:19:34.752294Z",
     "iopub.status.busy": "2026-08-13T02:19:34.752294Z",
     "iopub.status.idle": "2026-08-13T02:19:34.765034Z",
     "shell.execute_reply": "2026-08-13T02:19:34.765034Z"
    }
   },
   "outputs": [],
   "source": [
    "n_agents = 15\n",
    "rng = np.random.default_rng(100)\n",
    "homes = city.buildings_gdf[city.buildings_gdf['building_type'] == 'home']['id'].to_numpy()\n",
    "workplaces = city.buildings_gdf[city.buildings_gdf['building_type'] == 'workplace']['id'].to_numpy()\n",
    "\n",
    "agent_params = [\n",
    "    (f'agent_{i:04d}',\n",
    "     rng.choice(homes),\n",
    "     rng.choice(workplaces),\n",
    "     i)\n",
    "    for i in range(n_agents)\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "25602bf2",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-08-13T02:19:34.767036Z",
     "iopub.status.busy": "2026-08-13T02:19:34.767036Z",
     "iopub.status.idle": "2026-08-13T02:19:48.658913Z",
     "shell.execute_reply": "2026-08-13T02:19:48.658913Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 12 concurrent workers.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generating 15 agents in parallel...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done   2 out of  15 | elapsed:   11.9s remaining:  1.3min\n",
      "[Parallel(n_jobs=-1)]: Done   4 out of  15 | elapsed:   12.1s remaining:   33.4s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done   6 out of  15 | elapsed:   12.3s remaining:   18.5s\n",
      "[Parallel(n_jobs=-1)]: Done   8 out of  15 | elapsed:   12.4s remaining:   10.9s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  10 out of  15 | elapsed:   12.6s remaining:    6.2s\n",
      "[Parallel(n_jobs=-1)]: Done  12 out of  15 | elapsed:   12.6s remaining:    3.1s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generated 15 agents in 13.87s (0.92s per agent)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  15 out of  15 | elapsed:   13.8s finished\n"
     ]
    }
   ],
   "source": [
    "print(f\"Generating {n_agents} agents in parallel...\")\n",
    "start_time = time.time()\n",
    "\n",
    "results = Parallel(n_jobs=-1, verbose=10)(\n",
    "    delayed(generate_agent_trajectory)(params) for params in agent_params\n",
    ")\n",
    "\n",
    "generation_time = time.time() - start_time\n",
    "print(f\"Generated {n_agents} agents in {generation_time:.2f}s ({generation_time/n_agents:.2f}s per agent)\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "4682ebc3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-08-13T02:19:48.660915Z",
     "iopub.status.busy": "2026-08-13T02:19:48.660915Z",
     "iopub.status.idle": "2026-08-13T02:19:49.158483Z",
     "shell.execute_reply": "2026-08-13T02:19:49.158483Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saved sparse trajectories to data/trajectories_15_users\n"
     ]
    }
   ],
   "source": [
    "parallel_population = Population(city)\n",
    "for df, params in zip(results, agent_params):\n",
    "    identifier, home, work, seed = params\n",
    "    agent = Agent(identifier=identifier, city=city, home=home, workplace=work, seed=seed)\n",
    "    agent.sparse_traj = df.drop(columns=['home', 'workplace'])\n",
    "    parallel_population.add_agent(agent, verbose=False)\n",
    "\n",
    "parallel_population.reproject_to_mercator(sparse_traj=True)\n",
    "\n",
    "output_path = 'data/trajectories_15_users'\n",
    "parallel_population.save_pop(\n",
    "    sparse_path=str(output_path),\n",
    "    fmt='parquet'\n",
    ")\n",
    "print(f\"Saved sparse trajectories to {output_path}\")"
   ]
  }
 ],
 "metadata": {
  "jupytext": {
   "cell_metadata_filter": "all",
   "formats": "ipynb,py:percent"
  },
  "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": "3.10.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}