eth2economics/code/beaconrunner2049/observers/observed-br2049.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# \"It's only wafer-thin\" -- Preventing state explosion with observers\n",
    "\n",
    "Following the discussion [here](https://community.cadcad.org/t/mitigating-cadcad-overhead/140/8), we present an alternate version of the [Beacon Runner simulation](beacon_runner_2049.ipynb) where the cadCAD state is not deep-copied and retained at each step. This makes the simulation significantly faster, but means that we cannot rely on our previous process where \"statistics\" (e.g., what we called `active_stake_branch_0`) are obtained from the historical record contained in `df`.\n",
    "\n",
    "Instead, we define functions of the state (\"observers\") which are seamlessly included as state variables. We use a [fork of cadCAD](https://community.cadcad.org/t/mitigating-cadcad-overhead/140/5?u=barnabemonnot) where the state is shallow-copied at each step and make sure our observers return shallow values. For instance, we don't necessarily want to record the state of the `Network` and the two `BeaconState` objects, but we do want to see the evolution of the stake for both groups of validators in both partitions. The measured value is returned by a simple function of the current state, and can be recorded shallowly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import specs\n",
    "import importlib\n",
    "from eth2spec.config.config_util import prepare_config\n",
    "\n",
    "prepare_config(\".\", \"medium\")\n",
    "importlib.reload(specs)\n",
    "\n",
    "import network as nt\n",
    "import brlib\n",
    "\n",
    "import copy\n",
    "\n",
    "from cadCAD.configuration import Configuration\n",
    "from cadCAD.engine import ExecutionMode, ExecutionContext, Executor\n",
    "import pandas as pd\n",
    "\n",
    "from cadCADsupSUP import (\n",
    "    get_observed_initial_conditions, get_observed_psubs,\n",
    "    add_loop_ic, add_loop_psubs\n",
    ")\n",
    "\n",
    "genesis_state = brlib.get_genesis_state(1000)\n",
    "brlib.process_genesis_block(genesis_state)\n",
    "specs.process_slots(genesis_state, 1)\n",
    "genesis_state_2 = copy.deepcopy(genesis_state)\n",
    "\n",
    "set_a = nt.NetworkSet(validators=list(range(0,600)), beacon_state=genesis_state)\n",
    "set_b = nt.NetworkSet(validators=list(range(600,1000)), beacon_state=genesis_state_2)\n",
    "network = nt.Network(sets=list([set_a, set_b]))\n",
    "\n",
    "initial_conditions = {\n",
    "    'network': network\n",
    "}\n",
    "\n",
    "def active0(state):\n",
    "    return sum(\n",
    "        [state[\"network\"].sets[0].beacon_state.validators[validator_index].effective_balance\n",
    "         for validator_index in state[\"network\"].sets[0].validators]\n",
    "    ) / len(state[\"network\"].sets[0].validators)\n",
    "\n",
    "def inactive0(state):\n",
    "    return sum(\n",
    "        [state[\"network\"].sets[0].beacon_state.validators[validator_index].effective_balance\n",
    "         for validator_index in state[\"network\"].sets[1].validators]\n",
    "    ) / len(state[\"network\"].sets[1].validators)\n",
    "\n",
    "def active1(state):\n",
    "    return sum(\n",
    "        [state[\"network\"].sets[1].beacon_state.validators[validator_index].effective_balance\n",
    "         for validator_index in state[\"network\"].sets[1].validators]\n",
    "    ) / len(state[\"network\"].sets[1].validators)\n",
    "\n",
    "def inactive1(state):\n",
    "    return sum(\n",
    "        [state[\"network\"].sets[1].beacon_state.validators[validator_index].effective_balance\n",
    "         for validator_index in state[\"network\"].sets[0].validators]\n",
    "    ) / len(state[\"network\"].sets[0].validators)\n",
    "\n",
    "# We add the dictionary of observers\n",
    "observers = {\n",
    "    \"active_stake_branch_0\": active0,\n",
    "    \"inactive_stake_branch_0\": inactive0,\n",
    "    \"active_stake_branch_1\": active1,\n",
    "    \"inactive_stake_branch_1\": inactive1\n",
    "}\n",
    "\n",
    "# Write out our PSUBs as before\n",
    "psubs = [\n",
    "    # Step 1+2\n",
    "    {\n",
    "        'policies': {\n",
    "            'action': brlib.attest_policy\n",
    "        },\n",
    "        'variables': {\n",
    "            'network': brlib.disseminate_attestations\n",
    "        }\n",
    "    },\n",
    "    # Step 3+4\n",
    "    {\n",
    "        'policies': {\n",
    "            'action': brlib.propose_policy\n",
    "        },\n",
    "        'variables': {\n",
    "            'network': brlib.disseminate_blocks\n",
    "        }\n",
    "    }\n",
    "]\n",
    "\n",
    "initial_conditions = add_loop_ic(initial_conditions)\n",
    "psubs = add_loop_psubs(psubs)\n",
    "\n",
    "# We transform initial conditions and PSUBs to include initial observed values\n",
    "# and include observed values as state variables\n",
    "obs_ic = get_observed_initial_conditions(initial_conditions, observers)\n",
    "obs_psubs = get_observed_psubs(psubs, observers)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%capture\n",
    "\n",
    "simulation_parameters = {\n",
    "    'T': range(300),\n",
    "    'N': 1,\n",
    "    'M': {}\n",
    "}\n",
    "\n",
    "config = Configuration(initial_state=obs_ic,\n",
    "                       partial_state_update_blocks=obs_psubs,\n",
    "                       sim_config=simulation_parameters\n",
    "                      )\n",
    "\n",
    "exec_mode = ExecutionMode()\n",
    "exec_context = ExecutionContext(exec_mode.single_proc)\n",
    "executor = Executor(exec_context, [config])\n",
    "raw_result, tensor = executor.execute()\n",
    "df = pd.DataFrame(raw_result)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now get functions of our observers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Activity ratio of validators in the first partition\n",
    "df['activity_ratio_0'] = df['active_stake_branch_0'] * 60 / df['inactive_stake_branch_0'] / 40\n",
    "\n",
    "# Activity ratio of validators in the second partition\n",
    "df['activity_ratio_1'] = df['active_stake_branch_1'] * 40 / df['inactive_stake_branch_1'] / 60"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "All entries of the `\"network\"` column in `df` refer to the same object, which is the last update to the `Network` state variable. So we can just drop that column."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.drop(columns = [\"network\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>loop_time</th>\n",
       "      <th>active_stake_branch_0</th>\n",
       "      <th>inactive_stake_branch_0</th>\n",
       "      <th>active_stake_branch_1</th>\n",
       "      <th>inactive_stake_branch_1</th>\n",
       "      <th>run</th>\n",
       "      <th>substep</th>\n",
       "      <th>timestep</th>\n",
       "      <th>activity_ratio_0</th>\n",
       "      <th>activity_ratio_1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.588574e+09</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.500000</td>\n",
       "      <td>0.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.588574e+09</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1.500000</td>\n",
       "      <td>0.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.588574e+09</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1.500000</td>\n",
       "      <td>0.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.588574e+09</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1.500000</td>\n",
       "      <td>0.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.588574e+09</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1.500000</td>\n",
       "      <td>0.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>596</th>\n",
       "      <td>1.588575e+09</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>1.700000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>6.000000e+09</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>298</td>\n",
       "      <td>2.823529</td>\n",
       "      <td>3.555556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>597</th>\n",
       "      <td>1.588575e+09</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>1.700000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>6.000000e+09</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>299</td>\n",
       "      <td>2.823529</td>\n",
       "      <td>3.555556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>598</th>\n",
       "      <td>1.588575e+09</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>1.700000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>6.000000e+09</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>299</td>\n",
       "      <td>2.823529</td>\n",
       "      <td>3.555556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>599</th>\n",
       "      <td>1.588575e+09</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>1.700000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>6.000000e+09</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>300</td>\n",
       "      <td>2.823529</td>\n",
       "      <td>3.555556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>600</th>\n",
       "      <td>1.588575e+09</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>1.700000e+10</td>\n",
       "      <td>3.200000e+10</td>\n",
       "      <td>6.000000e+09</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>300</td>\n",
       "      <td>2.823529</td>\n",
       "      <td>3.555556</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>601 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        loop_time  active_stake_branch_0  inactive_stake_branch_0  \\\n",
       "0    1.588574e+09           3.200000e+10             3.200000e+10   \n",
       "1    1.588574e+09           3.200000e+10             3.200000e+10   \n",
       "2    1.588574e+09           3.200000e+10             3.200000e+10   \n",
       "3    1.588574e+09           3.200000e+10             3.200000e+10   \n",
       "4    1.588574e+09           3.200000e+10             3.200000e+10   \n",
       "..            ...                    ...                      ...   \n",
       "596  1.588575e+09           3.200000e+10             1.700000e+10   \n",
       "597  1.588575e+09           3.200000e+10             1.700000e+10   \n",
       "598  1.588575e+09           3.200000e+10             1.700000e+10   \n",
       "599  1.588575e+09           3.200000e+10             1.700000e+10   \n",
       "600  1.588575e+09           3.200000e+10             1.700000e+10   \n",
       "\n",
       "     active_stake_branch_1  inactive_stake_branch_1  run  substep  timestep  \\\n",
       "0             3.200000e+10             3.200000e+10    1        0         0   \n",
       "1             3.200000e+10             3.200000e+10    1        1         1   \n",
       "2             3.200000e+10             3.200000e+10    1        2         1   \n",
       "3             3.200000e+10             3.200000e+10    1        1         2   \n",
       "4             3.200000e+10             3.200000e+10    1        2         2   \n",
       "..                     ...                      ...  ...      ...       ...   \n",
       "596           3.200000e+10             6.000000e+09    1        2       298   \n",
       "597           3.200000e+10             6.000000e+09    1        1       299   \n",
       "598           3.200000e+10             6.000000e+09    1        2       299   \n",
       "599           3.200000e+10             6.000000e+09    1        1       300   \n",
       "600           3.200000e+10             6.000000e+09    1        2       300   \n",
       "\n",
       "     activity_ratio_0  activity_ratio_1  \n",
       "0            1.500000          0.666667  \n",
       "1            1.500000          0.666667  \n",
       "2            1.500000          0.666667  \n",
       "3            1.500000          0.666667  \n",
       "4            1.500000          0.666667  \n",
       "..                ...               ...  \n",
       "596          2.823529          3.555556  \n",
       "597          2.823529          3.555556  \n",
       "598          2.823529          3.555556  \n",
       "599          2.823529          3.555556  \n",
       "600          2.823529          3.555556  \n",
       "\n",
       "[601 rows x 10 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x118f38a30>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df[df.substep == 1].plot('timestep', ['activity_ratio_0', 'activity_ratio_1'])"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.8.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}