Datasets:
File size: 11,190 Bytes
4bd5225 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 | from __future__ import annotations
from typing import Any
import gymnasium as gym
from gymnasium import spaces
import numpy as np
from .actions import decode_continuous_action, decode_discrete_action, discrete_action_grid
from .config import BenchmarkConfig, load_config
from .data import BenchmarkData, load_benchmark_data
class SolarChainBenchmarkEnv(gym.Env):
"""Gymnasium environment for physics-constrained SolarChain market governance."""
metadata = {"render_modes": []}
def __init__(self, config: BenchmarkConfig | None = None, data: BenchmarkData | None = None):
super().__init__()
self.config = config or load_config()
self.data = data or load_benchmark_data(self.config.data_dir)
self._rng = np.random.default_rng(self.config.seed)
if self.config.action_mode == "discrete":
self.action_space = spaces.Discrete(len(discrete_action_grid(self.config)))
else:
self.action_space = spaces.Box(low=0.0, high=1.0, shape=(3,), dtype=np.float32)
self.observation_space = spaces.Box(low=-10.0, high=10.0, shape=(12,), dtype=np.float32)
self._step = 0
self._episode_start_hour = 0
self._liquidity = self.config.market.initial_liquidity
self._token_price = self.config.market.initial_token_price
self._peak_price = self._token_price
self._peak_liquidity = self._liquidity
self._prev_action = self._static_actual_action()
self._city_rewards = {city: 0.0 for city in self.data.cities}
self._last_info: dict[str, Any] = {}
def reset(self, *, seed: int | None = None, options: dict[str, Any] | None = None):
super().reset(seed=seed)
if seed is not None:
self._rng = np.random.default_rng(seed)
self._step = 0
self._episode_start_hour = self._sample_episode_start_hour(options)
self._liquidity = self.config.market.initial_liquidity
self._token_price = self.config.market.initial_token_price
self._peak_price = self._token_price
self._peak_liquidity = self._liquidity
self._prev_action = self._static_actual_action()
self._city_rewards = {city: 0.0 for city in self.data.cities}
self._last_info = {}
return self._observation(), {}
def step(self, action):
actual = self._decode_action(action)
reward_ratio, liquidity_ratio, burn_rate = map(float, actual)
absolute_hour = self._current_absolute_hour()
hour = absolute_hour % 24
rows = self.data.city_hour[self.data.city_hour["absolute_hour"].eq(absolute_hour)]
verified_mwh = float(rows["verified_W"].sum() / 1_000_000.0)
reported_mwh = float(rows["reported_W"].sum() / 1_000_000.0)
pmax_mwh = float(rows["pmax_W"].sum() / 1_000_000.0)
excess_mwh = float(rows["physics_excess_W"].sum() / 1_000_000.0)
rejected_reported_mwh = float(rows["rejected_reported_W"].sum() / 1_000_000.0)
raw_physics_record_rate = float(rows["violation_count"].sum() / max(rows["record_count"].sum(), 1))
trade_rows = self.data.trades[self.data.trades["absolute_hour"].eq(absolute_hour)]
demand_mwh = float(trade_rows["energy_purchased_MW"].sum())
if demand_mwh <= 0:
market_hour = self.data.market[self.data.market["absolute_hour"].eq(absolute_hour)]
demand_mwh = max(float(market_hour["total_verified_MW"].sum()), 0.001)
unsafe_supply = max(rejected_reported_mwh, excess_mwh, max(reported_mwh - pmax_mwh, 0.0))
backing_pressure = min(
reward_ratio + liquidity_ratio,
self.config.market.max_total_allocation,
) / max(self.config.market.max_total_allocation, 1e-9)
unsafe_backed_mwh = unsafe_supply * backing_pressure
backed_supply_mwh = verified_mwh + unsafe_backed_mwh
physics_rate = unsafe_backed_mwh / max(backed_supply_mwh, 1e-9)
liquidity_added = backed_supply_mwh * liquidity_ratio
reward_tokens = backed_supply_mwh * reward_ratio
effective_demand = demand_mwh * max(0.75, 1.0 - 0.80 * burn_rate)
available = self._liquidity + liquidity_added
matched = min(available, effective_demand)
unmet = max(effective_demand - available, 0.0)
self._liquidity = max(available - matched, 0.0)
self._peak_liquidity = max(self._peak_liquidity, self._liquidity)
liquidity_drawdown = 1.0 - self._liquidity / max(self._peak_liquidity, 1e-12)
slippage = float(effective_demand / max(available + 0.05, 0.05))
action_delta = float(np.linalg.norm(actual - self._prev_action, ord=1))
supply_gap = (verified_mwh - effective_demand) / max(effective_demand, 1e-6)
price_return = (
0.04 * supply_gap
+ 0.03 * matched
+ 0.10 * burn_rate
- 0.08 * slippage
- 0.02 * reward_tokens
+ float(self._rng.normal(0.0, 0.004))
)
price_return = float(np.clip(price_return, -0.35, 0.35))
self._token_price = max(0.05, self._token_price * (1.0 + price_return))
self._peak_price = max(self._peak_price, self._token_price)
token_drawdown = 1.0 - self._token_price / max(self._peak_price, 1e-12)
city_rewards = self._allocate_city_rewards(rows, reward_tokens, liquidity_ratio)
for city, value in city_rewards.items():
self._city_rewards[city] += value
fairness_variance = float(np.var(list(city_rewards.values()))) if city_rewards else 0.0
physics_penalty = 0.0 if self.config.no_physics_penalty else self.config.reward.physics_penalty * (physics_rate + unsafe_backed_mwh)
reward = (
matched
- self.config.reward.drawdown_penalty * liquidity_drawdown
- self.config.reward.action_jitter_penalty * action_delta
- self.config.reward.unmet_demand_penalty * unmet
- self.config.reward.fairness_penalty * fairness_variance
- physics_penalty
)
self._prev_action = actual
self._step += 1
truncated = self._step >= self.config.episode_steps
terminated = False
self._last_info = {
"absolute_hour": absolute_hour,
"episode_start_hour": self._episode_start_hour,
"hour": hour,
"reward_ratio": reward_ratio,
"liquidity_ratio": liquidity_ratio,
"burn_rate": burn_rate,
"verified_supply_MWh": verified_mwh,
"reported_supply_MWh": reported_mwh,
"physics_excess_MWh": unsafe_supply,
"unsafe_backed_MWh": unsafe_backed_mwh,
"artificial_liquidity_MWh": unsafe_backed_mwh * liquidity_ratio,
"physics_violation_rate": physics_rate,
"raw_physics_record_rate": raw_physics_record_rate,
"demand_MWh": demand_mwh,
"matched_energy_MWh": matched,
"unmet_demand_MWh": unmet,
"liquidity": self._liquidity,
"liquidity_depth": self._liquidity / max(self.config.market.initial_liquidity, 1e-9),
"token_price": self._token_price,
"max_drawdown": liquidity_drawdown,
"token_drawdown": token_drawdown,
"action_jitter": action_delta,
"slippage": slippage,
"city_rewards": city_rewards,
}
return self._observation(), float(reward), terminated, truncated, dict(self._last_info)
def _decode_action(self, action) -> np.ndarray:
if self.config.action_mode == "discrete":
return decode_discrete_action(int(action), self.config)
return decode_continuous_action(np.asarray(action, dtype=np.float32), self.config)
def _sample_episode_start_hour(self, options: dict[str, Any] | None = None) -> int:
if options and "start_hour" in options:
start_hour = int(options["start_hour"])
upper_bound = max(self.data.hour_count - self.config.episode_steps, 0)
return int(np.clip(start_hour, 0, upper_bound))
if self.data.hour_count <= self.config.episode_steps:
return 0
latest_start = self.data.hour_count - self.config.episode_steps
daily_starts = np.arange(0, latest_start + 1, 24, dtype=np.int64)
if len(daily_starts) == 0:
return 0
return int(self._rng.choice(daily_starts))
def _current_absolute_hour(self) -> int:
if self.data.hour_count <= 0:
return self._episode_start_hour + self._step
return min(self._episode_start_hour + self._step, self.data.hour_count - 1)
def _allocate_city_rewards(self, rows, reward_tokens: float, liquidity_ratio: float) -> dict[str, float]:
total_verified = max(float(rows["verified_W"].sum()), 1e-9)
city_rewards: dict[str, float] = {}
for row in rows.to_dict("records"):
city = str(row["city"])
share = float(row["verified_W"]) / total_verified
trust_discount = 1.0 - float(row["violation_rate"])
city_rewards[city] = reward_tokens * share * trust_discount + liquidity_ratio * share * 0.01
return city_rewards
def _observation(self) -> np.ndarray:
absolute_hour = self._current_absolute_hour()
hour = absolute_hour % 24
rows = self.data.city_hour[self.data.city_hour["absolute_hour"].eq(absolute_hour)]
market = self.data.market[self.data.market["absolute_hour"].eq(absolute_hour)]
verified_mwh = float(rows["verified_W"].sum() / 1_000_000.0)
reported_mwh = float(rows["reported_W"].sum() / 1_000_000.0)
pmax_mwh = float(rows["pmax_W"].sum() / 1_000_000.0)
violation_rate = float(rows["violation_count"].sum() / max(rows["record_count"].sum(), 1))
demand_mwh = float(self.data.trades[self.data.trades["absolute_hour"].eq(absolute_hour)]["energy_purchased_MW"].sum())
static_slippage = float(market["slippage_solarchain_pct"].mean()) if not market.empty else 0.0
gap = (verified_mwh - demand_mwh) / max(demand_mwh, 1e-6)
obs = np.array(
[
float(np.sin(2 * np.pi * hour / 24)),
float(np.cos(2 * np.pi * hour / 24)),
verified_mwh,
reported_mwh,
pmax_mwh,
gap,
self._liquidity,
self._token_price,
violation_rate,
static_slippage,
float(self._prev_action[0]),
float(self._prev_action[1]),
],
dtype=np.float32,
)
return np.clip(obs, -10.0, 10.0)
def _static_actual_action(self) -> np.ndarray:
return np.array(
[
self.config.market.static_reward_ratio,
self.config.market.static_liquidity_ratio,
self.config.market.static_burn_rate,
],
dtype=np.float32,
)
def latest_info(self) -> dict[str, Any]:
return dict(self._last_info)
def city_rewards(self) -> dict[str, float]:
return dict(self._city_rewards)
|