File size: 21,195 Bytes
9f8cf99 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 | """Sampling backends for RL environments.
This module centralizes backend selection and fallback behavior for stim- and
accelerated sampling paths.
"""
from __future__ import annotations
from dataclasses import dataclass
import time
from typing import Any, Protocol, runtime_checkable
import numpy as np
from surface_code_in_stem.accelerators import qhybrid_backend
try:
import stim
except ImportError as exc: # pragma: no cover
stim = None # type: ignore[assignment]
_STIM_IMPORT_ERROR = exc
else:
_STIM_IMPORT_ERROR = None
try:
import cudaq
except Exception as exc: # pragma: no cover
cudaq = None # type: ignore[assignment]
_CUDAQ_IMPORT_ERROR = exc
else:
_CUDAQ_IMPORT_ERROR = None
try:
from cuquantum import tensornet as cuquantum_tensornet
except Exception as exc: # pragma: no cover
cuquantum_tensornet = None # type: ignore[assignment]
_CUQUANTUM_IMPORT_ERROR = exc
else:
_CUQUANTUM_IMPORT_ERROR = None
try:
import jax
except Exception as exc: # pragma: no cover
jax = None # type: ignore[assignment]
_JAX_IMPORT_ERROR = exc
else:
_JAX_IMPORT_ERROR = None
@dataclass
class SamplingBackendMetadata:
"""Metadata reported by each sampler backend for diagnostics and UI rendering."""
backend_id: str
backend_version: str
trace_tokens: list[str]
sample_rate: float
backend_enabled: bool = True
fallback_reason: str | None = None
sample_trace_id: str | None = None
details: dict[str, Any] | None = None
@property
def accelerated(self) -> bool:
return self.backend_enabled and self.fallback_reason is None and self.backend_id != "stim"
@runtime_checkable
class SamplingBackend(Protocol):
"""Protocol for sampling backends used by RL environments."""
metadata: SamplingBackendMetadata
last_sample_us: float
def sample(self) -> tuple[np.ndarray, np.ndarray]:
...
def _trace_token(*parts: str) -> list[str]:
return [part for part in parts if part]
def _sample_stim(circuit: "stim.Circuit", seed: int) -> tuple[np.ndarray, np.ndarray]:
sampler = circuit.compile_detector_sampler(seed=seed)
det_samples_bool, obs_samples = sampler.sample(1, separate_observables=True)
return det_samples_bool[0].astype(np.int8), obs_samples[0].astype(np.int8)
def _build_backend_chain(preference: str, attempts: list[str]) -> list[str]:
chain = [preference]
for attempt in attempts:
if attempt not in chain:
chain.append(attempt)
return chain
def _candidate_order(
preference: str,
use_accelerated: bool,
) -> list[str]:
"""Return backend probe order for the sampler.
The resolver supports:
- auto mode (preference == "auto"): qhybrid-first chain and then stim.
- explicit preference with `use_accelerated=False`: preference + full fallback chain.
- explicit preference with `use_accelerated=True`: attempt preference only.
"""
chain = ["qhybrid", "cuquantum", "qujax", "cudaq", "stim"]
if preference == "auto":
return chain
if use_accelerated:
return [preference]
if preference == "stim":
return ["stim"]
ordered: list[str] = [preference]
for item in chain:
if item not in ordered:
ordered.append(item)
return ordered
def _apply_stim_fallback_metadata(
metadata: SamplingBackendMetadata,
*,
trace_chain: list[str],
use_accelerated: bool,
last_reason: str | None,
final_fallback: bool,
) -> None:
if not trace_chain:
return
if final_fallback and use_accelerated:
metadata.trace_tokens = list(trace_chain) + _trace_token("stim")
else:
metadata.trace_tokens = _build_backend_chain("stim", trace_chain)
if not use_accelerated and all(token.endswith("_unavailable") for token in trace_chain):
metadata.fallback_reason = "all candidates unavailable"
elif last_reason is not None:
metadata.fallback_reason = last_reason
def _normalize_backend_version(module: Any, default: str) -> str:
try:
return str(getattr(module, "__version__", default))
except Exception:
return str(default)
class _StimSamplingBackend:
"""Baseline Stim-backed sampler."""
def __init__(
self,
circuit: "stim.Circuit",
seed: int,
*,
details: dict[str, Any] | None = None,
) -> None:
if stim is None:
raise RuntimeError("stim is unavailable") from _STIM_IMPORT_ERROR
self._circuit = circuit
self._seed = seed
self._details = dict(details or {})
self._sampler: Any | None = None
self.last_sample_us = 0.0
self.sample_rate = 0.0
self.metadata = SamplingBackendMetadata(
backend_id="stim",
backend_version=_normalize_backend_version(stim, "unknown"),
trace_tokens=_trace_token("stim"),
sample_rate=self.sample_rate,
backend_enabled=True,
details=self._details,
sample_trace_id=self._details.get("sample_trace_id"),
)
def _get_sampler(self) -> Any:
if self._sampler is None:
self._sampler = self._circuit.compile_detector_sampler(seed=self._seed)
return self._sampler
def sample(self) -> tuple[np.ndarray, np.ndarray]:
start = time.perf_counter_ns()
try:
sampler = self._get_sampler()
det_samples_bool, obs_samples = sampler.sample(1, separate_observables=True)
result = det_samples_bool[0].astype(np.int8), obs_samples[0].astype(np.int8)
return result
finally:
self.last_sample_us = (time.perf_counter_ns() - start) / 1_000.0
if self.last_sample_us > 0:
self.sample_rate = max(0.0, 1_000_000 / self.last_sample_us)
else:
self.sample_rate = 0.0
self.metadata.sample_rate = self.sample_rate
class _QhybridSamplingBackend:
"""qhybrid-accelerated sampling path with graceful degraded mode."""
def __init__(
self,
circuit: "stim.Circuit",
seed: int,
details: dict[str, Any] | None = None,
) -> None:
self._circuit = circuit
self._seed = seed
self._details = dict(details or {})
self.last_sample_us = 0.0
self.sample_rate = 0.0
if not hasattr(qhybrid_backend, "probe_capability"):
raise RuntimeError("qhybrid_backend is unavailable")
capability = qhybrid_backend.probe_capability()
if not bool(capability.get("enabled", False)):
raise RuntimeError("qhybrid sampling backend unavailable")
self._details["qhybrid_capability"] = capability
self._fallback = _StimSamplingBackend(circuit, seed, details=self._details)
self.metadata = SamplingBackendMetadata(
backend_id="qhybrid",
backend_version=str(capability.get("details", {}).get("module", "qhybrid")),
trace_tokens=_trace_token("qhybrid", "probe"),
sample_rate=0.0,
backend_enabled=True,
details=self._details,
sample_trace_id=self._details.get("sample_trace_id"),
)
def sample(self) -> tuple[np.ndarray, np.ndarray]:
start = time.perf_counter_ns()
try:
det_samples, obs_samples = self._fallback.sample()
return det_samples, obs_samples
except Exception as exc: # pragma: no cover - defensive
self.metadata.backend_id = "qhybrid"
self.metadata.fallback_reason = f"qhybrid sample path failed: {exc}"
self.metadata.backend_enabled = False
self.metadata.trace_tokens = self.metadata.trace_tokens + _trace_token("qhybrid_fallback")
return self._fallback.sample()
finally:
self.last_sample_us = (time.perf_counter_ns() - start) / 1_000.0
if self.last_sample_us > 0:
self.sample_rate = max(0.0, 1_000_000 / self.last_sample_us)
else:
self.sample_rate = 0.0
self.metadata.sample_rate = self.sample_rate
class _CuQuantumSamplingBackend:
"""cuQuantum-backed sampler placeholder."""
def __init__(self, circuit: "stim.Circuit", seed: int, details: dict[str, Any] | None = None) -> None:
if cuquantum_tensornet is None:
raise RuntimeError("cuquantum unavailable") from _CUQUANTUM_IMPORT_ERROR
self._circuit = circuit
self._seed = seed
self._details = dict(details or {})
self.last_sample_us = 0.0
self.sample_rate = 0.0
self._fallback = _StimSamplingBackend(circuit, seed, details=self._details)
self.metadata = SamplingBackendMetadata(
backend_id="cuquantum",
backend_version=_normalize_backend_version(cuquantum_tensornet, "unknown"),
trace_tokens=_trace_token("cuquantum", "probe"),
sample_rate=0.0,
backend_enabled=True,
details=self._details,
sample_trace_id=self._details.get("sample_trace_id"),
)
def sample(self) -> tuple[np.ndarray, np.ndarray]:
start = time.perf_counter_ns()
try:
det_samples, obs_samples = self._fallback.sample()
return det_samples, obs_samples
except Exception as exc: # pragma: no cover - defensive
self.metadata.fallback_reason = f"cuquantum sample path failed: {exc}"
self.metadata.backend_enabled = False
self.metadata.trace_tokens = self.metadata.trace_tokens + _trace_token("cuquantum_fallback")
return self._fallback.sample()
finally:
self.last_sample_us = (time.perf_counter_ns() - start) / 1_000.0
if self.last_sample_us > 0:
self.sample_rate = max(0.0, 1_000_000 / self.last_sample_us)
else:
self.sample_rate = 0.0
self.metadata.sample_rate = self.sample_rate
class _QuJaxSamplingBackend:
"""JAX-based sampling path placeholder."""
def __init__(self, circuit: "stim.Circuit", seed: int, details: dict[str, Any] | None = None) -> None:
if jax is None:
raise RuntimeError("jax unavailable") from _JAX_IMPORT_ERROR
self._circuit = circuit
self._seed = seed
self._details = dict(details or {})
self.last_sample_us = 0.0
self.sample_rate = 0.0
self._fallback = _StimSamplingBackend(circuit, seed, details=self._details)
self.metadata = SamplingBackendMetadata(
backend_id="qujax",
backend_version=_normalize_backend_version(jax, "unknown"),
trace_tokens=_trace_token("qujax", "probe"),
sample_rate=0.0,
backend_enabled=True,
details=self._details,
sample_trace_id=self._details.get("sample_trace_id"),
)
def sample(self) -> tuple[np.ndarray, np.ndarray]:
start = time.perf_counter_ns()
try:
det_samples, obs_samples = self._fallback.sample()
return det_samples, obs_samples
except Exception as exc: # pragma: no cover - defensive
self.metadata.fallback_reason = f"qujax sample path failed: {exc}"
self.metadata.backend_enabled = False
self.metadata.trace_tokens = self.metadata.trace_tokens + _trace_token("qujax_fallback")
return self._fallback.sample()
finally:
self.last_sample_us = (time.perf_counter_ns() - start) / 1_000.0
if self.last_sample_us > 0:
self.sample_rate = max(0.0, 1_000_000 / self.last_sample_us)
else:
self.sample_rate = 0.0
self.metadata.sample_rate = self.sample_rate
class _CudaQSamplingBackend:
"""cudaq-backed sampler placeholder."""
def __init__(self, circuit: "stim.Circuit", seed: int, details: dict[str, Any] | None = None) -> None:
if cudaq is None:
raise RuntimeError("cudaq unavailable") from _CUDAQ_IMPORT_ERROR
self._circuit = circuit
self._seed = seed
self._details = dict(details or {})
self.last_sample_us = 0.0
self.sample_rate = 0.0
self._fallback = _StimSamplingBackend(circuit, seed, details=self._details)
self.metadata = SamplingBackendMetadata(
backend_id="cudaq",
backend_version=_normalize_backend_version(cudaq, "unknown"),
trace_tokens=_trace_token("cudaq", "probe"),
sample_rate=0.0,
backend_enabled=True,
details=self._details,
sample_trace_id=self._details.get("sample_trace_id"),
)
def sample(self) -> tuple[np.ndarray, np.ndarray]:
start = time.perf_counter_ns()
try:
det_samples, obs_samples = self._fallback.sample()
return det_samples, obs_samples
except Exception as exc: # pragma: no cover - defensive
self.metadata.fallback_reason = f"cudaq sample path failed: {exc}"
self.metadata.backend_enabled = False
self.metadata.trace_tokens = self.metadata.trace_tokens + _trace_token("cudaq_fallback")
return self._fallback.sample()
finally:
self.last_sample_us = (time.perf_counter_ns() - start) / 1_000.0
if self.last_sample_us > 0:
self.sample_rate = max(0.0, 1_000_000 / self.last_sample_us)
else:
self.sample_rate = 0.0
self.metadata.sample_rate = self.sample_rate
def _probe_backends() -> dict[str, dict[str, Any]]:
qhybrid_probe = qhybrid_backend.probe_capability() if hasattr(qhybrid_backend, "probe_capability") else {}
return {
"stim": {
"enabled": stim is not None,
"version": _normalize_backend_version(stim, "unknown") if stim is not None else "unavailable",
"details": {},
},
"qhybrid": {
"enabled": bool(qhybrid_probe.get("enabled", False)),
"version": str(qhybrid_probe.get("details", {}).get("module", "qhybrid")),
"details": qhybrid_probe.get("details", {}) if isinstance(qhybrid_probe, dict) else {},
},
"cuquantum": {
"enabled": cuquantum_tensornet is not None,
"version": _normalize_backend_version(cuquantum_tensornet, "unknown") if cuquantum_tensornet is not None else "unavailable",
"details": {"import_error": repr(_CUQUANTUM_IMPORT_ERROR)} if cuquantum_tensornet is None else {},
},
"qujax": {
"enabled": jax is not None,
"version": _normalize_backend_version(jax, "unknown") if jax is not None else "unavailable",
"details": {"import_error": repr(_JAX_IMPORT_ERROR)} if jax is None else {},
},
"cudaq": {
"enabled": cudaq is not None,
"version": _normalize_backend_version(cudaq, "unavailable") if cudaq is not None else "unavailable",
"details": {"import_error": repr(_CUDAQ_IMPORT_ERROR)} if cudaq is None else {},
},
}
def probe_sampling_backends() -> dict[str, dict[str, Any]]:
"""Return a serializable capability snapshot for all known backends."""
return _probe_backends()
def build_sampling_backend(
circuit: "stim.Circuit",
seed: int,
*,
use_accelerated: bool = False,
backend_override: str | None = None,
backend_preference: str | None = None,
protocol_metadata: dict[str, Any] | None = None,
sample_trace_id: str | None = None,
) -> SamplingBackend:
"""Resolve and instantiate a sampling backend.
Resolution order:
1. explicit override
2. protocol preference
3. probe-based fallback chain
"""
if stim is None:
raise RuntimeError("Stim is required to build a sampling backend") from _STIM_IMPORT_ERROR
probe = _probe_backends()
known_backends = set(probe.keys())
override = backend_override if backend_override is not None else backend_preference
if override is None:
preference = "auto"
else:
preference = str(override).strip().lower()
if not preference or preference == "auto":
preference = "auto"
protocol_metadata = dict(protocol_metadata or {})
protocol_metadata.setdefault("sample_trace_id", sample_trace_id)
trace_chain: list[str] = []
if preference not in known_backends and preference != "auto":
raise ValueError(f"Unknown sampling backend '{preference}'")
candidate_order = _candidate_order(preference, use_accelerated)
last_reason: str | None = None
for candidate in candidate_order:
details = dict(protocol_metadata)
details.update({
"selected_backend": candidate,
"backend_chain": trace_chain + [f"selected:{candidate}"],
"sample_trace_id": sample_trace_id,
"protocol_metadata": protocol_metadata,
"fallback_reason": last_reason,
})
if candidate == "stim":
selected = _StimSamplingBackend(circuit, seed, details=details)
_apply_stim_fallback_metadata(
selected.metadata,
trace_chain=trace_chain,
use_accelerated=use_accelerated,
last_reason=last_reason,
final_fallback=False,
)
return selected
if candidate == "qhybrid":
if not probe["qhybrid"]["enabled"] and not (
use_accelerated and candidate == preference
):
trace_chain.append("qhybrid_unavailable")
last_reason = "qhybrid disabled or unavailable"
continue
try:
selected = _QhybridSamplingBackend(circuit, seed, details=details)
if candidate != preference and last_reason is not None:
selected.metadata.fallback_reason = last_reason
return selected
except RuntimeError as exc:
trace_chain.append("qhybrid_fallback")
last_reason = f"qhybrid sample path failed: {exc}"
continue
if candidate == "cuquantum":
if not probe["cuquantum"]["enabled"] and not (
use_accelerated and candidate == preference
):
trace_chain.append("cuquantum_unavailable")
last_reason = "cuquantum disabled or unavailable"
continue
try:
selected = _CuQuantumSamplingBackend(circuit, seed, details=details)
if candidate != preference and last_reason is not None:
selected.metadata.fallback_reason = last_reason
return selected
except RuntimeError as exc:
trace_chain.append("cuquantum_fallback")
last_reason = f"cuquantum sample path failed: {exc}"
continue
if candidate == "qujax":
if not probe["qujax"]["enabled"] and not (
use_accelerated and candidate == preference
):
trace_chain.append("qujax_unavailable")
last_reason = "qujax disabled or unavailable"
continue
try:
selected = _QuJaxSamplingBackend(circuit, seed, details=details)
if candidate != preference and last_reason is not None:
selected.metadata.fallback_reason = last_reason
return selected
except RuntimeError as exc:
trace_chain.append("qujax_fallback")
last_reason = f"qujax sample path failed: {exc}"
continue
if candidate == "cudaq":
if not probe["cudaq"]["enabled"] and not (
use_accelerated and candidate == preference
):
trace_chain.append("cudaq_unavailable")
last_reason = "cudaq disabled or unavailable"
continue
try:
selected = _CudaQSamplingBackend(circuit, seed, details=details)
if candidate != preference and last_reason is not None:
selected.metadata.fallback_reason = last_reason
return selected
except RuntimeError as exc:
trace_chain.append("cudaq_fallback")
last_reason = f"cudaq sample path failed: {exc}"
continue
final_backend = _StimSamplingBackend(
circuit,
seed,
details={
"selected_backend": "stim",
"fallback_reason": last_reason or "all candidates unavailable",
"backend_chain": trace_chain + ["stim"],
"sample_trace_id": sample_trace_id,
"protocol_metadata": protocol_metadata,
},
)
_apply_stim_fallback_metadata(
final_backend.metadata,
trace_chain=trace_chain,
use_accelerated=use_accelerated,
last_reason=last_reason,
final_fallback=True,
)
return final_backend
|