You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: .agents/memory/workflow_and_quality.md
+1Lines changed: 1 addition & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -40,6 +40,7 @@ Use this file for changes that touch public API, tests, docs, or code quality.
40
40
- Benchmark external libraries in isolated environments and match precision/settings before making performance claims.
41
41
- Library plotting helpers should accept an optional `ax` and should not call `plt.show()` internally.
42
42
- In read-only or sandboxed environments, redirect caches such as `NUMBA_CACHE_DIR`, `MPLCONFIGDIR`, and mypy's cache directory in the shell command or local harness used for validation, not in tracked source files, unless the repo explicitly needs that behavior.
43
+
- If a test or import failure plausibly comes from sandbox restrictions or third-party runtime cache/process-pool setup, rerun the exact command with escalated permissions before attributing the failure to TensorCircuit logic; sandbox artifacts can mask the real failure mode.
43
44
- If `cotengra` fails at import time with `ImportError: cannot import name 'get_namespace' from autoray`, treat it as an environment dependency mismatch and upgrade `autoray` before debugging TensorCircuit's contraction code.
44
45
- On macOS sandboxed runs, `conda run -n <env> ...` can silently resolve to the base interpreter; verify `sys.executable` or call the env's Python directly when reproducing environment-sensitive issues.
45
46
- Cotengra hyper-optimization that reaches `joblib`/`loky` process-pool setup can fail inside the sandbox with `PermissionError` from `os.sysconf("SC_SEM_NSEMS_MAX")`; for end-to-end benchmark validation, rerun the exact benchmark command with escalated permissions rather than changing TensorCircuit logic.
Each script is named according to the whitepaper section it belongs to and is fully executable, demonstrating the power, flexibility, and performance of the framework.
This directory contains high-fidelity, reproducible simulations of key quantum computing and quantum information papers using **TensorCircuit-NG**. Each subfolder represents a specific paper reproduction.
4
+
5
+
## Directory Structure
6
+
7
+
For any new paper reproduction, follow the standardized naming and folder structure:
0 commit comments