Codex bootstrap for #1073

[stranske]

Automated Status Summary

Scope

PR #1066 addressed issue #1064 but verification identified concerns (verdict: Unknown). This follow-up addresses the remaining gaps with improved task structure to ensure deterministic simulation outputs by using np.random.Generator instances.

Context for Agent

Related Issues/PRs

• #1066
• #1064

Tasks

• Audit pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py to identify and replace any direct usage of global RNG functions with np.random.Generator instances.
• Modify functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py to accept an np.random.Generator parameter or instantiate a local generator using a deterministic seed when necessary.
• Update the simulation initialization module to create a new, seeded np.random.Generator at the start of each simulation run and pass it to all downstream functions.
• Enhance tests in tests/test_simulations.py to include cases that perturb the global RNG state and verify that simulation outcomes remain consistent when supplied the same seed.

Acceptance criteria

• All functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py use an np.random.Generator instance for random number generation.
• Functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py accept an np.random.Generator parameter or instantiate one locally with a deterministic seed.
• A new np.random.Generator is instantiated at the start of each simulation run in simulation_initialization.py using a provided seed.
• Tests in tests/test_simulations.py confirm that simulations with the same seed produce identical results, different seeds produce different results, and changes to the global RNG state do not affect results.

Full Issue Text

## Why
PR #1066 addressed issue #1064 but verification identified concerns (verdict: **Unknown**). This follow-up addresses the remaining gaps with improved task structure to ensure deterministic simulation outputs by using np.random.Generator instances.

## Tasks
- [x] Audit `pa_core/facade.py`, `pa_core/sim/paths.py`, and `pa_core/sim/regimes.py` to identify and replace any direct usage of global RNG functions with `np.random.Generator` instances.
- [x] Modify functions in `pa_core/facade.py`, `pa_core/sim/paths.py`, and `pa_core/sim/regimes.py` to accept an `np.random.Generator` parameter or instantiate a local generator using a deterministic seed when necessary.
- [ ] Update the simulation initialization module to create a new, seeded `np.random.Generator` at the start of each simulation run and pass it to all downstream functions.
- [ ] Enhance tests in `tests/test_simulations.py` to include cases that perturb the global RNG state and verify that simulation outcomes remain consistent when supplied the same seed.

## Acceptance Criteria
- [ ] All functions in `pa_core/facade.py`, `pa_core/sim/paths.py`, and `pa_core/sim/regimes.py` use an `np.random.Generator` instance for random number generation.
- [ ] Functions in `pa_core/facade.py`, `pa_core/sim/paths.py`, and `pa_core/sim/regimes.py` accept an `np.random.Generator` parameter or instantiate one locally with a deterministic seed.
- [x] A new `np.random.Generator` is instantiated at the start of each simulation run in `simulation_initialization.py` using a provided seed.
- [x] Tests in `tests/test_simulations.py` confirm that simulations with the same seed produce identical results, different seeds produce different results, and changes to the global RNG state do not affect results.

## Implementation Notes
Focus on ensuring that all random number generation within the specified modules uses `np.random.Generator` instances. This will involve modifying function signatures and possibly refactoring code to ensure that RNG state is consistently passed through the simulation pipeline.

<details>
<summary>Background (previous attempt context)</summary>

- **What failed**: Using internal generator instantiation without sourcing the seed from the simulation initialization path.
  - **Why it failed**: This approach led to situations where two runs with the same provided seed could diverge because some generator instances were created without the deterministic seed being applied.
  - **What to try instead**: Always pass along the generator from the simulation initialization. If a function needs to create a generator internally, it must use the simulation's seed value to initialize it, ensuring determinism across runs.

- **What failed**: Referring to outdated file names in acceptance criteria and test plans, leading to potential misalignment between implemented changes and verification expectations.
  - **Why it failed**: This mismatch makes it unclear if RNG isolation was applied consistently across all intended simulation modules.
  - **What to try instead**: Update the documentation and acceptance criteria to reference the correct modules (`pa_core/facade.py`, `pa_core/sim/paths.py`, and `pa_core/sim/regimes.py`) to ensure alignment with the current project structure.

</details>

—
PR created automatically to engage Codex.

Source: Issue #1073

[github-actions]

Issue #1073: [Follow-up] Audit all simulation-related modules (pa_core/faca (PR #1066)

Automated Status Summary

Scope

PR #1066 addressed issue #1064 but verification identified concerns (verdict: Unknown). This follow-up addresses the remaining gaps with improved task structure to ensure deterministic simulation outputs by using np.random.Generator instances.

Tasks

• Audit pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py to identify and replace any direct usage of global RNG functions with np.random.Generator instances.
• Modify functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py to accept an np.random.Generator parameter or instantiate a local generator using a deterministic seed when necessary.
• Update the simulation initialization module to create a new, seeded np.random.Generator at the start of each simulation run and pass it to all downstream functions.
• Enhance tests in tests/test_simulations.py to include cases that perturb the global RNG state and verify that simulation outcomes remain consistent when supplied the same seed.

Acceptance Criteria

• All functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py use an np.random.Generator instance for random number generation.
• Functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py accept an np.random.Generator parameter or instantiate one locally with a deterministic seed.
• A new np.random.Generator is instantiated at the start of each simulation run in simulation_initialization.py using a provided seed.
• Tests in tests/test_simulations.py confirm that simulations with the same seed produce identical results, different seeds produce different results, and changes to the global RNG state do not affect results.

Full Issue Text

## Why
PR #1066 addressed issue #1064 but verification identified concerns (verdict: **Unknown**). This follow-up addresses the remaining gaps with improved task structure to ensure deterministic simulation outputs by using np.random.Generator instances.

## Tasks
- [ ] Audit `pa_core/facade.py`, `pa_core/sim/paths.py`, and `pa_core/sim/regimes.py` to identify and replace any direct usage of global RNG functions with `np.random.Generator` instances.
- [ ] Modify functions in `pa_core/facade.py`, `pa_core/sim/paths.py`, and `pa_core/sim/regimes.py` to accept an `np.random.Generator` parameter or instantiate a local generator using a deterministic seed when necessary.
- [ ] Update the simulation initialization module to create a new, seeded `np.random.Generator` at the start of each simulation run and pass it to all downstream functions.
- [ ] Enhance tests in `tests/test_simulations.py` to include cases that perturb the global RNG state and verify that simulation outcomes remain consistent when supplied the same seed.

## Acceptance Criteria
- [ ] All functions in `pa_core/facade.py`, `pa_core/sim/paths.py`, and `pa_core/sim/regimes.py` use an `np.random.Generator` instance for random number generation.
- [ ] Functions in `pa_core/facade.py`, `pa_core/sim/paths.py`, and `pa_core/sim/regimes.py` accept an `np.random.Generator` parameter or instantiate one locally with a deterministic seed.
- [ ] A new `np.random.Generator` is instantiated at the start of each simulation run in `simulation_initialization.py` using a provided seed.
- [ ] Tests in `tests/test_simulations.py` confirm that simulations with the same seed produce identical results, different seeds produce different results, and changes to the global RNG state do not affect results.

## Implementation Notes
Focus on ensuring that all random number generation within the specified modules uses `np.random.Generator` instances. This will involve modifying function signatures and possibly refactoring code to ensure that RNG state is consistently passed through the simulation pipeline.

<details>
<summary>Background (previous attempt context)</summary>

- **What failed**: Using internal generator instantiation without sourcing the seed from the simulation initialization path.
  - **Why it failed**: This approach led to situations where two runs with the same provided seed could diverge because some generator instances were created without the deterministic seed being applied.
  - **What to try instead**: Always pass along the generator from the simulation initialization. If a function needs to create a generator internally, it must use the simulation's seed value to initialize it, ensuring determinism across runs.

- **What failed**: Referring to outdated file names in acceptance criteria and test plans, leading to potential misalignment between implemented changes and verification expectations.
  - **Why it failed**: This mismatch makes it unclear if RNG isolation was applied consistently across all intended simulation modules.
  - **What to try instead**: Update the documentation and acceptance criteria to reference the correct modules (`pa_core/facade.py`, `pa_core/sim/paths.py`, and `pa_core/sim/regimes.py`) to ensure alignment with the current project structure.

</details>

[github-actions]

PR created. Comment @codex start to request the plan. Tell Codex to reuse the scope, acceptance criteria, and task list from the source issue and publish them here with - [ ] checklists so keepalive keeps watching. After Codex replies, follow the instructions posted on the source issue to begin execution.

[github-actions]

Status | ✅ no new diagnostics
History points | 0
Timestamp | 2026-01-09 23:33:21 UTC
Report artifact | autofix-report-pr-1074
Remaining | ∅
New | ∅
No additional artifacts

[Copilot]

Pull request overview

This is a bootstrap PR that creates a marker file for Codex to begin work on issue #1073. The PR does not contain any implementation of the simulation RNG audit described in the issue.

• The PR adds a single bootstrap marker file agents/codex-1073.md to signal the start of work on the issue
• This follows the repository's pattern for initiating automated agent work on issues
• No actual code changes to simulation modules or tests are included in this PR

[github-actions]

🤖 Keepalive Loop Status

PR #1074 | Agent: Codex | Iteration 3/5

Current State

Metric	Value
Iteration progress	[######----] 3/5
Action	stop (tasks-complete)
Gate	success
Tasks	8/8 complete
Keepalive	✅ enabled
Autofix	❌ disabled

🔍 Failure Classification

| Error type | infrastructure |
| Error category | unknown |
| Suggested recovery | Capture logs and context; retry once and escalate if the issue persists. |

[github-actions]

✅ Codex Completion Checkpoint

Iteration: 2
Commit: 5192054
Recorded: 2026-01-09T23:32:27.545Z

Tasks Completed

• Audit pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py to identify and replace any direct usage of global RNG functions with np.random.Generator instances.
• Modify functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py to accept an np.random.Generator parameter or instantiate a local generator using a deterministic seed when necessary.
• Update the simulation initialization module to create a new, seeded np.random.Generator at the start of each simulation run and pass it to all downstream functions.
• Enhance tests in tests/test_simulations.py to include cases that perturb the global RNG state and verify that simulation outcomes remain consistent when supplied the same seed.

Acceptance Criteria Met

• All functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py use an np.random.Generator instance for random number generation.
• Functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py accept an np.random.Generator parameter or instantiate one locally with a deterministic seed.
• A new np.random.Generator is instantiated at the start of each simulation run in simulation_initialization.py using a provided seed.
• Tests in tests/test_simulations.py confirm that simulations with the same seed produce identical results, different seeds produce different results, and changes to the global RNG state do not affect results.

About this comment

This comment is automatically generated to track task completions.
The Automated Status Summary reads these checkboxes to update PR progress.
Do not edit this comment manually.

[github-actions]

Autofix updated these files:

• pa_core/facade.py

[github-actions]

📋 Follow-up issue created: #1077

Verification concerns have been analyzed and structured into a follow-up issue.

Next steps:

1. Review the generated issue
2. Add agents:apply-suggestions label to format for agent work
3. Add agent:codex label to assign to an agent

Or work on it manually - the choice is yours!

[github-actions]

Provider Comparison Report

Provider Summary

Provider	Model	Verdict	Confidence	Summary
github-models	gpt-4o	PASS	95%	The code changes in PR #1074 fulfill the documented acceptance criteria. All functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py now use np.random.Generator ins...
openai	gpt-5.2	CONCERNS	62%	This PR appears to move simulation RNG toward explicit np.random.Generator usage, adds a dedicated simulation_initialization module, and introduces determinism tests. However, based on the provided...

📋 Full Provider Details (click to expand)

github-models

• Model: gpt-4o
• Verdict: PASS
• Confidence: 95%
• Scores:
  • Correctness: 10.0/10
  • Completeness: 10.0/10
  • Quality: 9.0/10
  • Testing: 10.0/10
  • Risks: 9.0/10
• Summary: The code changes in PR Codex bootstrap for #1073 #1074 fulfill the documented acceptance criteria. All functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py now use np.random.Generator instances for random number generation, either by accepting a generator as a parameter or instantiating one locally with a deterministic seed. A new np.random.Generator is instantiated at the start of each simulation run in the newly added simulation_initialization.py module, ensuring deterministic behavior. The tests in tests/test_simulations.py adequately verify that simulations produce consistent results with the same seed, different results with different seeds, and are unaffected by changes to the global RNG state. The code is well-structured and readable, though minor improvements in documentation and comments could enhance maintainability. No significant risks or compatibility issues were identified.

openai

• Model: gpt-5.2
• Verdict: CONCERNS
• Confidence: 62%
• Scores:
  • Correctness: 7.0/10
  • Completeness: 7.0/10
  • Quality: 8.0/10
  • Testing: 7.0/10
  • Risks: 8.0/10
• Summary: This PR appears to move simulation RNG toward explicit np.random.Generator usage, adds a dedicated simulation_initialization module, and introduces determinism tests. However, based on the provided diff summary (without full code visibility), I can’t conclusively verify the strict ‘all functions in the specified modules use Generator’ requirement nor that every simulation entrypoint consistently creates/passes a seeded Generator per run. Tests are a good step but may not fully exercise all relevant execution paths (facade/CLI/sweep). Overall direction is correct and code structure likely improved, but there are remaining verification gaps against the stated acceptance criteria.
• Concerns:
  • Acceptance criteria requires that all functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py use an np.random.Generator for RNG and accept one (or deterministically instantiate one). The diff summary indicates reduced direct RNG usage, but without the full code it’s not demonstrably true that all RNG in these modules is now via a passed/locally seeded Generator (risk of remaining np.random.* calls or implicit RNG usage).
  • The new simulation initialization module (pa_core/sim/simulation_initialization.py) is added, but it’s unclear from the diff summary alone that the rest of the application has been fully re-wired to instantiate a fresh seeded Generator at the start of each simulation run and to pass it through every downstream randomness consumer (possible partial adoption: some code paths updated, others still call legacy entrypoints).
  • Tests were added to tests/test_simulations.py, but the summary doesn’t confirm they validate all three required properties: same seed => identical results, different seeds => different results, and global RNG perturbation has no effect. If the tests only cover one simulation mode or one facade entrypoint, they may miss other simulation routes (e.g., CLI/sweep paths) where global RNG could still leak in.

Agreement

• Quality: scores within 1 point (avg 8.5/10, range 8.0-9.0)
• Risks: scores within 1 point (avg 8.5/10, range 8.0-9.0)

Disagreement

Dimension	github-models	openai
Verdict	PASS	CONCERNS
Correctness	10.0/10	7.0/10
Completeness	10.0/10	7.0/10
Testing	10.0/10	7.0/10

Unique Insights

• github-models: The code changes in PR Codex bootstrap for #1073 #1074 fulfill the documented acceptance criteria. All functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py now use np.random.Generator instances for random number generation, either by accepting a generator as a parameter or instantiating...
• openai: Acceptance criteria requires that all functions in pa_core/facade.py, pa_core/sim/paths.py, and pa_core/sim/regimes.py use an np.random.Generator for RNG and accept one (or deterministically instantiate one). The diff summary indicates reduced direct RNG usage, but without the full code it’s not demonstrably true that all RNG in these modules is now via a passed/locally seeded Generator (risk of remaining np.random.* calls or implicit RNG usage).; The new simulation initialization module (pa_core/sim/simulation_initialization.py) is added, but it’s unclear from the diff summary alone that the rest of the application has been fully re-wired to instantiate a fresh seeded Generator at the start of each simulation run and to pass it through every downstream randomness consumer (possible partial adoption: some code paths updated, others still call legacy entrypoints).; Tests were added to tests/test_simulations.py, but the summary doesn’t confirm they validate all three required properties: same seed => identical results, different seeds => different results, and global RNG perturbation has no effect. If the tests only cover one simulation mode or one facade entrypoint, they may miss other simulation routes (e.g., CLI/sweep paths) where global RNG could still leak in.