fix: metrics layer cleanliness + matrix.py doc corrections by erinepshovel-code · Pull Request #43 · The-Interdependency/edcmbone

erinepshovel-code · 2026-05-12T19:35:34Z

Addresses issues #38 and #39.

Changes

compute.py (issue #39)

Removed unused prev_energy parameter from energy_step and updated the one call site in compute_round accordingly. The parameter was never read inside the function body.
Fixed _compute_D docstring: was "low bone density as proxy", now accurately describes the 1 - cosine_sim implementation.
Removed unused imports: Counter, math, and BoneToken.
Changed RoundMetrics.__init__ defaults for round_index, token_count, and bone_count from 0.0 (float) to 0 (int).

projection.py (issue #39)

Removed dead variable counts = Counter(t.token_count for t in turns) in gini_tbf — it was built and immediately discarded; the real work uses speaker_counts.
Added elif direction == "below" branch and a ValueError for unknown directions in fire_alerts, so a future "below" threshold fires correctly and any unexpected direction value raises immediately rather than silently not firing.

risk.py (issue #39)

Wrapped broken_return's return value in clamp() for consistency with every other risk function. The output is already in [0, 1] by construction, so this is a no-op safety wrapper.

matrix.py (issue #38)

Updated the κ (kappa) docstring: was "≥ 0, unbounded", now correctly states κ ∈ [0, 1] with a reference to energy_step() clamping and PROJECTION_MAP["DA"] requiring it.
Added a "Note on A_MATRIX vs compute.py" section to the module docstring explaining that A_MATRIX is currently documentation-shaped (actual weights are hardcoded in compute.py) and should eventually become the single source of truth.
Added # NOTE: E depends on Layer 1 metric R — Layer mixing; see issue #38 comment on the A_MATRIX["metrics"]["E"] entry.

Test plan

Run existing metric unit tests: pytest backend/tests/metrics/
Verify energy_step call sites pass only (prev_kappa, dissonance, alpha, delta_max) — no prev_energy.
Confirm fire_alerts with a "below" direction spec fires as expected.
Confirm broken_return return value is still in [0, 1] (clamp is a no-op but now explicit).

Closes #38
Closes #39

Generated by Claude Code

- compute.py: remove unused prev_energy param from energy_step and its call site; fix _compute_D docstring to reflect 1-cosine_sim impl; remove unused Counter/math/BoneToken imports; change RoundMetrics round_index/token_count/bone_count defaults from 0.0 to 0 (int) - projection.py: remove dead counts=Counter(...) variable in gini_tbf; add elif/ValueError for unknown direction in fire_alerts - risk.py: wrap broken_return return value in clamp() for consistency - matrix.py: update kappa docstring to kappa in [0,1]; add note that A_MATRIX is documentation-shaped and should become single source of truth; add Layer-mixing comment on E entry Closes #38, #39

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7c5e2da8aa

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Copilot

Pull request overview

This PR performs a small “cleanliness” pass across the metrics layer and corrects documentation around κ (kappa) and the metric matrix/projection definitions, addressing issues #38 and #39.

Changes:

Simplifies the RC-circuit helper by removing an unused prev_energy argument from energy_step() and updates the call site accordingly.
Removes dead/unused code in projection/risk helpers and hardens alert firing by supporting "below" thresholds and raising on unknown directions.
Updates matrix.py documentation to correctly describe κ ∈ [0, 1] and clarifies that A_MATRIX is currently documentation-only (not used by compute.py).

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
backend/src/edcmbone/metrics/compute.py	Removes unused `energy_step` parameter, cleans imports, adjusts defaults, and fixes `_compute_D` docstring.
backend/src/edcmbone/metrics/projection.py	Removes dead variable in `gini_tbf` and adds `"below"` + validation handling in `fire_alerts`.
backend/src/edcmbone/metrics/risk.py	Wraps `broken_return` output (and formula doc) in `clamp()` for consistency.
backend/src/edcmbone/metrics/matrix.py	Corrects κ range documentation and adds notes clarifying the current `A_MATRIX` vs implementation divergence.

Comments suppressed due to low confidence (1)

backend/src/edcmbone/metrics/compute.py:131

The module docstring still states stored tension k is "≥ 0" (unbounded), but energy_step() clamps kappa via clamp(...) which restricts it to [0, 1] (and matrix.py now documents κ ∈ [0, 1]). Please update the module docstring’s k/kappa range to match the implementation.

def energy_step(prev_kappa, dissonance, alpha=0.85, delta_max=0.3):
    """Compute one step of the RC-circuit energy model.

    s_{t+1} = alpha * s_t + E_t - delta_t
    delta_t  = min(delta_max, g(y_t, y_{t-1}))

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

energy_step(0.0, 0.0, dissonance=0.9, ...) passed prev_energy as the second positional arg, which now maps to dissonance — causing TypeError on the keyword duplicate. Drop the extra 0.0; new call is energy_step(0.0, 0.9, alpha=0.9, delta_max=0.1). Addresses Codex review comment on #43.

@codex

- Remove prev_energy=0.0 from compute_round signature and docstring - Remove prev_energy plumbing from compute_transcript (init + kwarg pass) - Module docstring: update k description from '>= 0' to 'kappa in [0,1]' to match the clamp() applied in energy_step Addresses Copilot review comment on PR #43 and erinepshovel-code @codex repair request (line 127).

…script plumbing Agent-Logs-Url: https://github.com/The-Interdependency/edcmbone/sessions/b891bdfe-243e-4c21-a39d-ba7b085c021f Co-authored-by: erinepshovel-code <250928284+erinepshovel-code@users.noreply.github.com>

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4294387457

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-05-13T01:42:01Z

 def compute_round(round_, prev_round=None, canon=None,
                  alpha=0.85, delta_max=0.3,
-                  prev_kappa=0.0, prev_energy=0.0, prev_entropy=0.0):
+                  prev_kappa=0.0, prev_entropy=0.0):
    """Compute the metric vector for a Round.


Keep compute_round compatible with prev_energy callers

Removing prev_energy from compute_round is a backward-incompatible API change for existing integrations that still pass prev_energy= (the parameter existed in the prior signature and this function is exported via edcmbone.metrics). In this commit those calls now fail immediately with TypeError: unexpected keyword argument 'prev_energy', even though the value was previously ignored, so preserving it as an optional no-op argument would avoid breaking downstream code.

Useful? React with 👍 / 👎.

@codex[agent] none downstream yet

Summary

Restored backward compatibility in compute_round by reintroducing an optional prev_energy parameter so downstream callers that still pass prev_energy= no longer fail with TypeError. The value is accepted as a no-op.

Updated the compute_round docstring to explicitly document prev_energy as a deprecated compatibility argument that is accepted but ignored.

Committed the fix on the current branch as commit 2735ef4 and created a follow-up PR message via make_pr per instructions.

Testing

✅ python -m py_compile backend/src/edcmbone/metrics/compute.py

View task →

erinepshovel-code · 2026-05-13T01:44:27Z

@copilot apply changes based on the comments in this thread

erinepshovel-code marked this pull request as ready for review May 13, 2026 01:26

Copilot AI review requested due to automatic review settings May 13, 2026 01:26

Copilot started reviewing on behalf of erinepshovel-code May 13, 2026 01:26 View session

chatgpt-codex-connector Bot reviewed May 13, 2026

View reviewed changes

Comment thread backend/src/edcmbone/metrics/compute.py

Copilot AI reviewed May 13, 2026

View reviewed changes

Comment thread backend/src/edcmbone/metrics/compute.py

Copilot started work on behalf of erinepshovel-code May 13, 2026 01:36 View session

erinepshovel-code and others added 2 commits May 12, 2026 18:38

Copilot finished work on behalf of erinepshovel-code May 13, 2026 01:39

chatgpt-codex-connector Bot reviewed May 13, 2026

View reviewed changes

Codex started work on behalf of erinepshovel-code May 13, 2026 01:44 View session

Codex finished work on behalf of erinepshovel-code May 13, 2026 01:46

Copilot started work on behalf of erinepshovel-code May 13, 2026 01:46 View session

erinepshovel-code merged commit f7a156d into main May 13, 2026
3 of 4 checks passed

erinepshovel-code deleted the claude/fix-metrics-layer branch May 13, 2026 01:47

Copilot stopped work on behalf of erinepshovel-code due to an error May 13, 2026 01:47
The session was cancelled by the user.

Navigation Menu

Uh oh!

fix: metrics layer cleanliness + matrix.py doc corrections - #43

fix: metrics layer cleanliness + matrix.py doc corrections#43
erinepshovel-code merged 4 commits into
mainfrom
claude/fix-metrics-layer

erinepshovel-code commented May 12, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot May 13, 2026

Uh oh!

erinepshovel-code May 13, 2026

Uh oh!

chatgpt-codex-connector Bot May 13, 2026

Uh oh!

erinepshovel-code commented May 13, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

erinepshovel-code commented May 12, 2026

Changes

compute.py (issue #39)

projection.py (issue #39)

risk.py (issue #39)

matrix.py (issue #38)

Test plan

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot May 13, 2026

Choose a reason for hiding this comment

Uh oh!

erinepshovel-code May 13, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot May 13, 2026

Choose a reason for hiding this comment

Summary

Uh oh!

erinepshovel-code commented May 13, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants