Repository navigation
D-CTX-5 + D-CTX-6: Morton visit order; render → observation → GadamerRevision - #1367
Conversation
Round 6 of the cognitive-shader-driver proof ladder. Four visit orders over the same 16x16 tile (Morton, row-major, 4x4 tiled, reversed): the f64 scatter changes only in its last bits with the order (max |delta| <= 1.1e-15), and an i64 accumulator at 2^-32 is order-free (identical across all four orders, within 1.4e-9 of f64). Trie locality via nibble_climb: Morton = tiled = reversed = 1.0588, row-major 1.2471. The fixed render allocates nothing. 7 tests; 4 disable runs red, one equivalent mutant recorded. Follow-up to #1364 review: the boundary probe no longer calls pairs weak/strong. Fixtures are negative-code / mid-negative / positive-code pairs, and the test states that the rendered boundary response changes deterministically with the signed pair-law code; no epistemic or semantic strength is derived from the measured order. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01U5ZJLtEBxeLPKX1pLRkHhc
|
Warning Review limit reachedYour organization has reached its usage spending cap. Adjust your spending cap in the billing tab. Next included review available in 58 minutes. View limit detailsLimit details: You’ve used the included review currently available. Your 67 included PR review attempts over the past 7 days set your current allowance at 1 review per hour. Review configuration: ⚙️ Run configuration
📒 Files selected for processing (7)
Warning Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption. Comment |
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: f7deed83-e3e3-4404-a416-a67e296d5dce) |
Round 7 of the cognitive-shader-driver proof ladder closes the loop to #1344 on the D-CTX surface. Belief is an InterpretiveHorizon over [u64; 4], one bit per pixel. The D-CTX-4 rendered witness is presented as an inherited interpretation with no root and is never adopted (Reinterpretation / NoIncrease); an observation reads only the resident palette and is one new independent root (HorizonExpansion or IndependentConfirmation). delta.resulting is adopted only on IncreaseEligible. Measured: two Fisher-Z laws render different fields and give identical final belief; replay is identical and re-presenting an observation is an Echo; a loop step allocates 0 B. The steepest rendered pixel on the split tile is one pixel off the material boundary, and render-guided search is not reliably better than a row-major scan (24 vs 34 on the split tile, 35 vs 34 on quadrants), so the order is pinned, not claimed useful. 7 tests, 5 disable runs red. strongest_boundary moves to support/boundary.rs with an address-predicate variant; boundary_measure_probe imports it unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01U5ZJLtEBxeLPKX1pLRkHhc
AdaWorldAPI
changed the title
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: 883938bb-5cda-4f8e-ab24-769413d9a729) |
|
Add Carrot credits or activate Agent usage billing to use Autopilot |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3d5a08cc74
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".
| let truth = boundary_oracle(tile).intersection(&inner_mask()); | ||
| let (mut h, mut steps, mut covered_at) = (prior(), 0, 0); | ||
| while let Some((next, _, _)) = step(tile, &field, &h) { |
There was a problem hiding this comment.
When a fixture has no true boundary in the inner region—such as the uniform tile and the split at x = 2—truth.is_subset_of(...) is already true before any observation, but episode checks it only after the first step and therefore reports covered_at = 1. row_major_coverage makes the same post-visit check, so the pinned (1, 1) results incorrectly claim that one observation was required to cover an empty truth set; initialize this case as covered at step zero (and return zero from the row-major comparison) so the experiment's coverage metric is accurate.
Useful? React with 👍 / 👎.
2 participants
Two commits, one per round. Each can be reviewed on its own.
Round 6 — D-CTX-5: does the visit order change the render?
examples/morton_order_probe.rsrenders the same 16×16 tile in four visit orders: Morton (the D-CTX-2 order), row-major, 4×4 tiled and reversed.f64scatteri64accumulator at 2⁻³²f64nibble_climb)The order is a locality choice, not a semantic one. If bit-exact order independence is ever required, the fix is the fixed-point accumulator, not a particular order.
7 tests. 4 disable runs are red; the 5th mutant is equivalent and cannot be caught.
Follow-up to the #1364 review:
boundary_response_changes_with_the_signed_pair_code;Round 7 — D-CTX-6: render → measurement → observation →
GadamerRevision→ replayexamples/observation_revision_probe.rscloses the loop to #1344.Belief state:
InterpretiveHorizon<(), [u64; 4]>, one bit per pixel:Encounters:
Reinterpretation/NoIncrease, never adoptedHorizonExpansionorIndependentConfirmation;delta.resultingadopteddelta.resultingis adopted only onIncreaseEligible.revision_index= steps; re-presenting every observation isEchoMeasured, not claimed useful:
7 tests. 5 disable runs are red:
strongest_boundarymoves tosupport/boundary.rswith an address-predicate variant.boundary_measure_probeuses it unchanged and its 7 tests are green.Board: D-CTX-4 → Shipped (#1364); D-CTX-5 and D-CTX-6 → In PR.
🤖 Generated with Claude Code
https://claude.ai/code/session_01U5ZJLtEBxeLPKX1pLRkHhc