D-HOLO: address-indexed algebra falsifiers H1, H5, H6 (measured); H6b, H8, H9 pre-registered by AdaWorldAPI · Pull Request #1550 · AdaWorldAPI/lance-graph

AdaWorldAPI · 2026-10-14T16:43:38Z

Research PR. It adds three probes (as examples only), their results, and three further pre-registered falsifiers. There is no production code, no new opcode and no executor.

Plan: .claude/plans/d-holo-address-indexed-algebra-v1.md. Board entry: .claude/board/entries/2026-10-14-d-holo-falsifiers.md.

Measured

All runs are on v4 (avx512f), with the tier printed by the program. Every arm matches a row-at-a-time oracle in every cell, and the mask-risc arms allocate nothing per query. 15 disable runs, all red.

After the Codex review, H1 and H5 were corrected and re-run (30fb38dd, 91ca4359): lowering recognition left the timed path, and H5 now charges the construction of its input masks. The verdicts below are the corrected ones; the plan keeps the first results too.

H1, K thought branches: mixed. The row-word arm stays flat in K (0.58–1.08× from K = 1 to 64). With construction spread over 100 queries, it beats "gate once, then AND per branch" only on the 4,096-row aperture. On the full tile, gate-once is faster (47.2 vs 54.9 µs). The verdict depends on whether coverage is stored per row; if per-thought planes are stored, gate-once wins.
H5, coordinate-native vs coordinate-crossing: confirmed at N = 100, not at N = 1.
- A prefix Range reads only the prefix's words. Charged with building the occupancy plane (about 110 µs), it is 31–37× faster than a key sweep at N = 100, but only 1.3–1.4× at N = 1, below the 5× rule.
- The 1 ns row-word popcount holds only when row words are the stored form.
- On a permuted key lane, Range disagrees with the oracle, so it needs a layout-order witness.
- The per-group plane inverse does not pay for itself on prefix-restricted queries within 1,000 queries; CSR does.
H6, closure of address restrictions: holds. Interval sets stay closed under ∩, ∪ and ¬. Their cost is flat as box area grows 256×, and they were the cheapest arm in all 24 cells. Quack's fold_ranges already covers the ∩ case.

On this VM, run-to-run noise reaches about 40 %; differences under 2× are reported as undecided.

Pre-registered, not run

H6b: canonical product form, Morton intervals, dense 8 KiB masks and a hybrid, compared on the Context × Operation reading of the u8:u8 tile. That reading is a contextual one, not a new ISA schema. The product form can exceed the dense mask (256 groups need 16 KiB), and the probe measures that rather than assuming it away.
H8: logical algebra, physical realization and semantic leakage, compared separately.
H9: relational emptiness decided without a fold. The pass condition is zero false empties.

The terminal representation selector is approved as a concept and held until H6b reports.

Also in this PR

A separate commit records the merged #1548 (D-NALP-4) in PR_ARC_INVENTORY and LATEST_STATE.

Related: #1549 (D-ORACLE-0, epistemic oracle fidelity). It is cross-referenced here and deliberately not combined with this work.

🤖 Generated with Claude Code

https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

…bra) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

…re any run) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

…al popcount, snapshot) with row oracle Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

…nt, perspectives, reverse memberOf with inverse build/update cost) with row oracle Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

Three arms evaluate four formulas over Morton boxes and a prefix on the dense tile, each checked against a row oracle: S on interval endpoints, M as Range ops in mask-risc, V as GeI32/LtI32 over x/y lanes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

H1 mixed: the row-word arm is flat in K but beats gate-once only with row-stored coverage and a narrow gate. H5 confirmed: prefix Range reads only its words, 292-573x faster than a sweep; the plane inverse does not pay back on prefix-restricted queries within 1,000 queries. H6 holds: interval sets stay closed under intersection, union and complement and were the cheapest arm in all 24 cells. Also fixes three clippy lints in the H1/H5 probes; H1 re-run, same verdicts. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

PR_ARC_INVENTORY and LATEST_STATE entries for #1548; D-NALP-4 row and the isa-truth entry status move from in-PR to merged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

H6b compares product normal form, Morton intervals, dense masks and a hybrid on the Context x Operation reading of the u8:u8 tile, including the correction that the product form can exceed the 8 KiB dense mask. H8 separates logical algebra, physical realization and semantic leakage. H9 tests relational emptiness decided without a fold (zero false empties). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

coderabbitai · 2026-10-14T16:43:49Z

Important

🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: Organization UI
Review profile: CHILL
Plan: Advanced
Run ID: 1d623516-41e5-4f34-8d66-958e57845292

Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

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.

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

chatgpt-codex-connector · 2026-10-14T18:31:01Z

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-14T18:33:46.467772Z	`21c110b`	Draft marked ready

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

coderabbitai · 2026-10-14T18:31:13Z

Add Carrot credits or activate Agent usage billing to use Autofix

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 21c110b424

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".

Codex review on #1550. H1 and H5 now store CompiledProgram and time only execution; Q3 also stops rebuilding its scan program per group. H5 charges the occupancy plane (packed from a presence lane) to both Q1 arms and reports totals at N = 1 and N = 100; Q2 measures both coverage orientations and reports totals for each. H1 measures row words built from planes, so both orientations are charged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

After the review fixes in 30fb38d. H5 Q1 holds at N = 100 (31-37x) but not at N = 1 (1.3-1.4x): building the occupancy plane dominates both arms. Q2's 1 ns holds only with row words stored. H1's verdict is unchanged; gate-once got faster without the repeated compile. The first results stay in the plan; the corrected ones are a new dated section. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

…ereg D-HOLO: pre-register H11 and H12, add connectedness to H6b; record #1550

claude added 8 commits October 14, 2026 16:39

plan: pre-register D-HOLO falsifiers H1, H5, H6 (address-indexed alge…

d2a9323

…bra) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

plan: amend D-HOLO pre-registration with the amortization count (befo…

5f63b88

…re any run) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

probe: D-HOLO H1 K-branch evaluation (per-branch, gate-once, position…

011ca97

…al popcount, snapshot) with row oracle Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

probe: D-HOLO H5 coordinate-native vs coordinate-crossing (prefix cou…

d84afc1

…nt, perspectives, reverse memberOf with inverse build/update cost) with row oracle Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

AdaWorldAPI marked this pull request as ready for review October 14, 2026 18:30

chatgpt-codex-connector Bot reviewed Oct 14, 2026

View reviewed changes

Comment thread crates/lance-graph-mask-risc/examples/holo_coord_probe.rs

Comment thread crates/lance-graph-mask-risc/examples/holo_k_branch_probe.rs Outdated

claude added 2 commits October 14, 2026 18:37

AdaWorldAPI merged commit 5cbb3fd into main Oct 14, 2026
18 checks passed

AdaWorldAPI mentioned this pull request Oct 14, 2026

D-HOLO: pre-register H11 and H12, add connectedness to H6b; record #1550 #1551

Merged

AdaWorldAPI added a commit that referenced this pull request Oct 14, 2026

Merge pull request #1551 from AdaWorldAPI/ccr-9f3b8075-74u47l-holo-pr…

9337da4

…ereg D-HOLO: pre-register H11 and H12, add connectedness to H6b; record #1550

Navigation Menu

Repository navigation

D-HOLO: address-indexed algebra falsifiers H1, H5, H6 (measured); H6b, H8, H9 pre-registered - #1550

D-HOLO: address-indexed algebra falsifiers H1, H5, H6 (measured); H6b, H8, H9 pre-registered#1550
AdaWorldAPI merged 10 commits into
mainfrom
ccr-9f3b8075-74u47l-holo

AdaWorldAPI commented Oct 14, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot commented Oct 14, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector Bot commented Oct 14, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot commented Oct 14, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

AdaWorldAPI commented Oct 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Measured

Pre-registered, not run

Also in this PR

Uh oh!

coderabbitai Bot commented Oct 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

chatgpt-codex-connector Bot commented Oct 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codex Review Summary

Uh oh!

coderabbitai Bot commented Oct 14, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

AdaWorldAPI commented Oct 14, 2026 •

edited

Loading

coderabbitai Bot commented Oct 14, 2026 •

edited

Loading

chatgpt-codex-connector Bot commented Oct 14, 2026 •

edited

Loading