Repository navigation
Execute scalar recipes on encoded sequential tapes - #3098
Merged
Mister-Meeseeks merged 1 commit intoOct 17, 2026
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
I execute the inherited ADD/SCALE recipes over complete aligned Gaussian packets with encoded widths/counts and fixed local word roles. I preserve borrowed operands, charge logical-zero scratch initialization and disposal, and check source/scratch return before appending each completed component.
I give a quantified sequential-tape refinement in
research/physical-scalar-stream/PROOF.md. For V=2hN and I recipe instructions per component, I obtain O((I+1)V) time and O(h+log N) local scratch, plus protected output. I include input loads, private source copies, counter setup, result replacement, rewind and erasure. I preserve literal intermediate guards, including NEG before HALF and target-walk overflow.I passed eight registered cold commands on Python 3.12.13/macOS arm64 by 00:20 UTC, using current main's updated runner: two-provider source compilation; all 31 literal templates under three policies on 84-bit Gaussian packets; 4,440 small signed prefix cases; 612 shift cases; later-field rejection/reuse; nested parent handoffs; and eight fault controls. Prefix/failure checks also passed under
-O.I coordinated the access/lifetime scope with Nick Petalas's Boolean-caller work in #3094. I retain his #3078 recipes and #3090 conditional capacity, Sagar Thakrar's #696 lowering, James Chang's #1869 ripple interface, #3091's signed tape arithmetic and the complete native/geometry/tape lineage in the unchanged archives. Jose Manuel (
jmbarrancoml), with substantial OpenAI Codex assistance, including analysis subagents.The 7,072 saving remains a logical real-instruction count. I have not proved a corresponding bit-time saving, supplied actual family alignment/routing or completed the caller. I leave the selected frontier unchanged and claim no independent human or formal verification.
I started from d8e0f83 and advanced to current main
18aacb215fc3717b06be8ceb9fca62aeca0d3e99before publication; priority6, selection and pinned scientific inputs remained unchanged.Canonical result: #2238 (comment)
Peer coordination: #2612 (reply in thread)