Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 18 additions & 1 deletion docs/ci-verification.md
Original file line number Diff line number Diff line change
@@ -1,8 +1,25 @@
# CI verification

## Current selected-family replay: #6030
## Current selected-family replay: #7000 with #7001

The [selected record](../certificates/selected-result.json) binds conditional
κ = 0.00363005817 to the [application review](research/pr-intake-evidence/pr7000-application/PROOF-REVIEW.md),
[validation receipts](research/pr-intake-evidence/pr7000-application/VALIDATION.json),
and [explicit source closure](../certificates/frontier7000-source.json).
`make frontier7000-verify` follows that exact-source selected replay. The review
reuses eleven authenticated preceding executions and records seven fresh serial
application, scalar and complete native-catalogue executions. The earlier
monolithic timeout is retained separately and is not completed coverage.
Consult the receipts for exact commands, source heads and limits.

The five inherited framework assumptions remain. No complete uniform word,
literal multiplier execution, independent human proof or formal-kernel proof
is asserted. Routine research submissions do not need to rerun this
selected-family replay. The sections below preserve historical checkpoints.

## Historical selected-family replay: #6030

The [historical checkpoint](../certificates/checkpoints/frontier6030.json) binds conditional
κ = 0.003125804 to the [proof review](research/pr-intake-evidence/pr6030-audit/PROOF-REVIEW.md),
[validation receipts](research/pr-intake-evidence/pr6030-audit/VALIDATION.json),
and [explicit source closure](../certificates/frontier6030-source.json).
Expand Down
108 changes: 108 additions & 0 deletions research/even-one-edge-maximizer/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,108 @@
# Inequivalent even maximizers with one matrix-unit edge

At pattern dimension `k=3` and matching number `ν=1`, two complete binary
projector pools attain the maximum joint-image dimension `n=4` and have
inequivalent actions. The new pool has two unordered pairs whose product has
trace one. The published maximizer has four. This is a scoped construction over
`F₂`.

## Model and construction

Write `P=xyᵀ` with `yᵀx=1`. Its pattern is the off-diagonal part of `P`.
For a pattern space `H`, the complete pool consists of all rank-one idempotents
whose patterns belong to `H \ {0}`. An edge `{i,j}` is present when `H` contains
`Eᵢⱼ` or `Eⱼᵢ`. The matching number is the maximum number of disjoint edges.
Equivalence of actions means simultaneous conjugacy by an invertible map on the
joint image, with the projectors allowed to be relabelled.

Use coordinates `a,b,f,g`, with corresponding unit vectors denoted by the same
letters. Take these four pairs of factors.

| Projector | Image `x` | Covector `y` |
| --- | --- | --- |
| `P` | `a` | `a+b` |
| `Q` | `a+b` | `b` |
| `R` | `a+f` | `a+g` |
| `S` | `a+b+g` | `b+f` |

Each pairing `yᵀx` is one. Put `H=span(E_ab, off(R), off(S))`. The patterns of
`P` and `Q` both equal `E_ab`. Partition the coordinates into `(a,b)` and
`(f,g)`. On the first two rows and last two columns, the patterns of `R` and
`S` restrict to `a gᵀ` and `(a+b) fᵀ`. In the reverse rectangle they restrict to
`f aᵀ` and `g bᵀ`. Both pairs of restrictions are independent. Every
pattern outside `span(E_ab)` therefore has a nonzero entry in each of the two
disjoint rectangles. It follows that `k=3` and `E_ab` is the sole matrix unit.

The only lifts of `E_ab` are `P` and `Q`. Indeed, the side conditions `x_a=1`
and `y_b=1` force all the other bits except `x_b,y_a` to vanish, and exactly one
of those two bits must be one. In the coset of `off(R)` modulo `span(E_ab)`,
the two nonzero cross rectangles uniquely fix `x` and `y` over `F₂`, yielding
`R`. The coset of `off(S)` gives `S` by the same argument. In the sum coset the
first rectangle has rank two, since `a` and `a+b` are independent. It cannot
be a rectangle of an outer product. This exhausts the nonzero pattern space.

The four images form a basis because `a` and `a+b` span the first two
coordinates and the other images have distinct fresh coordinates `f,g`.
The covectors are also a basis. Every nonzero invariant subspace must therefore
contain at least one of the four images, obtained by applying a projector to a
nonzero element of the subspace. Define an arrow `i → j` if `P_j x_i = x_j`.
The closed walk `P → R → S → Q → R → P` reaches all four images from each of its
vertices. Thus the action is irreducible, with `n=4`. The even maximum at these
parameters is four by the matching bound and its attainment in #7707.

## Inequivalence

For distinct projectors `P_i=x_i y_iᵀ` and `P_j=x_j y_jᵀ`,
`tr(P_i P_j)=(y_iᵀ x_j)(y_jᵀ x_i)`. The integer count of unordered pairs
with trace one is unchanged by simultaneous conjugation and relabelling.
The pairing matrix for the new pool, with entries `y_iᵀ x_j` and order
`P,Q,R,S`, is

```text
1 0 1 0
0 1 0 1
1 1 1 0
0 1 1 1
```

The trace-one pairs are `{P,R}` and `{Q,S}`. For the published maximizer,
#7707 uses `P,Q` with the two links `(b+f)(a+f)ᵀ` and `(b+g)(a+g)ᵀ`.
Each link has trace-one product with each of `P,Q`. The two links have
trace-zero product, as do `P,Q`. Its count is four, which separates the two
actions even under arbitrary linear changes of basis on the joint image.

## Reproduce

Run these from the repository root with CPython 3.12.15 and its standard library.

```sh
python3.12 -B research/even-one-edge-maximizer/verify.py
python3.12 -B -O research/even-one-edge-maximizer/verify.py
python3.12 -B scripts/research_package.py --package research/even-one-edge-maximizer
python3.12 -B scripts/check_research_commit.py --package research/even-one-edge-maximizer --check-source-history
```

The verifier scans all 120 width-four rank-one idempotents for the new pattern
space, checks all 15 nonzero starting vectors and reports the two trace-one
pairs. The written proof uses the published upper bound as a premise.

## Open next questions

- What are all action classes at `(k,ν,n)=(3,1,4)`? Start by enumerating spans
of `E_ab` and two realized pattern classes at width four, then classify
their complete pools up to conjugacy. Higher ambient widths require a
separate reduction.
- Can the new alphabet realize a rank-saving paid word? Start by specifying
a saturated exponent relation lattice and the full ambient endpoint. A
word, dirty-state restoration, native implementation and the complete
recursive invoice remain open. This construction supplies no new saving.

## Credit

The new construction and proof were produced by Nick Petalas. Ali Romman’s
[PR #7707](https://github.com/CrocSwap/integer-mult-bounds/pull/7707) supplies
its even-maximizer question, the baseline construction and the sharp value.
His [PR #7692](https://github.com/CrocSwap/integer-mult-bounds/pull/7692)
supplies the matching theorem and the two lifts of a matrix unit used in the
baseline. [TASK.json](TASK.json) identifies the exact predecessor proofs.
The new package is Apache-2.0.
18 changes: 18 additions & 0 deletions research/even-one-edge-maximizer/SOURCES.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,18 @@
{
"schema_version": 1,
"base_commit": "e54523517d0f03e58352e647c403a5af3e525a57",
"files": [
{
"path": "certificates/selected-result.json",
"usage": "provenance",
"git_blob": "75d77b328ea02ded535e7f23aee5b1e55d0bb1a8",
"sha256": "154c3d30455d778fd3343b6b4851f0fa4665f0f7625b0adb3675e2bafbbf6bf1"
},
{
"path": "docs/research/agent-task-catalog.json",
"usage": "provenance",
"git_blob": "3404fbdd5e7529a0674eef26114b5071e4586cec",
"sha256": "ead6cdc9d0af6d58e8a49744d8ec8c7c15e13d2ba93c276c434bcf9b51b93a81"
}
]
}
89 changes: 89 additions & 0 deletions research/even-one-edge-maximizer/TASK.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,89 @@
{
"schema_version": 1,
"task_id": "bit-new-register-model",
"priority_revision": 8,
"base_commit": "e54523517d0f03e58352e647c403a5af3e525a57",
"original_card": {
"id": "bit-new-register-model",
"question": "Can an explicitly larger or nonabelian register model escape the scoped three-register obstruction?",
"starting_points": [
"docs/research/research-family-status.md",
"research/bit-three-register-lower-bound/README.md",
"certificates/selected-result.json",
"research/bit-gl3-mixed-targets-length10/README.md",
"docs/research/community-efficient1000-review.md"
],
"methods": [
"Construct the smallest exact identity in a newly specified legal operation alphabet.",
"Derive a target-specific relaxation bound for that same explicit alphabet."
],
"acceptance_gate": "Name the changed hypothesis, all counted roles and a uniform identity; finite column tests alone do not certify it. A scoped local result may be retained without a complete supplier. Expanding it into further construction rounds requires the family-level budget and obligation gate in research-family-status.md; an optimistic passing relaxation alone is not acceptance.",
"first_step": {
"command": [
"python3",
"-B",
"scripts/research_selected_baseline.py"
],
"expected": "PASS_RESEARCH_BASELINE_BINDINGS_ONLY",
"inspect": "Resolve the current selected review, separate complex/ordinary BIT/admitted BIT/outer budgets, and compare the card against live handoffs before starting. Historical ceilings require exact source/model equivalence.",
"scope": "Source-binding orientation only; not a scientific replay or all-size proof. Materialize selected inputs if this checkout is sparse."
},
"first_experiment": "First compare the alphabet, role count and length with the fully replayed #1974 census and #2027 mixed-root tariff. Identify a changed premise and a legal native implementation; then price one smallest completed instance before searching larger words.",
"bottleneck_link": "A small-width completed saving changes the bit primitive; a finite-address identity alone is insufficient.",
"allowed_variations": [
"Choose a fixed larger register count or a specified nonabelian operation family."
],
"scope_change_triggers": [
"Remaining inside the already-excluded three-register abelian model.",
"Repeating the two-register GL3 census or a covered three-register length without changing a hypothesis; a formal group-algebra word without native conversion is a partial result."
],
"overlap_terms": [
"three-register",
"nonabelian",
"group algebra"
]
},
"scope_status": "changed",
"actual_scope": "Construct an inequivalent complete F2 projector pool for (k, matching number, joint-image dimension) = (3, 1, 4).",
"scope_change_reason": "Takes the first open question in PR #7707 at the smallest one-edge even maximum. The scope is classification of scalar alphabets before an endpoint word or native implementation.",
"bottleneck_link": "Inequivalent joint-image actions give different scalar alphabets for future binary endpoint-word searches at the same pattern and matching dimensions.",
"budget_minutes": 60,
"stop_rule": "Stop after the first inequivalent complete irreducible maximizer passes the exact width-four check.",
"outcome": "completed",
"evidence": "The written construction and trace comparison are in README.md. verify.py checks all 120 width-four rank-one idempotents and all 15 nonzero starting vectors in the new action, on CPython 3.12.15 with its standard library.",
"author": "Nick Petalas",
"ai_assistance": "Independent human and formal proof-kernel review are unclaimed.",
"open_questions": [
{
"question": "Which simultaneous-conjugacy classes occur at (k, matching number, joint-image dimension) = (3, 1, 4)?",
"why_it_matters": "A complete list would show which actions need separate word tests at the same dimension bound.",
"first_step": "At width four, form all spans of one matrix unit and two realized pattern classes, then classify their complete pools.",
"blocked_by": "Higher ambient widths need a separate reduction and are outside this construction."
},
{
"question": "Does the new alphabet admit a rank-saving paid word?",
"why_it_matters": "A complete irreducible action is only a candidate scalar alphabet.",
"first_step": "Specify a saturated exponent relation lattice and test a full ambient diagonal endpoint.",
"blocked_by": "Word chronology, dirty restoration, native realization and the complete recursive invoice remain open."
}
],
"predecessors": [
{
"pr": 7707,
"author": "aliromman",
"commit": "57e24923b609d2588b83321f530fa2a1e75ed364",
"proof": "https://github.com/aliromman/integer-mult-bounds/blob/57e24923b609d2588b83321f530fa2a1e75ed364/research/edge-matching-even-attainment/PROOF.md",
"sha256": "688efc2cd7c19e43c49193e94c1759c515dd79afbc4d832d62b27726f411daac",
"use": "Even maximum, open classification question and published comparison action. No runtime dependency."
},
{
"pr": 7692,
"author": "aliromman",
"commit": "c2870a38841f515db7203a0761cd77f014f7fb7a",
"proof": "https://github.com/aliromman/integer-mult-bounds/blob/c2870a38841f515db7203a0761cd77f014f7fb7a/research/edge-matching-image-bound/PROOF.md",
"sha256": "0e49106713d5b0dfe50bf193737edc0fa4ee7862eb6696b43a53ddf5e06bc026",
"use": "Matching bound and the two-lift edge pair. No runtime dependency."
}
],
"advance_claim": "https://github.com/CrocSwap/integer-mult-bounds/discussions/1166#discussioncomment-19718509"
}
28 changes: 28 additions & 0 deletions research/even-one-edge-maximizer/VERIFY.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,28 @@
{
"schema_version": 1,
"status": "ready",
"python": "3.12",
"requirements": [],
"inputs": [],
"commands": [
{
"argv": [
"{python}",
"-B",
"{package}/verify.py"
],
"timeout_seconds": 10,
"checks": "One matrix unit, exact complete pool, full-image irreducibility and trace-one pair count for the new witness."
},
{
"argv": [
"{python}",
"-B",
"-O",
"{package}/verify.py"
],
"timeout_seconds": 10,
"checks": "One matrix unit, exact complete pool, full-image irreducibility and trace-one pair count for the new witness."
}
]
}
76 changes: 76 additions & 0 deletions research/even-one-edge-maximizer/verify.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,76 @@
#!/usr/bin/env python3
"""Check the complete width-four witness and its trace invariant over F2."""

from itertools import combinations, product

WIDTH = 4
CARDINALITY = 1 << WIDTH
# The low-to-high bits represent a, b, f, g.
PROJECTORS = ((1, 3), (3, 2), (5, 9), (11, 6))


def require(condition, message):
if not condition:
raise ValueError(message)


def pairing(image, covector):
return (image & covector).bit_count() % 2


def off_pattern(image, covector):
pattern = 0
for row in range(WIDTH):
if image & (1 << row):
entries = covector & ~(1 << row)
pattern |= entries << (WIDTH * row)
return pattern


def span(generators):
vectors = {0}
for generator in generators:
vectors |= {vector ^ generator for vector in vectors}
return vectors


def reaches_all_images(start, images):
reached = {start}
for _ in range(len(PROJECTORS)):
reached |= {
image for image, covector in PROJECTORS
if any(pairing(vector, covector) for vector in reached)
}
return images <= reached


def main():
patterns = span(tuple(off_pattern(*projector) for projector in PROJECTORS))
require(len(patterns) == 8, "expected pattern dimension three")
units = {pattern for pattern in patterns if pattern.bit_count() == 1}
require(units == {1 << 1}, "expected the single unit E_ab")

factors = range(1, CARDINALITY)
idempotents = tuple((x, y) for x, y in product(factors, repeat=2) if pairing(x, y))
pool = {pair for pair in idempotents if off_pattern(*pair) in patterns - {0}}
require(len(idempotents) == 120, "incorrect width-four idempotent inventory")
require(pool == set(PROJECTORS), "complete pool differs from the four displayed lifts")

images = {image for image, _ in pool}
require(len(span(images)) == CARDINALITY, "joint image is not the full width-four space")
for vector in factors:
require(reaches_all_images(vector, images), f"nonspanning orbit at {vector}")

reciprocal = {
(i, j) for i, j in combinations(range(len(PROJECTORS)), 2)
if pairing(PROJECTORS[i][0], PROJECTORS[j][1])
and pairing(PROJECTORS[j][0], PROJECTORS[i][1])
}
require(reciprocal == {(0, 2), (1, 3)}, "unexpected trace-one pairs")
print("PASS_EVEN_ONE_EDGE_MAXIMIZER")
print(f"{len(idempotents)} idempotents scanned, {len(patterns) - 1} nonzero patterns, "
f"{len(pool)} lifts, {len(factors)} spanning orbits, {len(reciprocal)} trace-one pairs")


if __name__ == "__main__":
main()
6 changes: 5 additions & 1 deletion research/historical-selected-replay-gap/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -69,7 +69,11 @@ python3 -B scripts/research_package.py --package research/historical-selected-re

The direct command prints `PASS_HISTORICAL_SELECTED_REPLAY_GAP`. A registered
replay prints `PASS_REGISTERED_RESEARCH_PACKAGE`. Both require Git history
containing the two inspected commits.
containing the two inspected commits and the source and manifest objects at
`2485805a6d47f36ea83ddde6327d453ac82d44dc`. The verifier uses a temporary
sparse checkout of that commit, checks the declared source digests there, and
uses its planner and manifest inventory. The recorded rejections and selection
therefore remain reproducible after later changes to the live Makefile or CI guide.

## Open next questions

Expand Down
Loading