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
31 changes: 31 additions & 0 deletions research/p104-delayed-basis/PRICING-CHANGES.diff
Original file line number Diff line number Diff line change
@@ -0,0 +1,31 @@
--- PR5612/pricing.py
+++ pricing.py
@@ -12,8 +12,8 @@
from fractions import Fraction as Q
import local as L

-KAPPA, TESTED = Q('.00310299'), Q('.0031029905')
-PREVIOUS = Q('.00310184')
+KAPPA, TESTED = Q('.00310440'), Q('.0031044005')
+PREVIOUS = Q('.00310299')


def encode(x):
@@ -62,7 +62,7 @@
second_p[8*(j+1)] += 208
L.need((targets,paid) == (second_t,second_p), 'Independent full invoice')
L.need(max(targets) == 145 and max(paid) == 72, 'Maximum recursive ranks')
- L.need(sum(targets.values()) == 13030 and sum(paid.values()) == 66636, 'Complete field and call counts')
+ L.need(sum(targets.values()) == 13030 and sum(paid.values()) == 66618, 'Complete field and call counts')
L.need(sum(r*n for r,n in targets.items())-sum(r*n for r,n in paid.items()) == 760, 'Complete rank surplus')
return targets,paid,counts

@@ -150,7 +150,7 @@
L.need(wrong_cutoff[0] > 1, 'Old cutoff fails with this helper inventory')
bad=outer_module.ledger(dict(params,bit_saving=Q('.0027')))
L.need(any(x <= 0 for x in bad['strict_checks'].values()), 'Insufficient BIT contract rejected')
- return encode({'kappa':KAPPA,'kappa_decimal':'.00310299','previous_open_pr5608':PREVIOUS,'gain':KAPPA-PREVIOUS,
+ return encode({'kappa':KAPPA,'kappa_decimal':'.00310440','previous_open_pr5612':PREVIOUS,'gain':KAPPA-PREVIOUS,
'relative_gain_percent':100*(KAPPA/PREVIOUS-1),'raw_root_interval':[lower,upper],
'raw_headroom':[lower-KAPPA,upper-KAPPA],
'raw_ceiling':{'interval':[Q(0),cap],'derivative_upper':derivative_upper},'margin_at_previous':E.display(margin(PREVIOUS)),
22 changes: 22 additions & 0 deletions research/p104-delayed-basis/PRICING-EDITS.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,22 @@
[
[
"KAPPA, TESTED = Q('.00310299'), Q('.0031029905')",
"KAPPA, TESTED = Q('.00310440'), Q('.0031044005')"
],
[
"PREVIOUS = Q('.00310184')",
"PREVIOUS = Q('.00310299')"
],
[
"sum(paid.values()) == 66636",
"sum(paid.values()) == 66618"
],
[
"'kappa_decimal':'.00310299'",
"'kappa_decimal':'.00310440'"
],
[
"previous_open_pr5608",
"previous_open_pr5612"
]
]
44 changes: 44 additions & 0 deletions research/p104-delayed-basis/PROOF.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,44 @@
# Helper341 boundary composition

This composes Ali Romman's PR5602 identity, pinned at
`6c224aa720203c4f4615e6538d8038a275387db8`, with the published delayed/prepared
word. The source canonical SHA256 is
`df44a24231edd04954ebf066b1efb311441e5d9d48b64d03e555cbea087b25c6`.
This is a checked composition, not a new basis identity.

Immediately after the source's last read of341, event11285,
`ADD346,341,+1` at frame240, exact rational replay gives

`current341 - 2*current337 = x341 - 2*x93 - 2*x339 + 2*x351`.

Both current rows are at240. Initially93,339,341,351 are all at frame3.
Precompose the word with the three ordinary ADDs implementing

`x341 <- x341 + 2*x93 + 2*x339 - 2*x351`.

The update is invertible and does not change its source coordinates.
The original endpoint column of input341 is the unit helper341 column, so
this precomposition changes no other final output. After the last341 read,
subtract twice337 from341. The displayed relation now gives exactly the
original independent input `x341`.

There are no later reads of341. Its remaining eight destination operations,
including its rank5 ascent240-to127, can therefore be removed, and its final
frame declared240. All other operations remain unchanged. No new input,
temporary, owned lifetime, or zeroing operation is introduced.

The exact paid change is `-H5`. Helper341's productive target changes from
rank9 to rank4, giving target change `H4-H9`. Thus the signed funding gain
is `H4+H5-H9`, strictly positive for `H_r=r^t`, `0<t<1`. The target debit is
essential; counting only the removed paid transport would overstate the gain.

The separate independent agent calculation checked all756 original
input columns, all756 endpoint rows, every common frame and ownership
transition, and the unchanged INIT/CLEAR/COPY sequence. It confirms747
independent inputs,11 owned lifetimes, and peak2. Candidate canonical SHA256:
`22aa2b59ed2900dbd7ed6060586422cc6e72927cf7045727b68ec1b0b24b4209`.
The package verify.py separately checks native, caller, ordinary-cost and compiler obligations and records the full result in RESULT.json.

Independent audit prepared for flyingviper with substantial OpenAI Codex
assistance. Retain Ali Romman's PR5602 attribution and the composed sources'
notices. Apache-2.0.
96 changes: 96 additions & 0 deletions research/p104-delayed-basis/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,96 @@
# Conditional kappa 0.00310440 by composing a paid boundary change

This composes Ali Romman's helper341 boundary identity from [PR #5602](https://github.com/CrocSwap/integer-mult-bounds/pull/5602)
with the delayed/prepared construction in [PR #5612](https://github.com/CrocSwap/integer-mult-bounds/pull/5612).
The complete conditional endpoint increases from **0.00310299 to 0.00310440**,
a gain of **0.00000141**. The basis identity is credited to its author;
this contribution verifies that it survives the intervening shared relays,
changed contacts, independent splits and owned preparation.

Three initial coframed ADDs and one final coframed correction replace eight
operations on helper341. Its last rank5 transport disappears, and its
productive target changes from rank9 to rank4. The local funding gain is
`H4+H5-H9`, where `Hr=r^(1-kappa)`. The complete gain is18 times that expression.
The target debit is included; the removed transport alone is not the gain.

The exact raw crossing is **(0.00310440547057,0.00310440547058)**, tested at
0.0031044005. The full invoice has **66,618 paid calls,13,030 productive
targets and rank surplus760**. The word has25,525 events,747 independent
inputs,539 helpers,nine owned roles,11 owned lifetimes and peak occupancy two.
No independent input or owned lifetime is added. Every retained input is
preserved exactly, and every owned role returns zero.

Read [PROOF.md](PROOF.md), [the pinned witness](WITNESS.json), [the small constructor](construct.py)
and [the complete receipt](RESULT.json). The specialized constructor must agree
exactly with the unchanged published PR5602 constructor on this new source.
Only the causal cut index changes, from11151 to11285; the boundary covectors,
helper identity and chart remain exactly the same.

## Reproduce

Python3.12 and its standard library suffice:

```sh
python3.12 -B scripts/research_package.py --package research/p104-delayed-basis
```

Both registered modes print `PASS_P104_DELAYED_BASIS_KAPPA_00310440`.
The launcher prints `PASS_REGISTERED_RESEARCH_PACKAGE`. Replay authenticates the
official24-file PR5612 export, its retained PR5608 export and33-file framework;
checks the exact published witness, two complete constructions, all756 endpoint
rows and dense transpose columns, every signed native map, the full physical
caller, complete invoice and ordinary charges, two interval engines at the
tested endpoint, the compiler and58 outer predicates. Five controls reject
missing input correction, missing cut correction, missing owned initialization,
old target credit and an unpaid transport. Complete decompressed candidate
payloads are compared; compressed source archives remain byte-pinned.

The inherited prepared checker is reused unchanged and checked against its
retained source adaptation. [PRICING-EDITS.json](PRICING-EDITS.json) and
[PRICING-CHANGES.diff](PRICING-CHANGES.diff) declare the only five price changes:
endpoint, comparator, paid count, output decimal and comparator label.
The original mathematical inequalities and compiler reserves remain unchanged.

## Scope and ceiling

The five inherited framework assumptions remain: native all-column realization,
owned erasure, uniform compiler closure, precision and fixed-tape costs, and
analytic recovery. The helper-endpoint contract remains. The enormous uniform
word and complete rounded multiplier are not materialized. No independent
human or proof-kernel acceptance is claimed, and the selected record is unchanged.

The signed moment is strictly decreasing on[0,0.0034]. The fixed-invoice raw
ceiling is the interval above, leaving5.47057e-9 to5.47058e-9 above the claimed
endpoint. This is not an optimum over other constructions. Logarithm power1024,
tail decay39/40 and the complete outer/ordinary conditions are retained.

## Sources and independent checks

Base: canonical main `d7633ce230196a1d5d061347de7a0fd8be512ed6`, priority revision7.
PR5612 is pinned at `faa9f6dbe6b8cc4a996d3edeca71823056453e1a`;
PR5602 at `6c224aa720203c4f4615e6538d8038a275387db8`. Exact archives,
source files, hashes and original notices are retained through[SOURCES.json](SOURCES.json).
Replay requires no foreign Git objects, network or local workspace files.

Two independent agents checked the compatibility: one derived the causal
witness and compared complete histograms; another implemented separate exact
scalar/frame/ownership replay over all756 original columns and checked the
ordered basis updates. This is agent verification, not independent human review.

Credit Ali Romman for PR5602's identity and constructor, independent splits,
frame changes and native/caller/price machinery; Scratch-net for prepared573;
Pierre-Loic Doulcet/hexapode for commuting contacts; flyingviper for shared and
delayed relays; Joel Pulikkan/GamingPuzzled, DreamingOfClouds, thu-nmrc,
Gareth Davidson, David Arias, Cinna Davis, Jacob Alan Sussman, zxv, Cao Yifan
and all retained geometry, BIT and analytic contributors. This composition
and verification were prepared for flyingviper with substantial OpenAI Codex
assistance and independent agents. Apache-2.0.

## Open next questions

Can a changed independent split cut make another published boundary identity
compatible without losing more target credit than it saves? First derive a
current-value relation after the helper's last source read; preserve its full
arbitrary-input boundary basis and debit every changed productive rank.
The other five PR5602 witnesses are not claimed compatible here. Structured
questions and the assignment refinement are in[TASK.json](TASK.json).
Loading