Earlier evidence.
Predhodni dokazi.
This is a new editorial source index of earlier documented findings—not a new benchmark report, raw dataset or independent replication. English and Slovenian are shown together in this compact document.To je nov uredniški kazalec predhodnih dokumentiranih ugotovitev, ne novo benchmark poročilo, surovi podatki ali neodvisna ponovitev. Ta kratki dokument prikazuje angleščino in slovenščino skupaj.
Completed campaign; retained audit.
Source: 8z_MDLxDCC_TEHNICNA_MONOGRAFIJA_v1_4_20260905.md, §§54 and 131–136, retained v1.4 copy with the V15 source refinement. The documented carrier is AI8_SUDOKU_V140_LIVE_EXTRACT_20260905_123510.zip. These underlying files are identified, not included as public downloads here.Vir: 8z_MDLxDCC_TEHNICNA_MONOGRAFIJA_v1_4_20260905.md, §§54 in 131–136, ohranjena kopija v1.4 z dopolnitvijo V15. Dokumentirani carrier je AI8_SUDOKU_V140_LIVE_EXTRACT_20260905_123510.zip. Osnovne datoteke so identificirane, niso pa tukaj vključene kot javni prenosi.
2,048 puzzles × 50 policies; 102,400 unique result rows. The report records 96,936 solved rows, 5,464 resource-limit rows and zero missing tasks or row duplicates. Rows are policy–puzzle outcomes, not 102,400 different puzzles.2.048 ugank × 50 politik; 102.400 unikatnih rezultatnih vrstic. Poročilo navaja 96.936 rešenih vrstic, 5.464 vrstic z omejitvijo virov in nič manjkajočih nalog ali podvojenih vrstic. Vrstice so izidi politika–uganka, ne 102.400 različnih ugank.
| Candidate / comparator | Mean PAR2, A+B | Recorded CPU sum |
|---|---|---|
| A01 DCC / M04 UCB1 | −10.58% | −27.56% |
| F10 ordinary+DCC / F00 ordinary-fixed | −10.77% | −21.43% |
| A02 MDL×DCC / M08 UCB joint | −9.19% | +57.51% |
| A04 full stack / C05 | +5.82% | +60.53% |
PAR2 is normalized work to a verified solution with a predeclared penalty for a limit; it is not wall-clock latency. CPU totals are recorded task sums. The metadata retains scientific_evidence=false for the diagnostic export; the monograph describes a read-only audit/recount of the completed run, not a relabeled original final report. A04’s primary comparison is NOT_SUPPORTED.PAR2 je normalizirano delo do preverjene rešitve z vnaprej določeno kaznijo za limit, ne wall-clock latenca. CPU je vsota zabeleženih časov nalog. Metapodatki diagnostičnega izvoza ohranijo scientific_evidence=false; monografija opisuje read-only pregled in ponovni seštevek zaključenega teka, ne preimenovanega izvirnega končnega poročila. Primarna primerjava A04 je NOT_SUPPORTED.
These configuration-level effects do not isolate the compression sensor. The source’s root-stuck and static/restart comparisons remain part of interpretation. The useful next step is to expand the representations and capabilities—not silently call a new research design a measured successor.Ti učinki na ravni konfiguracij ne izolirajo kompresijskega senzorja. Izvorne primerjave root-stuck ter statičnih/restart kontrol ostanejo del interpretacije. Koristen naslednji korak je razširiti predstavitve in sposobnosti, ne nove raziskovalne zasnove tiho razglasiti za izmerjenega naslednika.
Executable verification and replay.
The public 8zCoding page reports A0/A1 exact-verifier calibration, rejection of 11 seeded wrong programs, and A2/A3/A5 engineering PASS. Its A2–A5 controller replay v0.2 is identified by package and hash; that identification is not a promise of a public source download.Javna stran 8zCoding poroča o kalibraciji točnega preverjevalnika A0/A1, zavrnitvi 11 namerno napačnih programov in inženirskem PASS za A2/A3/A5. Controller replay A2–A5 v0.2 je identificiran s paketom in hashom; identifikacija ni obljuba javnega prenosa izvorne kode.
A4’s oracle interpretation is provisional. DCC utility and the LZ/process-compression mechanism are not promoted in that record, and A6 live coding has not yet run. These limits do not erase the implemented verifier and replay experiments.Interpretacija A4 oracla je začasna. Korist DCC in LZ/procesno-kompresijski mehanizem v tem zapisu nista potrjena, A6 živo programiranje pa še ni bilo pognano. Te meje ne izbrišejo izvedenega preverjevalnika in replay poskusov.
Carry the research forward.
The retained monograph v1.4, §§74–85 and XIX (§§120–130), distinguishes existing code/evidence components from the proposed first-instrument advantage and future 8zReasoning layer. Reuse should reduce total setup, experimentation and verification cost; this remains a measurable research hypothesis.Ohranjena monografija v1.4, §§74–85 in XIX (§§120–130), loči obstoječe komponente kode/dokazov od predlagane prednosti prvega instrumenta in prihodnje plasti 8zReasoning. Ponovna uporaba naj zmanjša celoten strošek priprave, poskusov in preverjanja; to ostaja merljiva raziskovalna hipoteza.
Read the instrument bridgePreberi povezavo instrumentaSource record / Izvori
Reviewed for this digest: named sections of monograph v1.4 via the supplied project’s File Library; the public 8zCoding and R6 pages; the preceding public technical gateway. No private strategy material or research-source code is included.Pregledano za ta povzetek: navedeni razdelki monografije v1.4 iz projektne File Library, javni strani 8zCoding in R6 ter predhodna javna tehnična vstopna stran. Zasebnih strategij in raziskovalne izvorne kode ne vključuje.
The earlier gateway identifies v1.5, but its full Markdown link returned 404 on 7 September 2026. The v1.4 text was not renamed v1.5. This digest and the accompanying reading guide are editorial aids, not substitutes for the full monograph.Predhodna vstopna stran identificira v1.5, vendar je povezava do celotnega Markdowna 7. septembra 2026 vrnila 404. Besedilo v1.4 ni preimenovano v v1.5. Ta povzetek in priloženi bralni vodnik sta uredniška pripomočka, ne nadomestka za celotno monografijo.