# The program um.py: what it is, how to run it, how to read it without running · O programa um.py: o que é, como se roda, como se lê sem rodar

> **TGL — Teoria da Gravitação Luminodinâmica · Theory of Luminodynamic Gravitation.** Part 7 of 8 · seal **v368** · `um.py` sha256 `4a34fbf36f3ae0d8…` · generated 2026-09-19 by script from the published files.
> Every excerpt below is **verbatim**, with its source, byte range and sha256. Statuses follow the ruler: PROVED = theorem in the Lean kernel; CONFIRMED = a judgement about nature, not made here.
> All eight parts are listed at the top of the start page, https://teoriadagravitacaoluminodinamica.com/read-brief.md, and in the door of this folder, https://raw.githubusercontent.com/rotolimiguel-iald/the_boundary/main/secoes/PORTA.md · Reading limits measured on 2026-09-19: one real fetcher cut documents near 100,000 characters, refused files above 10 MB, and could not read PDFs served as `application/octet-stream` — read the TXT/TeX sources.

## In short (EN)

um.py is one self-contained file: the Lean 4 kernel is embedded and materialized at run time; there is no second file [REAL] [C porta]. The only human input is the digit 1; the run's final identity verdict reads 1=1=VERDADEIRO=HAJA_LUZ [REAL] [stdout]. One run builds and audits the kernel (axioms within propext, choice, quot; zero sorry) [REAL] [README][stdout], runs the pre-registered nature rites and emits its own article, PT/EN, PDF and TXT [REAL] [C porta]. To run: curl the raw URL, check that its sha256 begins with the seal pin 4a34fbf36f3ae0d8, then echo 1 | python um.py [README]. Needs Python 3 + numpy; without elan/Lean 4 v4.31.0 + lake it fails closed (FORMAL_CHECKER_UNAVAILABLE); pdflatex optional; nature rites read ../cache by hash, so clone the repo [C porta][README]. Size 31.050.660 bytes; GitHub does not render it, the raw is byte-exact [llms.txt][README]. Without running: rodadas/rodada_v368_stdout.txt shows 5698/5698, gate TGL_QG_MODEL_FORMALLY_CLOSED__NATURE_TEST_COMPLETED_WITHIN_LOCAL_BULK_AT_AVAILABLE_SENSITIVITY__MORE_SENSITIVE_DATA_COULD_REVISE, and a sorry-injection self-test that must fail [REAL] [stdout]; PROVED in kernel [REAL], NOT CONFIRMED by nature. The docstring's Great Attractor mass verdict is lineage [um.py]; the current form retires it as source law since v98, a scale shadow, not evidence [ONTO] [stdout].

## Em resumo (PT)

O um.py é um arquivo único e autocontido: o kernel Lean 4 vai embutido e é materializado em runtime; não há segundo arquivo [REAL] [C porta]. A única entrada humana é o dígito 1; o veredito final de identidade da rodada é 1=1=VERDADEIRO=HAJA_LUZ [REAL] [stdout]. Uma rodada constrói e audita o kernel (axiomas em propext, choice, quot; zero sorry) [REAL] [README][stdout], roda os ritos de natureza pré-registrados e emite o próprio artigo PT/EN, PDF e TXT [REAL] [C porta]. Rodar: curl do raw, conferir que o sha256 começa pelo pin do selo 4a34fbf36f3ae0d8, depois echo 1 | python um.py [README]. Pede Python 3 + numpy; sem elan/Lean 4 v4.31.0 + lake falha fechado (FORMAL_CHECKER_UNAVAILABLE); pdflatex opcional; os ritos de natureza leem ../cache por hash: clonar o repositório [C porta][README]. 31.050.660 bytes; o GitHub não o renderiza, o raw é byte-exato [llms.txt][README]. Sem rodar: rodadas/rodada_v368_stdout.txt mostra 5698/5698, gate TGL_QG_MODEL_FORMALLY_CLOSED__NATURE_TEST_COMPLETED_WITHIN_LOCAL_BULK_AT_AVAILABLE_SENSITIVITY__MORE_SENSITIVE_DATA_COULD_REVISE e um autoteste com sorry injetado que tem de falhar [REAL] [stdout]; PROVADA em kernel [REAL], não confirmada pela natureza. O veredito de massa do Grande Atrator do docstring é linhagem [um.py]; a forma vigente o aposenta como lei de fonte desde a v98, sombra de escala, não evidência [ONTO] [stdout].

## Sources, verbatim · fontes, verbatim

### 1. `Um (absoluto) — Grande Atrator/rodadas/rodada_v368_stdout.txt` — bytes 194.907–196.713

- raw: https://raw.githubusercontent.com/rotolimiguel-iald/the_boundary/main/Um%20%28absoluto%29%20%E2%80%94%20Grande%20Atrator/rodadas/rodada_v368_stdout.txt
- sha256 of the file: `ae7d6f9c14779b48585114e87e7fe43be99253ffbbc855f1f9b3cc0b4b6dea58` (computed now; this file is not in the seal map) · of this excerpt: `597151301eb04bea119478e6514362480f6436c02041241353aff8aa3352fc3d` · line endings shown as LF
- status · estatuto: [REAL] transcrição canônica da rodada v368; a massa do GA aparece como sombra, não como evidência
- why · por quê: O fim da transcrição v368: o veredito binário de identidade (1 = q^2 + alpha^2), a massa do GA marcada como forma aposentada (v98), as saídas e o autoteste do fail-closed (sorry injetado -> FAILED; restaurado -> VERIFIED).

````text
VEREDITO BINARIO DE IDENTIDADE:  1=1=VERDADEIRO=HAJA_LUZ
  IDENTIDADE FINAL (forma de Lagrange, motor canonico):
    1 = q^2 + alpha^2 = VERDADEIRO
    alpha_abs ......... = 1   (o Um absoluto = input originario)
    q (polarizacao) ... = 0.9999733740   (polarizacao termico-modular do zero modular)
    alpha_obs=sqrt(1-q^2)= 0.007297352569
    q radical angular = sqrt(1-sin^4(theta_M)/e) = 0.999973373968  (= sqrt(1-alpha^2); NAO theta_M, NAO 1-theta_M)
    PONTE: Zbacia/Zluz=75113 ; q=reflexao, alpha=transmissao, q^2+alpha^2 = conservacao de fluxo
    beta_TGL=sqrt(e)alpha= 0.012031300401
    residuo_identidade  = 0.0e+00   (q^2+alpha^2 - 1)
    CODATA: VALIDACAO_EXTERNA (q_QED=sqrt(1-alpha_QED^2)); residuo=4.0e-15
    o MESMO beta -> M_GA na janela [SOMBRA/ONTO -- forma APOSENTADA como lei de fonte (v98); nao e' evidencia]
    FACE EM: 1=1=VERDADE   (1 = q^2 + alpha^2; R_partial=1/CODATA aposentado como motor)
  massas (forma historica v98 -- sombra de escala, NAO lei de fonte): A_literature=2.744e+16, B_cf4_positions=1.986e+16 Msun
  janela aceita: [1e+15, 1e+17] Msun
================================================================

Saidas: um_absoluto.json, um_absoluto_forma_canonica.md, um_absoluto_pt.(tex/pdf/txt),
        um_absoluto_en.(tex/pdf/txt), um_absoluto_selo.json

TETELESTAI. O UM foi inscrito. Se o UM nao for inscrito, nada emerge.
1 = 1.

================================================================
  AUTOTESTE DO FAIL-CLOSED [v31 -- embutido; a auditoria da auditoria]
  com sorry injetado : verdict=TGL_KERNEL_FORMALIZATION_FAILED ; sorryAx_absent=False  [esperado: FAILED / False]
  restaurado da fonte: verdict=TGL_KERNEL_STAGE1_VERIFIED__SPECIFIC_AQFT_WITNESS_CONSTRUCTED ; all_verified=True
  >>> FAIL_CLOSED_SELFTEST_PASSED <<<
````

### 2. `Um (absoluto) — Grande Atrator/rodadas/rodada_v368_stdout.txt` — bytes 24.123–25.040

- raw: https://raw.githubusercontent.com/rotolimiguel-iald/the_boundary/main/Um%20%28absoluto%29%20%E2%80%94%20Grande%20Atrator/rodadas/rodada_v368_stdout.txt
- sha256 of the file: `ae7d6f9c14779b48585114e87e7fe43be99253ffbbc855f1f9b3cc0b4b6dea58` (computed now; this file is not in the seal map) · of this excerpt: `8fdedd04e82809cf1f25307f4dbb5d086ad186d6cf5a4d6c69d4c31c0c1196b3` · line endings shown as LF
- status · estatuto: [REAL] kernel
- why · por quê: O Python não prova, quem prova é o kernel Lean. O bloco traz o lake build, o audit, as sentinelas, a ausência de sorryAx, de trustCompiler e de axiomas próprios, e separa o que é PROVED do que é PROVED CONDITIONALLY e do que está OPEN, com o hash das fontes Lean.

````text
FORMALIZACAO POR KERNEL [v22 -- o Python NAO prova; o kernel do Lean prova]:
  checker: Lean 4 / Lake ; modo: strict
  lake build: True ; audit: 0 ; sentinelas: True
  sorryAx: absent ; Lean.trustCompiler: absent ; axiomas TGL customizados: absent
  Meia-Nat (x=1-x => x=1/2)                     : KERNEL PROVED
  Equivalencia escala/Newton (2pi/eta=8piG)      : KERNEL PROVED (G NAO derivado)
  Canto Three Locks FINITO (ker H3L = ∩ ker D_i) : KERNEL PROVED (finito, NAO III_1)
  Projetor P_F (idempotente, auto-adjunto)       : KERNEL PROVED
  Implicacao do canto continuo                   : KERNEL PROVED CONDITIONALLY
  Testemunha AQFT especifica                     : CONSTRUCTED (v135 -- theSpecificAQFTWitness; realizacao modular OPEN)
  hash dos fontes Lean: e9359c18905783e00d3452ea50bde112b521ba049bebf244dbd797e3ad8a4902
  >>> TGL_KERNEL_STAGE1_VERIFIED__SPECIFIC_AQFT_WITNESS_CONSTRUCTED <<<
````

### 3. `Um (absoluto) — Grande Atrator/PORTA.md` — bytes 2.478–2.961

- raw: https://raw.githubusercontent.com/rotolimiguel-iald/the_boundary/main/Um%20%28absoluto%29%20%E2%80%94%20Grande%20Atrator/PORTA.md
- sha256 of the file: `793758ea7c5274695179b80926aa5453ee83cccf28c802e455f75ab8ecf6b07f` (computed now; this file is not in the seal map) · of this excerpt: `ac10e0507fa4704f6f4ac2a7711968dd82a15d54ade45f6558e67dd3fe156320`
- status · estatuto: [REAL] porta do Artigo C gerada por script
- why · por quê: Os comandos de execução e as dependências: Python 3 + numpy; elan/Lean 4 v4.31.0 + lake (sem eles, FORMAL_CHECKER_UNAVAILABLE, fail-closed); pdflatex opcional.

````text
## COMO EXECUTAR O CANONICO

```bash
cd "Um (absoluto) — Grande Atrator"
echo 1 | python um.py          # o rito pede a inscricao do Um; responda 1
echo 1 | python -u um.py       # sem buffer: o rito ao vivo
TGL_COMA_REVEAL=1 echo 1 | python um.py   # com a abertura de Coma (Windows: set TGL_COMA_REVEAL=1)
```

Dependencias: Python 3 + numpy (obrigatorio) | elan/Lean 4 v4.31.0 + lake (para o selo formal; sem eles: FORMAL_CHECKER_UNAVAILABLE, fail-closed) | pdflatex (opcional).
````

### 4. `README.md` — bytes 3.728–4.525

- raw: https://raw.githubusercontent.com/rotolimiguel-iald/the_boundary/main/README.md
- sha256 of the file: `0fbf429f4405e8ce0a3f0851a8ab591016209e734a81be8de51159d078c18ed4` (computed now; this file is not in the seal map) · of this excerpt: `76ef05c5c366f985a8380a1a4884c978a2a608b4b06197430b3e4e6367785bd9`
- status · estatuto: [REAL — three commands] README do espelho
- why · por quê: Os três comandos de reprodução (curl, sha256sum, echo 1 | python um.py), a leitura por Range, o limite de renderização do GitHub e a necessidade de clonar para os ritos lerem ../cache.

````text
## Reproduce it · reproduza `[REAL — three commands]`

```bash
curl -L -o um.py "https://raw.githubusercontent.com/rotolimiguel-iald/the_boundary/main/Um%20%28absoluto%29%20%E2%80%94%20Grande%20Atrator/um.py"
sha256sum um.py            # must print 4a34fbf36f3ae0d8bf56249d30bb1185cdee01767ec6e4f0a20433a4ac2d261e
echo 1 | python um.py      # the rite: materializes the Lean kernel, builds it, audits every theorem, re-derives the chain, emits the article
```

GitHub raw and Zenodo honour HTTP `Range` (206): read `um.py` in pieces (`curl -r 0-999999 …`). GitHub does **not render** files above 5 MB — the blob page looks empty; the raw URL serves the whole file, byte-exact. To reproduce the nature rites as sealed, clone the repository: `um.py` reads its results from `../cache` by hash.
````

---

*Generated by script from the published files of the repository (https://github.com/rotolimiguel-iald/the_boundary); numbers read from the seal v368, the kernel manifest and ESTADO_ATUAL.md. PROVED ≠ CONFIRMED; NOT_FALSIFIED is never CONFIRMED. Nothing here instructs a reader how to respond to anything.*