four.mino.mobi · papers

How a paper gets made

Paper 1, being written in the open. Everything below is read from the paper's own files as they stand: the claims' values are recomputed from the baked table on every build, and a test fails the build if any claim stops holding. Nothing here is a description of the work written after the fact.

Read the current draft →the paper as it stands: introduction, Parts I–III, figures and references, regenerated from the code on every change; every reference to the supplement links into it PDF

Version 1 →frozen on 2026-11-11: the copy to cite and share, which never changes, with its own supplement, PDFs and data PDF

The supplement →the notation and appendices A–K: the arguments and methods in full, the checks against our world, chemistry's starting data, response, and how to reproduce every number PDF

Paper 2, in outline →fusion, stars, abundances and gravity in four dimensions: the claim map where the nuclear and stellar results are collected as they land, each with what holds it up and what it still needs

Paper 2, section 2: the windows →which strong and weak forces let a four-dimensional universe burn hydrogen: figure 1, what it shows, and how every line and point on it was computed (draft)

Paper 2, section 3: stars that live →how long four-dimensional stars burn, what a star like our Sun makes at each strength of the nuclear force and of gravity, and why they get convective cores without a CNO cycle: figure 2 and its methods (draft)

The rule

The paper's subject is a choice. In four dimensions the law between charges can't be inherited: Gauss's law leaves no atom, so one has to be picked, and what it costs has to be paid in the open. Paper 1 picks 1/r, says why, and measures what that choice does by solving the whole periodic table twice, once under the law and once with the shell theorem forced back on.

A number in the paper is never typed. It's read from a file the solver baked, and a test checks it against that file. The claims, the figures and eventually the text are generated, so the prose can't drift from the numbers. References are cited only once their details have been confirmed from the source or an index; seen-in-passing doesn't count. Anyone can solve the atoms again and get the same numbers to the last digit.

The pipeline

From solver to PDF. Each stage's status is kept in process.json.

The claims, checked now

The figures

Drawn by papers/p1/figures/make.mjs from the bake and the law code on every build; captions from the same file. Hover a point for its value.

The text so far

Each section's source is in papers/p1/text/, with every number a token the build fills from the code or the bake; a token it doesn't know stops the build.

The references

Each source marked by how far it has been checked: read the source itself, index its details confirmed from a publisher's or index's record, cited only seen cited in another work, memory not yet looked up. Only read and index may be cited.

The data

Generated from the bake by papers/p1/build.mjs; the field guide is data/README.md. Hashes are SHA-256 of the files as served.

Solve it again

The reproduction re-solves the paper's atoms with the same function that baked them and compares, configuration exactly and energies to 10⁻⁴ hartree. It ships with the solvers in the open mirror (MIT), and the mirror's workflow runs the sample on every change.

node papers/p1/reproduce.mjs                 # a sample: each 4D row's first, second and noble gas, and six of ours
node papers/p1/reproduce.mjs --z4=78,138     # chosen 4D elements
node papers/p1/reproduce.mjs --all           # all 176 (hours on one machine)

Last run: 14 of 14 reproduced to the bit, 18 s on four cores (2026-11-04).

Before a word of prose

Journal

The outline and figure plan · the widgets · the table itself · the literature register · the code