What it shows
The record states one part of the result this way:
The published record says, word for word (an excerpt)
the counts went DOWN when padding was removed.
In plain words: in the record’s latest run, the SMT solver proved 27 of 27 required statements, and refuted all 8 planted false claims. An older excerpt in the same record shows a smaller required count; the live run is the one quoted here.
The Lean files behind the theorem cards on the home page are checked separately. A run of the lab’s Lean proof gate, which builds the packaging files among them and audits their axioms, at 2026-09-12T00:43:36Z, exited 0. For the timing-channel file, the lab’s axiom report records it as compiled with return code 0, for the exact source this site serves. The cone-sufficiency file has no build record on this page.
Why it matters
A bound is only as good as the statement it rests on. When those statements are checked by an outside tool, and the check is shown able to fail, a reviewer can trust each encoded statement as written; the step from statement to bound is argued on each result page, not machine-checked.
What is ours, and what is not
The SMT solver is an outside tool (see the prior art below). What is ours is the set of statements, their encoding, and the harness that requires planted false claims to be refuted.
Who should care
- Technical reviewers who need to know which of our results rest on checked statements, and which do not.
- Teams that want to re-run the checks: the record names the command.
The limits, in the record’s words
The published record says, word for word (an excerpt)
Formal proof establishes the encoded proposition, not the fidelity of the physics abstraction. ValidStep-style regime hypotheses are assumptions, not discharged facts.
The published record says, word for word (an excerpt)
Z3 was 26/26 until a tautological unsat[False] check was removed as padding.
The published record says, word for word (an excerpt)
Every documented alternative count is discarded in favour of the live run: cvc5 36/36 and 46/46 are set aside, as are Lean 6/16/17/26/56. The estate-external claim of '8,567 Lean 4 machine-checked proof jobs' is a FABRICATION and appears nowhere in this repository.
In plain words: a proof checks the statement as written, not whether the statement describes the physics. The record also says that an earlier list of checked Lean theorems pointed at a proof file that was never compiled, so those theorems were taken out of the count. The counts went down when padding was removed, and every other count that once circulated is set aside in favour of the live run.
Open source for this step
Tools and datasets we publish for the package step of building a multi-chip package. They are the checkers around this work, not a copy of the result itself.
- physics-lint: One command that checks a folder of physics models against a fixed set of named physical rules, with findings straight into CI.
- maxwell-lint: Flags a coupling extractor whose answers no passive set of conductors could produce.
- sparam-lint: Is your signal-response model physically possible? Five physical laws checked from the command line.
- interval-core: The interval arithmetic core behind our proofs over whole families of layouts.
- touchstone-tools: Read, write and convert Touchstone files, the standard text files that record how signals pass through a package's connections, and refuse to write one that cannot be read back.
- physics-lint-mcp: The physics checks, callable by an AI agent.
- physics-lint-action: A GitHub Action that fails the build when a model breaks one of a fixed set of named physical rules.
- Signal-response validity corpus: A labelled corpus of physically invalid signal-response networks, and a scorer that grades any checker against it.
- screening-ceiling: The screening-ceiling family as an open dataset.