kernel-Int.fib_cassini
- Kind
- kernel-term
- Status
- checked
Supports: `Int.fib_cassini` is admitted with exactly the type quoted in `formal.statement`, read directly out of the kernel's environment via the exact `render_lean` route `int_theorem_inventory` uses. No committed test in this repository asserts `Int.fib_cassini`'s statement verbatim the way `det2_fib_is_cassini_through_det2` does for `Rat.det2_fib` or the Wilson checker does for `Int.wilson` -- this evidence row is the closest available substitute: it re-derives the rendered type from the trusted kernel on every run and fails the moment that type changes, rather than trusting the doc comment in `fibonacci.rs`.
test "$(scripts/cargo-serialized.sh run --release -p axeyum-lean-kernel --example int_theorem_inventory -- fib_cassini 2>&1 | grep -Fc "((x0 : AxNat) -> Eq.{1} Int (Int.sub (Int.mul (Int.ofNat (AxNat.fib (AxNat.succ (AxNat.succ x0)))) (Int.ofNat (AxNat.fib x0))) (Int.mul (Int.ofNat (AxNat.fib (AxNat.succ x0))) (Int.ofNat (AxNat.fib (AxNat.succ x0))))) (Int.pow (Int.neg Int.one) (AxNat.succ x0)))")" -ge 1 Evidence notes
Run 2026-08-24 against this tree: exit 0. A deliberately wrong substring (a string that appears nowhere in the tool's output) run through the identical command exits 1, confirming the grep is not a vacuous pass. This is the strongest available check that does not require adding a new pinned test to `crates/`, which this lane's brief forbids touching.