lemmaroutine
Finite buffered comparison test for sums
Finite test accepts every and only , with digits
fixed-precision comparison for the numerical interface
Lemma·
lem:arith-buffered-test·§10.5 Indexed ranks, sums, scales, and structural leavesStatement (paper)
Let , let and have known magnitude bound ,
and suppose . There is a deterministic
finite test that accepts every and accepts only .
It uses -place digit queries after normalization, with
intrinsic and suspended local lengths . Its result is
independent of any subsequent numerical output clock.
In plain English
For and with known magnitude bound and , there is a deterministic finite test that accepts every and accepts only . It uses -place digit queries after normalization, has intrinsic and suspended local lengths , and its outcome does not depend on any later numerical output clock.
Why it's here
This gives a comparison with a factor-two buffer: inputs clearly above are accepted, inputs at or below are rejected, and the gap in between may go either way. Because is bounded below by , finitely many digits suffice to decide, so the test can be used inside the numerical interface at fixed precision without depending on output clocks. It supplies the buffered tests that the uniform numerical interface relies on.
Classified routine:
A standard approximate-comparison argument: since u is bounded below by 2^{-κd} and the factor-two gap is wide, truncating s and u to O(d) binary places and comparing the approximations decides the buffered comparison. The only new part is the bookkeeping of the paper's own length and clock accounting.
Uses
Nothing else in the paper
Used by
Sources
PaperOpen
- Section
- §10.5 Indexed ranks, sums, scales, and structural leaves
- Page
- —
- Label
lem:arith-buffered-test- Line
sections/arithmetic.tex:698
Leannot aligned
Phase 2 aligns this node to a Lean declaration.