claimroutineinferred_dep
Clock gap pays band path length
Band path cost is at most , giving linear clock-gap storage
yields the storage bound of the multiplication lemma
Claim·§10.3 Multiplication by minimum-index bands·inside lemMultiplication with linear clock-gap storage
Statement (paper)
For an original input digit at an operand place in a positive band , the full path cost through the bands up to , which is at most for a fixed constant , is bounded by for fixed constants ; the zero band has constant depth and only constant overshoot. Consequently the suspended storage at is , and every intrinsic path and local word width is .
In plain English
For an input digit at operand place in a positive band , the total path cost through bands up to (at most ) is bounded by , and the zero band costs only a constant. Hence suspended storage at is and all intrinsic path and local word widths are .
Why it's here
This is the accounting step that ties the band circuit's path length to the clock gap , so the long product expansion costs no more than a linear term in . Combined with the path-only storage bound of the depth-first evaluation lemma and the per-band depth and width bounds from the gate-indexing claim, it gives the storage bound stated in the multiplication lemma.
Classified routine:
A bookkeeping step inside the multiplication lemma: it combines the per-band depth and width bounds from the gate-indexing claim with the path-only storage bound of depth-first circuit evaluation, using the fact that band index D grows linearly in the clock gap t-j.
Uses
Used by
Nothing in the paper
Sources
PaperOpen
- Section
- §10.3 Multiplication by minimum-index bands
- Page
- —
- Label
—- Line
sections/arithmetic.tex:382
Leannot aligned
Phase 2 aligns this node to a Lean declaration.
Friction
- infoinferred_depoutline: "Clock gap pays band path length" rests on claim-lem-arith-multiplication-4