claimroutineinferred_dep

Many light indices force collisions

Many light indices force ∑na2≥MHLH\sum n_a^2\ge MHL_H, giving saving −cLH-cL_H
collision case of the term dichotomy
Claim·§3.4 A comparison with two alternatives·inside propDeterminant comparison with two savings

Statement (paper)

Suppose that, for a choice of terms with indices aρa_\rho, at least ηM\eta M of the indices obey wa≤AHwa\le AH. Then ∑ana2≥(ηM)2/DA(H)=MHLH\sum_a n_a^2\ge (\eta M)^2/D_A(H)=MHL_H, and, after division by MHMH, the bound given by Lemma  for this term choice is at most −cLH+Ean+ρH-cL_H+E_{\rm an}+\rho_H.

In plain English

If, for some choice of terms, at least ηM\eta M of the indices aρa_\rho satisfy wa≤AHwa\le AH, then ∑ana2≥(ηM)2/DA(H)=MHLH\sum_a n_a^2\ge(\eta M)^2/D_A(H)=MHL_H. After dividing by MHMH, the collision lemma's bound for that term choice is at most −cLH+Ean+ρH-cL_H+E_{\rm an}+\rho_H.

Why it's here

Light indices live in the simplex wa≤AHwa\le AH, which has only DA(H)D_A(H) lattice points, so by pigeonhole (Cauchy–Schwarz) many rows must share indices, giving a large ∑ana2\sum_a n_a^2. Feeding this into the collision lemma yields the −cLH-cL_H saving for these term choices, which is the first of the two cases the comparison proposition needs to beat the arithmetic lower bound.
Classified routine:
Cauchy–Schwarz applied to the at most DA(H)D_A(H) light indices gives the collision count; substituting it into the earlier det-collision lemma yields the stated saving.

Sources

PaperOpen
Section
§3.4 A comparison with two alternatives
Page
—
Label
—
Line
sections/determinant.tex:393
Leannot aligned

Phase 2 aligns this node to a Lean declaration.

Friction

  • info
    inferred_dep
    outline: "Many light indices force collisions" rests on claim-det-comparison-1
  • info
    inferred_dep
    outline: "Many light indices force collisions" rests on det:collision