claimroutineinferred_dep
Binary-search median is accurate
Binary-search upper median satisfies
produces the final deterministic approximation
Claim·§12.2 A dyadic approximation at the specified start·inside thmDeterministic logspace approximation of acceptance probability
Statement (paper)
Let be the number of retained encodings and . The binary search over outputs the least integer such that at least retained encodings satisfy , which is the upper median of the multiset counted with repetitions. Since more than three quarters of its members satisfy Equation , so does the median, and hence satisfies .
In plain English
With retained encodings and , the binary search over returns the upper median of the values , and satisfies .
Why it's here
Since more than three quarters of the retained encodings give an accurate dyadic value, the median must also be accurate, so a deterministic median computation over the enumerated list gives the required approximation. It combines the exact enumeration of retained encodings with the accuracy of the individual truncated values to finish the proof of the parent theorem.
Classified routine:
Standard median-amplification argument: if more than half (here three quarters) of the values lie in an interval, the median does too, and a logspace binary search over counts finds it. It follows directly from the accuracy claim (claim-thm-fixed-precision-3) and the enumerability claim (claim-thm-fixed-precision-4).
Uses
Used by
Nothing in the paper
Sources
PaperOpen
- Section
- §12.2 A dyadic approximation at the specified start
- Page
- —
- Label
—- Line
sections/conclusion.tex:230
Leannot aligned
Phase 2 aligns this node to a Lean declaration.
Friction
- infoinferred_depoutline: "Binary-search median is accurate" rests on claim-thm-fixed-precision-3
- infoinferred_depoutline: "Binary-search median is accurate" rests on claim-thm-fixed-precision-4