Documentation
¶
Overview ¶
Command float64table writes float64_table.go:
- the 128-bit truncated powers of five that bound Float64's leading-digit enclosure: each entry holds the leading 128 bits of 5^q, so 5^q = (M + ε)·2^s with 2^127 ≤ M < 2^128 and 0 ≤ ε < 1, where s = ⌊q·log2 5⌋ − 127;
- 5^(13j) and 2^(29j) as exact base-10^9 limbs, so an exact dyadic comparison expands its point with one short multiplication.
Usage: go run ./tools/float64table > float64_table.go
Click to show internal directories.
Click to hide internal directories.