This is the mail archive of the
glibc-bugs@sourceware.org
mailing list for the glibc project.
[Bug math/15899] exp2f() is not monotonic
- From: "joseph at codesourcery dot com" <sourceware-bugzilla at sourceware dot org>
- To: glibc-bugs at sourceware dot org
- Date: Tue, 27 Aug 2013 23:23:54 +0000
- Subject: [Bug math/15899] exp2f() is not monotonic
- Auto-submitted: auto-generated
- References: <bug-15899-131 at http dot sourceware dot org/bugzilla/>
http://sourceware.org/bugzilla/show_bug.cgi?id=15899
--- Comment #3 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
I don't think monotonicity is a practical goal for normal floating-point
implementation techniques (a sequence of intermediate computations in the
same type as the function concerned, typically using a polynomial
approximation, where the total error accumulation can be bounded but
rounding errors for different input values cannot readily be related to
each other) without use of precision extension to achieve error bounds
such that monotonicity is implied by the error bounds alone. Unless you
have pointers to existing proofs of monotonicity in the literature for
such implementations, that do not rely on any sort of exhaustive search.
--
You are receiving this mail because:
You are on the CC list for the bug.