Re: [boost] Proposal: Safe Double

From: Dave Steffen (steffend_at_[hidden])
Date: 2000-04-26 21:03:39

Kevin Atkinson writes:
 > It seams like safe comparison should really be an operation of the
 > FPU.....

Didn't I see a discussion in this forum a while back of "limit" or
"bound" based computation (I can't remember the exact term)? The idea
being, that every value is really _two_ double-precision numbers,
which represent the upper and lower bounds of what the computation
could have / should have / ought to have produced?

That seems to me like a better way of expressing what you want, on the
principle that what you _really_ want isn't _really_ possible on
finite-precision machines, right? ;-) 

Or did I really miss the point?

--------------------------------------------------------------------------
Dave Steffen                      Wave after wave will flow with the tide
Dept. of Physics                    And bury the world as it does
Colorado State University         Tide after tide will flow and recede
steffend_at_[hidden]        Leaving life to go on as it was...
                                                        - Peart / RUSH
"The reason that our people suffer in this way.... 
is that our ancestors failed to rule wisely".   -General Choi, Hong Hi