Marc:
The numerical library currently returns
(expt 0 1.)
0
(/ 0 1.)
0
(/ 0 1.+1.i)
0
This small patch has it return
(expt 0 1.)
0.
(/ 0 1.)
0.
(/ 0 1.+1.i)
0.+0.i
which is what I believe R6RS does.
Gambit now passes most of the arithmetic and trigonometric and transcendental function tests in the "Test Programs" part of
http://www.cesura17.net/~will/Professional/Standards/Scheme/R6RS/
The remaining differences are accuracy issues. I'll need to recode the complex inverse hyperbolic and inverse trigonometric routines as in Kahan, W. "Branch Cuts for Complex Elementary Functions, or Much Ado About Nothing's Sign Bit".
Brad
Afficher les réponses par date
Thanks for the patch.
Marc
On 2013-09-24, at 2:11 PM, Bradley Lucier lucier@math.purdue.edu wrote:
Marc:
The numerical library currently returns
(expt 0 1.)
0
(/ 0 1.)
0
(/ 0 1.+1.i)
0
This small patch has it return
(expt 0 1.)
(/ 0 1.)
(/ 0 1.+1.i)
0.+0.i
which is what I believe R6RS does.
Gambit now passes most of the arithmetic and trigonometric and transcendental function tests in the "Test Programs" part of
http://www.cesura17.net/~will/Professional/Standards/Scheme/R6RS/
The remaining differences are accuracy issues. I'll need to recode the complex inverse hyperbolic and inverse trigonometric routines as in Kahan, W. "Branch Cuts for Complex Elementary Functions, or Much Ado About Nothing's Sign Bit".
Brad <exact-zero.patch>
-
Bradley Lucier -
Marc Feeley