[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Bug-gsl] [bug #47347] Complex tangent failure for -1000i
|
From: |
Yu Liu |
|
Subject: |
[Bug-gsl] [bug #47347] Complex tangent failure for -1000i |
|
Date: |
Mon, 12 Sep 2016 04:51:09 +0000 (UTC) |
|
User-agent: |
Mozilla/5.0 (X11; CrOS x86_64 8350.68.0) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/52.0.2743.116 Safari/537.36 |
Follow-up Comment #1, bug #47347 (project gsl):
Please see test_tan.c for trouble shooting process and gsl_complex_tan.diff
for a proposed fix.
1. This proposed fixed is based on 7141c7adb8100207ef01ff87a94d5179103fcebc
which add a branch in gsl_complex_tanh to handle a similar issue. In this
commit, tanh was modified to have a separate branch to handle large real part
in input. This is what gsl_complex_tan.diff is doing in essense. In
test_tan.c, this idea is implemented in `gsl_complex_tan_2` for testing
purposes.
2. glibc implementation
(https://github.com/lattera/glibc/blob/a2f34833b1042d5d8eeb263b4cf4caaea138c4ad/math/s_ctan.c)
converts everything to exponential when the imagineary part is big. Is this a
better implementation?
_______________________________________________________
Reply to this item at:
<http://savannah.gnu.org/bugs/?47347>
_______________________________________________
Message sent via/by Savannah
http://savannah.gnu.org/