Sat, 12 Sep 2015 14:16:39 +0000 |
schloegl |
use macro ISNAN() instead of C++'s isnan() - because it supports all floating point formats not just double
octave-forge
|
Sat, 12 Sep 2015 07:15:01 +0000 |
schloegl |
resulting values should be really converted to output data type
octave-forge
|
Thu, 18 Jun 2015 15:09:49 +0000 |
schloegl |
check for sparse matrices and and convert to full if needed
octave-forge
|
Sat, 24 Jan 2015 01:20:11 +0000 |
schloegl |
address compiler warnings on functions with no return argument
octave-forge
|
Fri, 10 Jan 2014 13:34:15 +0000 |
schloegl |
update docu of TSA and NaN package: push version to 4.2.7 and 2.5.8 for TSA and NaN package, resp.
octave-forge
|
Fri, 11 Nov 2011 07:47:37 +0000 |
schloegl |
return output for case of no output variable
octave-forge
|
Wed, 27 Jul 2011 22:20:09 +0000 |
schloegl |
use only single test for (D1==1)
octave-forge
|
Wed, 20 Apr 2011 09:16:06 +0000 |
schloegl |
update contact e-mail and www address
octave-forge
|
Tue, 19 Apr 2011 07:26:15 +0000 |
schloegl |
move #PRAGMA OMP PARALLEL into #PRAGMA OMP PARALLEL FOR - this fixes SegFaulting on ML7.11
octave-forge
|
Tue, 15 Mar 2011 20:31:04 +0000 |
schloegl |
OpenMP support included
octave-forge
|
Thu, 27 Jan 2011 17:10:36 +0000 |
schloegl |
fix web address
octave-forge
|
Thu, 27 Jan 2011 16:39:17 +0000 |
schloegl |
replace <inttypes.h> with <stdint.h>
octave-forge
|
Wed, 28 Jul 2010 09:28:54 +0000 |
schloegl |
addresses undesired side-effect off in-place sorting of data
octave-forge
|
Sun, 10 Jan 2010 22:05:59 +0000 |
schloegl |
use *.cpp instead of *.c
octave-forge
|
Tue, 13 Oct 2009 22:43:28 +0000 |
schloegl |
fix version check for defining mwSize; set default ACC_LEVEL 0
octave-forge
|
Thu, 01 Oct 2009 12:13:51 +0000 |
schloegl |
default accuracy level set to 0; level 2 and 3 exchanged; added test to determine optimal accuracy level
octave-forge
|
Wed, 30 Sep 2009 22:18:27 +0000 |
schloegl |
any combination of naive and Kahan summation with double and extended can be controlled through flag_accuracy_level
octave-forge
|
Wed, 30 Sep 2009 08:18:02 +0000 |
schloegl |
increase accuracy by using Kahan's summation formula - this reduces the error by a factor of N; first tests show a penalty of about 40% in terms of speed; currently, this is only implemented for stride=1 (summation along 1st dimension)
octave-forge
|
Thu, 24 Sep 2009 16:09:34 +0000 |
schloegl |
extended accuracy for stride >1
octave-forge
|
Wed, 29 Jul 2009 23:14:03 +0000 |
schloegl |
more efficient (smaller) code; improve docu
octave-forge
|
Mon, 27 Jul 2009 20:29:27 +0000 |
schloegl |
avoid seg-fault in empty input argument
octave-forge
|
Tue, 30 Jun 2009 12:15:57 +0000 |
schloegl |
using internally long double for increased accuracy
octave-forge
|
Wed, 27 May 2009 21:51:02 +0000 |
schloegl |
support ml72 and earlier
octave-forge
|
Thu, 07 May 2009 09:33:46 +0000 |
schloegl |
NaN-tb: performance test of alternative algorithm, and optimization for speed; support complex matrices; bug fix in sumskipnan_mex
octave-forge
|
Wed, 06 May 2009 12:48:29 +0000 |
schloegl |
NaN-tb: avoid seg-fault (when size(x,DIM)=0, and empty arg3)
octave-forge
|
Wed, 06 May 2009 11:08:58 +0000 |
schloegl |
NaN-toolbox: support weighting of samples
octave-forge
|
Tue, 05 May 2009 10:06:21 +0000 |
schloegl |
fix type for size values
octave-forge
|
Mon, 20 Apr 2009 14:28:51 +0000 |
schloegl |
return sum even if no output argument
octave-forge
|
Tue, 24 Mar 2009 09:42:13 +0000 |
schloegl |
replace NaN-check (x==x) with (\!isnan(x))
octave-forge
|
Wed, 11 Mar 2009 08:23:44 +0000 |
schloegl |
minimize cache misses - increases performance; support for complex data moved to sumskipnan.m
octave-forge
|
Mon, 09 Mar 2009 11:25:34 +0000 |
schloegl |
mex support improved: tests show a significant advantage in terms of speed
octave-forge
|
Thu, 05 Mar 2009 16:05:10 +0000 |
schloegl |
move if-decisions outside of loops
octave-forge
|
Thu, 05 Mar 2009 12:42:02 +0000 |
schloegl |
rename sumskipnan such that it does not shadow sumskipnan.m
octave-forge
|