Re: MacOS X 10.3.9 build but load-save.m test failure

John W. Eaton wrote:
> On 28-Apr-2005, Dmitri A. Sergatskov wrote:
>
> | EXEC: ../src/octave -f -q -p .:../src//:./../scripts// 
> ./octave.test/linalg/expm-2.m
> | 
> | Octave Output:
> | ans = 0
> | FAIL: octave.test/linalg/expm-2.m
>
> The test is:
>
>   arg = [1, 1; 0, 1];
>   result = [2.718281828459045, 2.718281828459045;
>             0.000000000000000, 2.718281828459045];
>   all (all (abs (expm (arg) - result) < 2*eps))

But the error I get is :

Octave Output:
error: eig: argument must be a square matrix
error: near line 1 of file `./octave.test/linalg/eig-5.m'
child process exited abnormally

???

> My guess is it is not computing a result to the expected precision.
>
> What is the max difference on your system?  If it is reasonably small
> (say < 256*eps or so) then probably we should just increase the
> tolerance in the test file.

The "make check" used to be OK up to 2.1.67 (the latest I run). Was the test 
changed lately?

Anyway:


octave:6> format long
octave:7> abs (expm (arg) - result)
ans =

   4.44089209850063e-16   4.44089209850063e-16
   0.00000000000000e+00   4.44089209850063e-16

octave:8> 2*eps
ans =  4.44089209850063e-16

> jwe

Dmitri.
--



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------