[Coin-ipopt] Re: IPOPT doubts

Andreas Waechter andreasw at watson.ibm.com
Thu Feb 3 13:29:14 EST 2005


Hi Luciano,

I had a look at the output you sent.  (By the way, since this is going to
a mailing list, please don't attach large files to postings - you could
send large files to my direct email address, andreasw at watson.ibm.com,
even thought I usually prefer the mailing list for Ipopt-related
questions.)

It seems to me that you tried to specify the HSL library using the
--with-hsl=... flag for configure?  I suspect that you wanted to specify
something like

-L/Ipopt/OTHERS/hsl -lhsl

but you might have forgotten to include quotations?  In other words, you
should write

--with-hsl='-L/Ipopt/OTHERS/hsl -lhsl'

On the other hand, it seems that you want to use the HSL library that is
compiled by the Ipopt makefiles.  In that case, all you have to do is to
put the source files (usually ma27ad.f and mc19ad.f) into
Ipopt/OTHERS/HSL, and the configure script should find the source files
and compile them automatically.

Since you didn't include any HSL subroutine, you got the 97 error message.
If you put the source files for the Harwell routines into the right place
and run the configure script and make again, the example should work.

I hope it helps,

Andreas

Andreas Waechter
IBM T.J. Watson Research Center
P.O. Box 218 (Rt. 134)
Yorktown Heights, NY 10598
USA

On Thu, 3 Feb 2005, Ing. Luciano Salvietti Cignetti wrote:

> Hello, Mr. Watson
>
> I have this doubts regarding IPOPT:
>
> 1. I run the ./configure on the cygwin prompt and the message "configuration
> succesful appears" but I am not certain if I installed blas, lapack and hsl
> correctly because as you can see on image #2, on the attached file, it seems
> that it doesnt find the harwell libraries...
>
> 2. Also I was wondering if you could tell me a way to check if IPOPT is working
> correctly. I know there is a C example on the Examples directory, and I have
> also tried to run it but I get a 97 error, regarding the missing of MA47,
> which if not wrong is a harwell subroutine.
>
> Thanks for your time,
>
>
> Regards,
>
>
> Eng. Luciano Salvietti Cignetti
> Research Cathedra on Industrial Engineer
> Monterrey Institute of Technology
> Monterrey, Mexico
>
>
>




More information about the Coin-ipopt mailing list