[Ipopt] solving cops-3 problems with ipopt

[Dieter Kraft]

Andreas Waechter wrote:
> Hi Dieter,
>
> I have not used the COPS scripts, so I don't know how to use cops_solve.
>
> However, what I see is that you let AMPL write a file called 
> ipopt.opt, and it will contain the line
>
> "option ipopt_options 'outlev 1 maxit 5000 max_cpu_time 1800';
>
> But ipopt.opt is the Ipopt options file, which is read by the Ipopt 
> code (unless you change its name using the Ipopt/AMPL option 
> "option_file_name"), and only works if you have in each line an Ipopt 
> options keyword with the assigned value.  Note that outlev and maxit 
> are options for Ipopt only available from AMPL (for compatibility 
> reasons). So, instead your ipopt.opt file could read
>
> print_level 7
> max_iter 5000
> max_cpu_time 1800
>
> I hope this helps,
Thanks, Andreas,
yes, this is a solution.
Regards
Dieter

>
> Andreas
>
> On Sun, 16 Aug 2009, Dieter Kraft wrote:
>
>> Hi all subscribers, hi Andreas,
>>
>> I want to solve cops-3 problems with ipopt-3.7.0 using
>> $COPS_HOME/bin/cops_solve
>> I copied $COPS_HOME/solvers/minos/initial to
>> $COPS_HOME/solvers/ipopt/initial and modified it to
>>
>> #!/usr/bin/perl -w
>>
>> open OPTIONS, "> ipopt.opt";
>> print OPTIONS "option ipopt_options 'outlev 1 maxit 5000 max_cpu_time
>> 1800';\n";
>> close OPTIONS;
>>
>> What I get is the following output in
>> $COPS_HOME/output/model/camshape.par1.ipopt.A0.R0.out
>>
>>
>> Presolve eliminates 3 constraints.
>> Adjusted problem:
>> 50 variables, all nonlinear
>> 100 constraints; 247 nonzeros
>>     51 nonlinear constraints
>>     49 linear constraints
>> 1 linear objective; 50 nonzeros.
>>
>> Ipopt 3.7.0: outlev 1
>> maxit 5000
>> max_cpu_time 1800
>>
>> Exception of type: OPTION_INVALID in file "IpOptionsList.cpp" at line 
>> 638:
>> Exception message: Read Option: "option". It is not a valid option. 
>> Check
>> the list of available options.
>> ampl_ipopt.cpp: Error in second Initialize!!!!
>> <BREAK>
>>
>> COPS runtime: 0
>>
>> which is not what I expect nor want
>>
>> Thanks for any help
>> Dieter
>>
>> -- 
>> Dieter Kraft
>> Munich University of Applied Sciences
>> kraft at hm.edu
>> www.lrz-muenchen.de/~dkraft
>>
>> _______________________________________________
>> Ipopt mailing list
>> Ipopt at list.coin-or.org
>> http://list.coin-or.org/mailman/listinfo/ipopt
>>
>>
>


-- 
Dieter Kraft
kraft at hm.edu
www.lrz-muenchen.de/~dkraft