[Ipopt] Error in eq_mult_calculator

Johan Akesson jakesson at control.lth.se
Wed Sep 9 01:58:19 EDT 2009


Hi Andreas,

Thanks for your reply.

> 
> The output you sent is from a setting where there are no degrees of 
> freedom: you have as many equality constraints as you have free 
> variables. This means, all Ipopt will do is try to find a feasible point.
> 
This is indeed my intention: I'm solving a square problem which 
corresponds to "simulating" the plant over a fixed interval. I sometimes 
  solve such problems prior to solving the actual optimization problem. 
The issue here is that I am generating C code from Modelica model 
descriptions using the jmodelica.org compiler. Since the compiler is 
still under development I usually verify that I get the expected 
solution when solving a square problem which I can compare to simulation 
results derived in an independent software.

> The output shows that Ipopt does actually succeeds with that task, and 
> the final EXIT message confirms that.  The error message you see can be 
> safely ignored (all Ipopt is trying to do there is to compute some 
> constraint multipliers, but those don't mean much in this context).
> 

Ok, good.

Best
/Johan

> So, all is fine.  But you should keep in mind that you are actually not 
> optimizing if you have no degrees of freedom, you are just solving a 
> square nonlinear system of nonlinear equality constraints.
> 
> Regards,
> 
> Andreas
> 
> 
> On Mon, 7 Sep 2009, Johan Akesson wrote:
> 
>> Hello,
>>
>> I'm trying to solve a dynamic optimization problem discretized using
>> collocation with Ipopt. In some configurations (like the one below,
>> where I'm solving a square problem) I get the error
>>
>> "Cannot recompute multipliers for feasibility problem.  Error in
>> eq_mult_calculator"
>>
>> The problem may be badly scaled (optimization variables in the range 0.5
>> to 8e6). Before I start working rescaling the model, can anyone confirm
>> that poorly scaled systems can give this error?
>>
>> Best
>> /Johan
>>
>>
>>
>>
>> This is Ipopt version 3.5.4, running with linear solver ma27.
>>
>> Number of nonzeros in equality constraint Jacobian...:    36365
>> Number of nonzeros in inequality constraint Jacobian.:        0
>> Number of nonzeros in Lagrangian Hessian.............:        0
>>
>> Total number of variables............................:    13666
>>                      variables with only lower bounds:     2914
>>                 variables with lower and upper bounds:        0
>>                      variables with only upper bounds:        0
>> Total number of equality constraints.................:    13666
>> Total number of inequality constraints...............:        0
>>         inequality constraints with only lower bounds:        0
>>    inequality constraints with lower and upper bounds:        0
>>         inequality constraints with only upper bounds:        0
>>
>> iter    objective    inf_pr   inf_du lg(mu)  ||d||  lg(rg) alpha_du
>> alpha_pr  ls
>>    0  1.0000000e+00 1.30e+01 1.00e+00   0.0 0.00e+00    -  0.00e+00
>> 0.00e+00   0
>>    1  1.0000000e+00 1.78e-05 1.14e+01  -2.6 4.21e+04    -  9.90e-01
>> 1.00e+00h  1
>>    2  1.0000000e+00 1.18e-05 6.11e+01  -4.6 1.46e+04    -  1.00e+00
>> 1.00e+00h  1
>>    3  1.0000000e+00 7.59e-06 1.74e+02  -5.2 9.91e+03    -  1.00e+00
>> 1.00e+00h  1
>>    4  1.0000000e+00 4.26e-06 2.06e+02  -7.2 6.48e+03    -  1.00e+00
>> 1.00e+00h  1
>>    5  1.0000000e+00 2.23e-06 1.94e+02 -11.0 6.95e+03    -  1.00e+00
>> 1.00e+00h  1
>>    6  1.0000000e+00 7.14e-07 6.74e+02 -11.0 3.87e+04    -  1.00e+00
>> 1.00e+00h  1
>>    7  1.0000000e+00 1.31e-07 3.96e-01 -11.0 2.38e+03    -  1.00e+00
>> 1.00e+00h  1
>>    8  1.0000000e+00 1.61e-07 2.61e-02 -11.0 3.08e+03    -  1.00e+00
>> 1.00e+00h  1
>>    9  1.0000000e+00 1.56e-07 6.28e-05 -11.0 2.90e-01    -  1.00e+00
>> 1.00e+00H  1
>> iter    objective    inf_pr   inf_du lg(mu)  ||d||  lg(rg) alpha_du
>> alpha_pr  ls
>>   10  1.0000000e+00 1.58e-07 4.71e-05 -11.0 2.93e-03    -  1.00e+00
>> 2.50e-01h  3
>>   11  1.0000000e+00 1.19e-07 2.66e-09 -11.0 8.34e-04    -  1.00e+00
>> 1.00e+00h  1
>>   12  1.0000000e+00 1.19e-07 1.75e-08 -11.0 1.09e-03    -  1.00e+00
>> 1.00e+00H  1
>>   13  1.0000000e+00 1.19e-07 2.02e-08 -11.0 2.78e-03    -  1.00e+00
>> 1.25e-01h  4
>>   14  1.0000000e+00 1.19e-07 4.89e-08 -11.0 3.02e-03    -  1.00e+00
>> 2.50e-01h  3
>>   15  1.0000000e+00 1.19e-07 5.53e-08 -11.0 2.02e-03    -  1.00e+00
>> 1.25e-01h  4
>>   16  1.0000000e+00 1.19e-07 5.54e-08 -11.0 3.48e-03    -  1.00e+00
>> 1.95e-03h 10
>> Cannot recompute multipliers for feasibility problem.  Error in
>> eq_mult_calculator
>>
>> Number of Iterations....: 16
>>
>>                                    (scaled)                 (unscaled)
>> Objective...............:   1.0000000000000000e+00    
>> 1.0000000000000000e+00
>> Dual infeasibility......:   5.5443463001490997e-08    
>> 5.5443463001490997e-08
>> Constraint violation....:   1.1920928955078125e-07    
>> 4.4305623703166930e-06
>> Complementarity.........:   0.0000000000000000e+00    
>> 0.0000000000000000e+00
>> Overall NLP error.......:   1.1920928955078125e-07    
>> 4.4305623703166930e-06
>>
>>
>> Number of objective function evaluations             = 71
>> Number of objective gradient evaluations             = 18
>> Number of equality constraint evaluations            = 71
>> Number of inequality constraint evaluations          = 0
>> Number of equality constraint Jacobian evaluations   = 18
>> Number of inequality constraint Jacobian evaluations = 0
>> Number of Lagrangian Hessian evaluations             = 0
>> Total CPU secs in IPOPT (w/o function evaluations)   =      1.515
>> Total CPU secs in NLP function evaluations           =      3.439
>>
>> EXIT: Solved To Acceptable Level.
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> -- 
>> Johan Åkesson, PhD
>> Deptm. of Automatic Control    E-mail: johan.akesson at control.lth.se
>> Faculty of Engineering         WWW:    www.control.lth.se/user/jakesson
>> Lund University                Phone:  +46 46 2228797
>> Box 118                        Fax:    +46 46 138118
>> 221 00  LUND
>>
>> _______________________________________________
>> Ipopt mailing list
>> Ipopt at list.coin-or.org
>> http://list.coin-or.org/mailman/listinfo/ipopt
>>
>>
>>


-- 
Johan Åkesson, PhD
Deptm. of Automatic Control    E-mail: johan.akesson at control.lth.se
Faculty of Engineering         WWW:    www.control.lth.se/user/jakesson
Lund University                Phone:  +46 46 2228797
Box 118                        Fax:    +46 46 138118
221 00  LUND



More information about the Ipopt mailing list