[Ipopt] Restoration Failed with one constaint only

[Lewis I]

Hi Andreas, 

Thank you so much for your reply!
It is not too last, because I still have the same problem now.

For your first answer (Linear Independence Constraint), if some of my constraints
are really linear dependent, what can i do?

And why sometime I can solve with these constraints, but sometimes fail (in
different pattern, but it generated from same set of basic constraint functions)?
Does it mean that linear dependence constraints may not 100% cause problems?

Sorry for asking some simple questions, I am a student who works on some
researches that need to use optimization tools.
Thank you so much.


Regards, 
Lewis


----- Original Message ----
From: Andreas Waechter <andreasw at watson.ibm.com>
To: Lewis I <lewis369lewis at yahoo.com>
Cc: ipopt at list.coin-or.org
Sent: Wed, February 3, 2010 12:15:14 AM
Subject: Re: [Ipopt] Restoration Failed with one constaint only

Hi Lewis,

I guess my answer is a little late, but maybe it still helps...

This error (restoration phase failed at almost feasible points) can have 
several causes:

1. Your problem is not satisfying the Linear Independence Constraint 
Qualification, i.e., at the point the algorithm is converging to, the 
gradients of the (active) constraints are linearly dependent.  If you say 
you have only one constraints, is the gradient of this constraints zero 
(or very small) at the point the algorithm fails?

2. Your problem is very ill-conditioned

3. You have NaN/Inf in your model

4. The linear solver has repeatedly fatal errors in that iteration.

Regards,

Andreas

On Sat, 16 Jan 2010, Lewis I wrote:

> Dear all,
>
> I have check my derivative with derivative checker, no problem found.
> I have finally solved it by setting the tol to 1e-6, and set the 
> mu_strategy to adaptive.What is happening?
> thx
>
> Regards,
> Lewis
> --- On Sat, 1/16/10, Lewis I <lewis369lewis at yahoo.com> wrote:
>
>> From: Lewis I <lewis369lewis at yahoo.com>
>> Subject: [Ipopt] Restoration Failed with one constaint only
>> To: ipopt at list.coin-or.org
>> Date: Saturday, January 16, 2010, 6:21 PM
>> Dear all,
>>
>> I have the following problems when I ran my problem,
>> my problem only hv 24 variables, 1 obj. function  and
>> 1 constraint.
>> The constraint is the function of 18 variables from that 24
>> variables.
>>
>>
>> Log:
>> ______________________________________________________________________
>>
>> 458 2.3928240e+003 2.01e+003 3.40e+008 -11.0
>> 3.29e+001    -  1.00e+000 1.00e+000w  1
>>  459 5.9069378e+003 1.66e+003 8.61e+002 -11.0
>> 2.13e+001    -  1.00e+000 1.00e+000w  1
>> Restoration phase is called at point that is almost
>> feasible,
>>   with constraint violation 7.310133e-013. Abort.
>>
>> Number of Iterations....: 459
>>
>>                
>>                
>>    (scaled)         
>>        (unscaled)
>> Objective................: 
>> 9.4569553206125093e+001   9.4569553206125093e+001
>> Dual infeasibility......: 
>> 9.1622460109465753e-001   9.1622460109465753e-001
>> Constraint violation....: 
>> 7.3101334986695766e-013   5.8207660913467407e-011
>> Complementarity.........: 
>> 0.0000000000000000e+000   0.0000000000000000e+000
>> Overall NLP error........: 
>> 2.2388529800870424e-005   9.1622460109465753e-001
>>
>>
>> Number of objective function evaluations   
>>          = 3081
>> Number of objective gradient evaluations   
>>          = 462
>> Number of equality constraint evaluations   
>>         = 3289
>> Number of inequality constraint evaluations   
>>       = 0
>> Number of equality constraint Jacobian
>> evaluations   = 461
>> Number of inequality constraint Jacobian evaluations = 0
>> Number of Lagrangian Hessian evaluations   
>>          = 0
>> Total CPU secs in IPOPT (w/o function
>> evaluations)   =      4.375
>> Total CPU secs in NLP function evaluations   
>>        =      0.016
>>
>> EXIT: Restoration Failed!
>> _______________________________________________________________
>> Regards,
>> Lewis
>>
>>
>>      
>>
>> _______________________________________________
>> Ipopt mailing list
>> Ipopt at list.coin-or.org
>> http://list.coin-or.org/mailman/listinfo/ipopt
>>
>
>
>
>
>
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt
>
>
>