[Coin-ipopt] primal-dual feasible point

Adela Pagès Bernaus pages25300 at hotmail.com
Wed Jan 16 07:28:33 EST 2008

Dear all,
Looking at the iterations output I've seen that usually the complementarity, lg(mu), is very small in comparison to primal and dual infeasibilities. My experience from a self-coded IPM in Matlab was that these three values should have the same order of magnitude. Is there any option in Ipopt which first tries to reduce primal and dual infeasibility?
PS: I'm still working on finding a good warm start point which I suspect that should be primal and dual feasible but I don't care much about complementarity. 
PS2: An example of the output I get:
iter objective inf_pr inf_du lg ( mu ) || d || log ( rg ) alpha_du alpha_pr ls     
70r-2.7020846e+08 1.16e+07 8.92e+02 -4.5 1.36e+05 -2.2 3.49e-01 7.04e-02f 1   
71r-2.7020848e+08 1.16e+07 8.92e+02 -4.5 5.14e+04 -1.8 1.00e+00 6.90e-06f 1 
72r-2.7321264e+08 1.16e+07 8.91e+02 -4.5 1.53e+05 -2.2 5.10e-01 1.68e-01f 1  
73r-2.7321269e+08 1.16e+07 8.91e+02 -4.5 5.77e+04 -1.8 1.00e+00 1.37e-05f 1 
74r-2.7321659e+08 1.16e+07 8.91e+02 -4.5 1.72e+05 -2.3 5.35e-01 1.65e-04f 1 
Here primal infeasibility is 10^7 and complementarity is 10^-5, and I suspect that we are spending too much iterations trying to leave this situation (the primal steps are very small): until iteration 923 primal infeasibility swings around 10^6 and in 10 iterations goes to 10^-6. 
Express yourself instantly with MSN Messenger! Download today it's FREE!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/ipopt/attachments/20080116/152bcbc2/attachment.html 

More information about the Coin-ipopt mailing list