[ADOL-C] Incorrect calculus of hessian

Christian ChristianLupus at gmx.de
Tue Jun 28 15:23:54 EDT 2011


Hello to everybody,

I am new to the comunity of programmers wit mathematical background. But now I 
needed an optimization over several variables and therefore I looked at ipopt 
and adolc.

I have now written a program to optimize a control problem. In fact, if I use 
the quasi-Newton approch and the first order derivatives calculated with adolc 
all goes right.

If I enable the calculus of the hessian and the derivative checker of ipopt, I 
getthe result that there seems to be seveeral errors in the hessian.

Ipopt tells me that only the constraints 12 to 35 are wrong. The others seems 
to be ok.Therefoe I guess that the problem is not the initerface with the rest 
of the code.

I have given the code here:

class decription: http://codepad.org/rRYUOTCT
implementation: http://codepad.org/VPOo6Ohz
additional for make it runnable: http://codepad.org/mHIfBz3k

I know some of the parameters does not make sense (e.g. 50 iterations might 
semm a bit low,...) but I wanted just to test the derivatove generation of 
adolc

My system:
adolc-2.2.1
ipopt-3.9.3
gcc-4.4.5
Debian Sqeueeze system

OK, I hope, you can help me

Thanks in advance
Christian
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
Url : http://list.coin-or.org/pipermail/adol-c/attachments/20110628/b1df42e3/attachment.bin 


More information about the ADOL-C mailing list