[ADOL-C] sparse_jac return unreasonable NaN's

[Andrea Walther]

Hi,

a quote from a following up email:

>  The second one is just below the first one. What's in it is setting option[3] to 0 can avoid this problem. Though I don't understand what's the connection between them.

If J is a Jacobian matrix one can either compute a compressed
representation using

J * B

with some so called seed matrix B and the forward mode of AD.
Alternatively, one may compute

W * J

with another seed matrix and the reverse mode of AD.


So far, we tested the row compression only on a limited set
of test problems. If setting option[3] to 0
solves your problem, the column compression works OK but there is
somewhere an error in the row compression which might be either in
ADOL-C or in ColPack.

As soon as possible I will have a detailed look at the example
that you send me to detect the part of the ADOL-C Colpack
coupling that causes the error.

Hope this helps for the time being. Otherwise feel free to email again.

Best regards,

Andrea Walther

-- 
Prof. Dr. Andrea Walther
Lehrstuhl fuer Mathematik und ihre Anwendungen
Institut fuer Mathematik
Universitaet Paderborn
Warburger Str. 100
33098 Paderborn

Email: andrea.walther at uni-paderborn.de
Phone: ++49 5251 602721
Fax:   ++49 5251 603728

**********