<div dir="ltr">Hi Christian,<div><br></div><div>From the result, I didn't see any incorrectness. In B=HS, there H is the whole Hessian, not just the lower half.</div><div>When recovering H from B, using direct recovery guarantees that every entry in H can be directly copied from an entry in B. ColPack handles the details. I don't think there is a simple relation one can say other than that. </div><div><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Wed, Nov 5, 2014 at 8:09 AM, Christian Wolf <span dir="ltr"><<a href="mailto:ChristianLupus@gmx.de" target="_blank">ChristianLupus@gmx.de</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hello,<br>
<br>
I am trying to build/use ADOL-C and I am running into the following issues:<br>
<br>
1. I built colpack with MPI (--enable-openmp) support. To allow ADOL-C to compile I have to use<br>
> LIBS=-lgopm ./configure ....<br>
2. The documentation seems (at least with respect to the sparse driver) a bit outdated/buggy.<br>
3. (The main bug (?)) As far as I understand things the compresses hessian matrix (called for example "B") must fullfill the equation B = H S where S is the seed matrix. Please correct me if this is not correct.<br>
<br>
I tried with various builds (2.4.1 and trunk) to compile and run the program ADOL-C/examples/additional_<u></u>examples/sparse/sparse_<u></u>hessian. Just by looking at the output (no modifications to the SVN version) of the default output I get<br>
<br>
Sparsity pattern of Hessian:<br>
0: 0 1<br>
1: 0 1 4<br>
2: 2 5<br>
3: 3<br>
4: 1 4 5<br>
5: 2 4 5<br>
<br>
Seed matrix<br>
<br>
0: 1.0000 0.0000 0.0000 0.0000<br>
1: 0.0000 1.0000 0.0000 0.0000<br>
2: 1.0000 0.0000 0.0000 0.0000<br>
3: 1.0000 0.0000 0.0000 0.0000<br>
4: 1.0000 0.0000 0.0000 0.0000<br>
5: 0.0000 0.0000 1.0000 0.0000<br>
<br>
<snip><br>
<br>
H<br>
<br>
0: 1.0000 0.0000 0.0000 0.0000 0.0000 0.0000<br>
1: 5.0000 3.5787 0.0000 0.0000 0.0000 0.0000<br>
2: 0.0000 0.0000 1.8000 0.0000 0.0000 0.0000<br>
3: 0.0000 0.0000 0.0000 -0.9602 0.0000 0.0000<br>
4: 0.0000 -0.5820 0.0000 0.0000 95.4314 0.0000<br>
5: 0.0000 0.0000 22026.4658 0.0000 79.4007 88169.4721<br>
<br>
compressed H: (Here called B)<br>
<br>
0: 1.0000 5.0000 0.0000 0.0000<br>
1: 4.4180 3.5787 0.0000 0.0000<br>
2: 1.8000 0.0000 22026.4658 0.0000<br>
3: -0.9602 0.0000 0.0000 0.0000<br>
4: 95.4314 -0.5820 79.4007 0.0000<br>
5: 22105.8665 0.0000 88169.4721 0.0000<br>
<br>
Clearly the sparsity pattern is correct (while H shows only the lower triangle). On manually evaluating the value B[1][0] I get a value of 5.<br>
<br>
In [1] the author writes that there are several ways to choose the seed matrtix. At the moment it seems to be constructed that the i-th row of B and the j-th column of the seed matrix must be multiplyed (matrix or scalar product) to obtain H[i][j]. This is (as far as I think) related to the fact that at the moment the direct approach is used (only permutation of identy matrix).<br>
Can I rely on this fact that there will be no change in the future to the default behavior?<br>
<br>
Thanks a lot<br>
Christian<br>
<br>
<br>
[1] "Computing Sparse Hessians with Automatic Differentiation" from ANDREA WALTHER at the Institute of Scientific Computing, Technische Universität Dresden<br>
<br>
______________________________<u></u>_________________<br>
ADOL-C mailing list<br>
<a href="mailto:ADOL-C@list.coin-or.org" target="_blank">ADOL-C@list.coin-or.org</a><br>
<a href="http://list.coin-or.org/mailman/listinfo/adol-c" target="_blank">http://list.coin-or.org/<u></u>mailman/listinfo/adol-c</a></blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature"><div dir="ltr">Mu Wang, Ph.D. Candidate<div><div>HASS 257, Department of Computer Science</div><div>Purdue University -- West Lafayette</div></div><div><a href="mailto:wangmu0701@gmail.com">wangmu0701@gmail.com</a></div></div></div>
</div>