Re: Patch to qp() to support extra arg specifying maximum iterations

[Joshua Redstone]

Hi Ben,
Yeah I forgot the patch.  It is attached now.
I was going to look at optimget/set, but I'm getting errors when i checked out the mercurial tree and tried compiling octave.  Where should I post to get help?
The errors are:

g++  -I. -I.. -I../liboctave -I../src -I../libcruft/misc  -DHAVE_CONFIG_H -mieee-fp -Wall -W -Wshadow -Wold-style-cast -g -O2 -rdynamic \
    -L..  -fPIC  -o octave \
    main.o  \
    -L../liboctave -L../libcruft -L../src -Wl,-rpath -Wl,/usr/local/lib/octave-3.1.51+ \
    -loctinterp -loctave  -lcruft   \
         \
        \
        \
     -lreadline  -lncurses -ldl -lm  -L/usr/lib/gcc/i486-linux-gnu/4.3.2 -L/usr/lib/gcc/i486-linux-gnu/4.3.2/../../../../lib -L/lib/../lib -L/usr/lib/../lib -L/usr/lib/gcc/i486-linux-gnu/4.3.2/../../.. -lgfortranbegin -lgfortran -lm
../liboctave/liboctave.so: undefined reference to `pthread_mutexattr_destroy'
../src/liboctinterp.so: undefined reference to `XOpenDisplay'
../liboctave/liboctave.so: undefined reference to `pthread_mutexattr_init'
../libcruft/libcruft.so: undefined reference to `ssymm_'
../liboctave/liboctave.so: undefined reference to `pthread_mutexattr_settype'
../libcruft/libcruft.so: undefined reference to `chemm_'
../libcruft/libcruft.so: undefined reference to `zhemm_'
../src/liboctinterp.so: undefined reference to `XScreenNumberOfScreen'
../libcruft/libcruft.so: undefined reference to `dsymm_'
collect2: ld returned 1 exit status

for comparison, octave-3.0.3 compiled fine for me.
Thanks,
Josh

On Thu, Feb 5, 2009 at 10:06 AM, Ben Abbott <address@hidden> wrote:

On Thursday, February 05, 2009, at 12:36PM, "Joshua Redstone" <address@hidden> wrote:

>On Wed, Feb 4, 2009 at 4:41 PM, Ben Abbott <address@hidden> wrote:
>
>>
>> On Feb 4, 2009, at 5:52 PM, John W. Eaton wrote:
>>
>>  On  1-Feb-2009, Ben Abbott wrote:
>>>
>>> | I tried to merge your changes, and modified the proper ChangeLog.
>>> | Using the final result, the first test fails for me (I've attached the
>>> | changeset).
>>>
>>> Instead of adding options as individual arguments, I think we should
>>> be using optimset.
>>>
>>> jwe
>>>
>>
>> Good point. I'll put it on my long list of things I hope to get to ;-)
>>
>> However, if someone (Josh?) has the initiative, no need to wait on me!
>>
>> Ben
>>
>I don't mind making a small change to use optimset, but I couldn't find any
>documentation on it except for some terse matlab stuff.
>In the meantime, I updated the test - the H is positive definite, so
>hopefully should be more robust across platforms, though I'm fuzzy on the
>qp() internals wrt math.
>Attached is diff for qp.m
>Josh
>

Josh, dd you forget the attachment?

Regarding optimset/get these have been added and improved recently. If you don't have mercurial setup already, you might try doing so. See the "Devepment Sources" section at the link below.

 http://www.gnu.org/software/octave/download.html

... or just download the files you'd like to change using the web interface.

 http://hg.savannah.gnu.org/hgweb/octave/file/38968f09f7ca/scripts/optimization/

To do was John suggests, the needed changed will be in qp.m (optimset should not need to be modified).

My understanding is that fsolve already supports what John suggests be done for qp.

Ben

qp3.diff
Description: Text Data