Does anyone have some reliable numbers regarding the maximum number of decision variables for bintprog? I am maxing out around 3600, with a Aeq constraint matrix - mostly zeros around the same size.

I am on WinXP 32bit - 64bit may help some I know but not an option as I am at work.

Thanks.

Hello Jeff,
What do you mean when you say "maxing out"? Do you run out of memory after 
several iterations or do you hit one of the algorithm limits without finding 
an integer feasible solution or something else?

Best Regards
Derya
"Jeff Baker" <jbaker75@gmail.com> wrote in message 
news:hggnqc$gvt$1@fred.mathworks.com...
> Does anyone have some reliable numbers regarding the maximum number of 
> decision variables for bintprog? I am maxing out around 3600, with a Aeq 
> constraint matrix - mostly zeros around the same size.
>
> I am on WinXP 32bit - 64bit may help some I know but not an option as I am 
> at work.
>
> Thanks. 

Hi Derya, I am not entirely sure. I do reach an out of memory error, but I believe it is on the first iteration. What is the largest number of decision variables you have used in practice. One problem I am interested in is a simple unit commitment algorithm (I work in the energy sector) - e.g. 100 possible generators, 36 hours - is it on or off, with additional startup, min on, max off constraints. That is very simple, but a relatively large number of variables. In reality you have more generators and are looking at a larger time frame.

"Derya Ozyurt" <dozyurt@mathworks.com> wrote in message <hgo367$fhh$1@fred.mathworks.com>...
> Hello Jeff,
> What do you mean when you say "maxing out"? Do you run out of memory after 
> several iterations or do you hit one of the algorithm limits without finding 
> an integer feasible solution or something else?
> 
> Best Regards
> Derya
> "Jeff Baker" <jbaker75@gmail.com> wrote in message 
> news:hggnqc$gvt$1@fred.mathworks.com...
> > Does anyone have some reliable numbers regarding the maximum number of 
> > decision variables for bintprog? I am maxing out around 3600, with a Aeq 
> > constraint matrix - mostly zeros around the same size.
> >
> > I am on WinXP 32bit - 64bit may help some I know but not an option as I am 
> > at work.
> >
> > Thanks. 
> 

Hi Jeff,
If you get an out of memory error, I suspect it happens during the 
pre-allocation. If so, 3600 variables corresponds roughly to 60 MB. When you 
type >>memory at the command line what is the "Maximum possible array size"? 
Is it around 60 MB?

In any case, bintprog can be effective for problems with much smaller size 
(possibly less than 500 variables).

Best Regards,
Derya


"Jeff Baker" <jbaker75@gmail.com> wrote in message 
news:hgpnf7$ku5$1@fred.mathworks.com...
> Hi Derya, I am not entirely sure. I do reach an out of memory error, but I 
> believe it is on the first iteration. What is the largest number of 
> decision variables you have used in practice. One problem I am interested 
> in is a simple unit commitment algorithm (I work in the energy sector) - 
> e.g. 100 possible generators, 36 hours - is it on or off, with additional 
> startup, min on, max off constraints. That is very simple, but a 
> relatively large number of variables. In reality you have more generators 
> and are looking at a larger time frame.
>
> "Derya Ozyurt" <dozyurt@mathworks.com> wrote in message 
> <hgo367$fhh$1@fred.mathworks.com>...
>> Hello Jeff,
>> What do you mean when you say "maxing out"? Do you run out of memory 
>> after several iterations or do you hit one of the algorithm limits 
>> without finding an integer feasible solution or something else?
>>
>> Best Regards
>> Derya
>> "Jeff Baker" <jbaker75@gmail.com> wrote in message 
>> news:hggnqc$gvt$1@fred.mathworks.com...
>> > Does anyone have some reliable numbers regarding the maximum number of 
>> > decision variables for bintprog? I am maxing out around 3600, with a 
>> > Aeq constraint matrix - mostly zeros around the same size.
>> >
>> > I am on WinXP 32bit - 64bit may help some I know but not an option as I 
>> > am at work.
>> >
>> > Thanks.
>>