How make Integer linear programming optimization, intlinprog built-in function, takes binary data or even single as its input rather than double?

I am using intlinprog built-in function using binary decision variables and I run out of memory cause the matrices I am building are of type double. When I cast them as single, unit8 and so on I also get this error that matrices should be double. I am clearing my workspace of some temporary variables but I still run of memory. Is there any way to stay with only binary data rather than double to save up some memory?

Answer 1

[Matt J]

You could convert your matrices to type sparse, if they contain mostly zeros.

Answer 2

[John D'Errico]

Sorry. intlinprog requires doubles. According to doc intlinprog, even single is not an option. Just wanting software to do what it was not written to do is so rarely successful.

You could write your own solver of course. Or you could find a solver out there that allows you to provide input in the form you want, then learn how to call it.