Angus Leeming wrote:
Correct. Note that X is modified by the call to lu_factorize and y is modified by the call to lu_substitute.
Thanks. Unfortunately I'm stuck with boost-1.30.2 and it doesn't have lu_* in it. I have lu.hpp but I get compile errors with bcc-5.6.4 in lu.hpp hence why I wondered if there was a way to do it using ublas, but not lu_*.
If not, is there another way of solving this without using lu_*?
Sure. Use the matrix solution method of your choice. Does the MTL not interoperate with uBlas?
Don't know what MTL is. I have a look. Thanks
Well, if matrix X isn't too big, then the lu_ methods are just what you need.
X is typically ~500 by 5 (don't know if that counts as big or not, last time I did matrix work, we only used 3x3 and 4x4 so to me it seems pretty huge, but we were doing stuff by hand). I'll have to have a look around for another library if there is no way to do it with ublas without using lu_*. Thanks for your help Russell