@Carl Love Yes, I figured that later... It's a pity, this first vector has hundreds of non zero entries. I guess Adri's answer gives probably the best result then. Anyway, thank you for your help!!

@Carl Love Yes, I figured that later... It's a pity, this first vector has hundreds of non zero entries. I guess Adri's answer gives probably the best result then. Anyway, thank you for your help!!

Thank you a lot, Carl.

This really is the best solution : it gives a list of solutions (a basis, I guess), instead of the awful vector given by LinearSolve, and one of them had only 3 non-zero entries. For comparison, setting all free parameters to 0 (after LinearSolve) with eval(c,indets(c)=0); gave a solution with 102 non-zero entries.On top of that, you learnt me several shortcuts I didn't know about :)

Thank you a lot, Carl.

This really is the best solution : it gives a list of solutions (a basis, I guess), instead of the awful vector given by LinearSolve, and one of them had only 3 non-zero entries. For comparison, setting all free parameters to 0 (after LinearSolve) with eval(c,indets(c)=0); gave a solution with 102 non-zero entries.On top of that, you learnt me several shortcuts I didn't know about :)

So far this method gave me a solution with no non zero coordinate, but whose every coordinate was really close to 0. What I am looking for is actually a solution with integer coordinates (for some reason I know that there must be one).

So far this method gave me a solution with no non zero coordinate, but whose every coordinate was really close to 0. What I am looking for is actually a solution with integer coordinates (for some reason I know that there must be one).

Thank you Adri, it works - though I had to remove the ~, which provoked an internal error.

Thank you Adri, it works - though I had to remove the ~, which provoked an internal error.

Thank you so much !
This eval function seems to be quite useful in many situations.
Have a nice day.

Thank you so much !
This eval function seems to be quite useful in many situations.
Have a nice day.

Thank you also for this answer, it is highly instructive ! I would thumb it up if I were allowed to.

Thank you also for this answer, it is highly instructive ! I would thumb it up if I were allowed to.

@Markiyan Hirnyk This is surprising. Though I had not tried linalg, since on later versions it is recommended to use LinearAlgebra.

@Markiyan Hirnyk This is surprising. Though I had not tried linalg, since on later versions it is recommended to use LinearAlgebra.

I wrote most parts of the program under maple 15 at my office, and then ran it at home, under maple 12. It seems that you're right : maple 15 does handle large Matrices, while 12 didn't. Thank you very much !