matsolve

Function: matsolve
Section: linear_algebra
C-Name: gauss
Prototype: GG
Help: matsolve(M,B): solution of MX=B (M matrix, B column vector or matrix).
Doc: Let $M$ be a left-invertible matrix and $B$ a column vector
 such that there exists a solution $X$ to the system of linear equations
 $MX = B$; return the (unique) solution $X$. This has the same effect as, but
 is faster, than $M^{-1}*B$. Uses Dixon $p$-adic lifting method if $M$ and
 $B$ are integral and Gaussian elimination otherwise. When there is no
 solution, the function returns an $X$ such that $MX - B$ is nonzero
 although it has at least $\#M$ zero entries:
 \bprog
 ? M = [1,2;3,4;5,6];
 ? B = [4,6,8]~; X = matsolve(M, B)
 %2 = [-2, 3]~
 ? M*X == B
 %3 = 1
 ? B = [1,2,4]~; X = matsolve(M, [1,2,4]~)
 %4 = [0, 1/2]~
 ? M*X - B
 %5 = [0, 0, -1]~
 @eprog\noindent Raises an exception if $M$ is not left-invertible, even if
 there is a solution:
 \bprog
 ? M = [1,1;1,1]; matsolve(M, [1,1]~)
  ***   at top-level: matsolve(M,[1,1]~)
  ***                 ^------------------
  *** matsolve: impossible inverse in gauss: [1, 1; 1, 1].
 @eprog\noindent The function also works when $B$ is a matrix and we return
 the unique matrix solution $X$ provided it exists.

Variant: For integral input, the function
 \fun{GEN}{ZM_gauss}{GEN M,GEN B} is also available.

Function: _ZM_inv_worker
C-Name: ZM_inv_worker
Prototype: GG
Section: programming/internals
Help: worker for ZM_inv

Function: _ZabM_inv_worker
C-Name: ZabM_inv_worker
Prototype: GGG
Section: programming/internals
Help: worker for ZabM_inv