matrixqz

Function: matrixqz
Section: linear_algebra
C-Name: matrixqz0
Prototype: GDG
Help: matrixqz(A,{p=0}): if p>=0, transforms the rational or integral mxn (m>=n)
 matrix A into an integral matrix with gcd of maximal determinants coprime to
 p. If p=-1, finds a basis of the intersection with Z^n of the lattice spanned
 by the columns of A. If p=-2, finds a basis of the intersection with Z^n of
 the Q-vector space spanned by the columns of A.
Doc: $A$ being an $m\times n$ matrix in $M_{m,n}(\Q)$, let
 $\text{Im}_\Q A$ (resp.~$\text{Im}_\Z A$) the $\Q$-vector space
 (resp.~the $\Z$-module) spanned by the columns of $A$. This function has
 varying behavior depending on the sign of $p$:

 If $p \geq 0$, $A$ is assumed to have maximal rank $n\leq m$. The function
 returns a matrix $B\in M_{m,n}(\Z)$, with $\text{Im}_\Q B = \text{Im}_\Q A$,
 such that the GCD of all its $n\times n$ minors is coprime to
 $p$; in particular, if $p = 0$ (default), this GCD is $1$.

 If $p=-1$, returns a basis of the lattice $\Z^n \cap \text{Im}_\Z A$.

 If $p=-2$, returns a basis of the lattice $\Z^n \cap \text{Im}_\Q A$.

 \misctitle{Caveat} ($p=-1$ or $-2$) For efficiency reason, we do not compute
 the HNF of the resulting basis.

 \bprog
 ? minors(x) = vector(#x[,1], i, matdet(x[^i,]));
 ? A = [3,1/7; 5,3/7; 7,5/7]; minors(A)
 %1 = [4/7, 8/7, 4/7]   \\ determinants of all 2x2 minors
 ? B = matrixqz(A)
 %2 =
 [3 1]

 [5 2]

 [7 3]
 ? minors(%)
 %3 = [1, 2, 1]   \\ B integral with coprime minors
 ? matrixqz(A,-1)
 %4 =
 [3 1]

 [5 3]

 [7 5]

 ? matrixqz(A,-2)
 %5 =
 [3 1]

 [5 2]

 [7 3]

 @eprog