The matrix is input as an N rows (data) by M columns (variables) PG. On return, the matrix is decomposed to A = U * W * Vt. If M<N, then an error will be registered. In this case, augment the "A" PG with rows of zero values. The input matrices must be A[M,N], U[M.N] and V[N,N]. The length of the W VV is set by sSVDecompose_PGU to N. The Pagers must be type &lt;define&gt;GS_DOUBLE&lt;/define&gt;. Terminates if: U is not M by N. (Taken from size of A) V is not N by N. (Taken from #columns in A). PGs, VV are not &lt;define&gt;GS_DOUBLE&lt;/define&gt;. Dummies are treated as 0 values.

public static void SVRecompose(
	CGXNETCore gxNetShared,
	CPG pg_u,
	CVV vv_w,
	CPG pg_v,
	double min_w,
	CPG pg_a
)

Public Shared Sub SVRecompose ( 
	gxNetShared As CGXNETCore,
	pg_u As CPG,
	vv_w As CVV,
	pg_v As CPG,
	min_w As Double,
	pg_a As CPG
)

public:
static void SVRecompose(
	CGXNETCore^ gxNetShared, 
	CPG^ pg_u, 
	CVV^ vv_w, 
	CPG^ pg_v, 
	double min_w, 
	CPG^ pg_a
)

static member SVRecompose : 
        gxNetShared : CGXNETCore * 
        pg_u : CPG * 
        vv_w : CVV * 
        pg_v : CPG * 
        min_w : float * 
        pg_a : CPG -> unit