fcblok Subroutine

The routine supervises the PLU factorization with pivoting of
the scaled rows of an almost block diagonal matrix.

The almost block diagonal matrix is stored in the arrays
BLOKS and INTEGS.

The FACTRB routine carries out steps 1,..., LAST of Gauss
elimination, with pivoting, for an individual block.

The SHIFTB routine shifts the remaining rows to the top of
the next block.

14 February 2007

Carl DeBoor

Carl DeBoor,
A Practical Guide to Splines,
Springer, 2001,
ISBN: 0387953663,
LC: QA1.A647.v27.

Input/output, real ( kind = 8 ) BLOKS(*).  On input, the almost 
block diagonal matrix A to be factored.  On output, the
factorization of A.

Input, integer ( kind = 4 ) INTEGS(3,NBLOKS), describes the block 
structure of A.

Input, integer ( kind = 4 ) NBLOKS, the number of blocks in A.

Output, integer ( kind = 4 ) IPIVOT(*), which will contain pivoting 
information.  The dimension of IPIVOT is the sum ( INTEGS(1,1:NBLOKS) ).

Workspace, real SCRTCH(*), of length maxval ( integs(1,1:NBLOKS) ).

Output, integer ( kind = 4 ) IFLAG, error flag.
= 0,  in case matrix was found to be singular;
= (-1)**(number of row interchanges during factorization), otherwise.