1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
|
subroutine dsmsp(nrb,ncb,nca,b,mrb,a,nela,inda,c,mrc)
c multiply a full matrix stored in b on right by a sparse
c matrix stored in a,inda. Put result in c.
c*** input
c nrb actual number of rows in b.
c ncb actual number of columns in b and row of a
c nca actual number of columns in a and c
c b a two-dimensional array containing all the
c elements of the first matrix.
c mrb row-dimension of b in calling routine.
c a a one-dimensional array containing the non-zero elements
c of the first matrix,arranged row-wise, but not
c necessarily in order within rows.
c nela number of non-zero elements in a
c inda(i) 1<=i<=nra number of non-zero elements in row i of a.
c inda(nra+i) 1<=i<nela column index of i'th non-zero element of a.
c mrc row dimension of c in calling routine.
c***output
c c a two-dimensional array containing all the
c elements of the product matrix.
c!
c Copyright INRIA
double precision a(*), b(mrb,ncb), c(mrc,nca)
integer inda(*)
c nra,nca are number of rows,columns in a.
nra = ncb
c nrc,ncc are number of rows,columns in c.
nrc = nrb
ncc = nca
c clear c to zero.
10 do 30 i=1,nrc
do 20 j=1,ncc
c(i,j) = 0.0d0
20 continue
30 continue
c n2 will be pointer to end of row k of a.
n2 = 0
c k will be row-index for a.
do 60 k=1,nra
c pick out number of non-zeros in row k of a.
nir = inda(k)
if (nir.eq.0) go to 60
c if no non-zeros skip processing of row k.
c n1 will point to start of row k in a,n2 to end of row.
n1 = n2 + 1
n2 = n2 + nir
c process row k of a,i.e.
c form all products of b(i,k) with non-zero a(k,j),
c add into c(i,j)
c l points to non-zero elements in row k of a.
do 50 l=n1,n2
j = inda(nra+l)
do 40 i=1,nrb
c(i,j) = c(i,j) + b(i,k)*a(l)
40 continue
50 continue
60 continue
return
end
|
