Subscript triplets facilitate the selection of subsets of arrays, called array sections, which are often required in array assignments. Subscript triplets take the form

      start index : end index : stride

indicating the first element, last element and the jump between each element. This can be a really powerful shorthand in programs. For example, selecting the interior (non-boundary) points from an array becomes much simpler when expressed in Fortran 90's triplet notation:

      FORTRAN 77                Fortran 90 

      REAL A(10,10),B(10,10)    REAL, DIMENSION(10,10) :: A,B
      DO 1 J=1,8                A(1:8,1:8) = B(2:9,2:9)
        DO 2 I=1,8
          A(I,J) = B(I+1,J+1)
   2    CONTINUE
   1  CONTINUE

Array constants can be defined using this notation:

 REAL, DIMENSION(1024) :: a 
 a(1:1024) = 1.0        ! same as a = 1.0
 a(:1024:2) = 2.0       ! a(1)=a(3)=...= 2.0

Another option is to use the array constructor notation:

 a = (/ (i,i=1,1024) /)   ! a(1)=1,a(2)=2,...
 a = (/ (i,i=1,1024,4) /) ! a(1)=1,a(5)=5,...

Note that arrays in F90 expressions should be conformable, that is, they should have the same rank (number of dimensions) and the same extent in every dimension. Scalars are conformable with any array.