The variables marked with the "D" are decision variables. This is

evident because they have index j3 equal to 1 and, by examining the

index outputs, we see that for j3 the value 1 must be a decision. This

output set assignment easily converts to a do-loop type iterative

solution procedure such as:

 C READ DECISIONS AND TEARS

 READ (5,5000) Y

 5000 FORMAT (10F8.0)

 10 Do 20 IF=l,2

 DO 20 11=1,2

 DO 20 12=1,2

 IF (IF .EQ. 1) X(I1,I2) = F(X,Y,I1,I2)

 IF (IF .EQ. 2) Y(I1,I2+1) = G(X,Y,I1,I2)

 20 CONTINUE

 C CHECK CONVERGENCE

 If y not converged, go to 10

The convergence check would probably check the change in the torn

variables y from iteration to iteration, convergence being defined as

an arbitrarily small change in y. The functions "F" and "G" would be

whatever f and g happen to be.

 Although these restrictions on the allowable output set assign-

ments have been necessary to facilitate the writing of iterative

solution procedures using do-loops, they do not weaken to any extent the

solution procedure deriving power of the algorithms which operate under

them. In fact these restrictions point the way, as it were, to the

types of output set assignments engineers traditionally have employed.

For example, in modeling a distillation column the material balance