Naming sets

It is customary in mathematical models to write index sets as 1,...,N or the like. Instead of translating this directly into Mosel code like the following:

 declarations
  x: array(1..N) of mpvar
 end-declarations

 sum(i in 1..N) x(i) >= 10

it is recommended to name index sets:

 declarations
  RI = 1..N
  x: array(RI) of mpvar
 end-declarations

 sum(i in RI) x(i) >= 10

The same remark holds if several loops or operators use the same intermediate set(s). Instead of

 forall(i in RI | isodd(i)) x(i) is_integer
 forall(i in RI | isodd(i)) x(i) <= 5
 sum(i in RI | isodd(i)) x(i) >= 10

which calculates the same intermediate set of odd numbers three times, it is more efficient to define this set explicitly and calculate it only once:

 ODD:= union(i in RI | isodd(i)) {i}

 forall(i in ODD) x(i) is_integer
 forall(i in ODD) x(i) <= 5
 sum(i in ODD) x(i) >= 10

Alternatively, loops of the same type and with the same index set(s) may be regrouped to reduce the number of times that the sets are calculated:

 forall(i in RI | isodd(i)) do
  x(i) is_integer
  x(i) <= 5
 end-do
 sum(i in RI | isodd(i)) x(i) >= 10