Re: [Help-glpk] Building set of sets from plain set

[Andrew Makhorin]

>> The idea is to introduce a special operator, which is equivalent to
>> the following:

>> T[*,*] := empty;
>> for all (i,j,k,l,m) in S do
>>     T[k,i] := T[k,i] union {(j,m,l)};

>> Any suggestions about the syntax?

> Something like follows:

> set T{k in K, i in I}, data S : 3 1 : 2 5 4;

> that means that 3rd and 1st components of 5-tuple in S correspond,
> resp., to 1st and 2nd indices of T while 2nd, 5th, and 4th components
> of the 5-tuple give 3-tuple included in a plain set assigned to
> corresponding member of T.

> The statement above can be implemented as if there were data for T
> in the data section.

Just a comment.

Suppose you deal with a flow network G = (V, E), where V is a set
of nodes and E in V x V is a set of (directed) arcs. In MathProg this
may look as follows:

set V;
set E within V cross V;

To specify the flow conservation constraints you have to write:

s.t. foo{i in V}: sum{(i,j) in E} x[i,j] - sum{(j,i) in E} x[j,i] = 0;

and if E is large, evaluation of all foo[i] takes too long time.

Using the proposed feature the same constraints might be written as
follows:

set OUT{i in V}, data E : 1 : 2;
/* OUT[i] is a set of nodes incident to node i thru outgoing arcs */

set IN{i in V}, data E : 2 : 1;
/* IN[i] is a set of nodes incident to node i thru incoming arcs */

s.t. foo{i in V}: sum{j in OUT[i]) x[i,j] - sum{j in IN[i]} x[j,i] = 0;

In the latter case all foo[i] are evaluated efficiently.