This model implements a compact subtour elimination constraint set. It is proposed by Miller, Tucker and Zemlin:
u(i)-u(j) + n x(i,j) <= n - 1
Although easy to formulate in GAMS, it is not very intuitive. This formulation is not suited for large models (the MIP takes a long time to solve).
The GAMS model first checks whether our 6 city problem from tsp1 gives the correct objective. Then a 13 city problem is solved (the first 13 out of 17 from the br17 model).