*   Automatically generated include file that initializes problem at a
*   feasible point, using observations.
*   1) Reset all variables with bounds and equations of importance

    option kill=pe;
    option kill=ce;
    option kill=se;
    option kill=x;
    option kill=F;
    option kill=h;
    option kill=dx;
    option kill=ds;
    option kill=cx;
    option kill=cs;
    option kill=ro;
    option kill=pi;
    option kill=z;

*   2) Restore bounds on transport costs

    ce.up(i,j) = 200 $ (AD(i,j) and (ord(i) lt ord(j)));

*   3) Starting point for costs is observation

    ce.l(i,j) $ (AD(i,j) and (ord(i) lt ord(j))) = c(i,j);
    se.l(i) = so(i);

*   4) Find a corresponding feasible price vector by solving TP problem

    solve TPmin using nlp minimising z;

*   pe.l(t,i) = fp + h.m(t,i) $ im(i);
    pe.l(t,i) = p("t1","i1") + h.m(t,i) $ im(i);

*   5) Initialise the first order conditions using duals of TP problem

    pi.l(t,i,j) $ AD(i,j)
        = ce.l(i,j) $ (ord(i) lt ord(j))
        + ce.l(j,i) $ (ord(i) gt ord(j)) + pe.l(t,i) - pe.l(t,j);

    ro.l(t,i) = se.l(i) - sum(tnext(t,t1), pe.l(t1,i)) + pe.l(t,i);