hi @sshin23 just to fix what looks like a (very) minor issue as dt should T/N (or 1/N, depending in any case on N as, otherwise, changing the grid size implies changing the time horizon)

BTW: the mayer term in the cost is a bit strange (just looks like the reweighted final grid point of the Lagrange quadrature, but for the control):

    objective(c, 0.5 * Qf * (x[N+1, j] - d)^2 for (j, Qf, d) in itr2)

do you have a detailed ref. towards the problem?

cc: @amontoison