Closed jokochems closed 6 months ago
While in principle this is functional, there are some numerical issues here.
When the model is resolved for a large instance, the lp files contains rows that contain something like 0 = 0.0004 ...
Thus, ultimately this leads the solver to detect model infeasibility. I have to think about a suitable way to address this ...
I'll give it another shot using rounding of investment results and forcing them to be within bounds. If this fails as well, this will no longer be pursued.
It is still not working. The reason for that is that non-negativity seems to be not ensured using the given rounding precision. I'll have to take one more look into that, I guess.
I have an idea for an alternative solution:
nominal_value
.nominal_value
does change over time, one would have to alter the max
time series as well accordingly.I'll address that somewhat soon in a dedicated pull request.
Closed as it is not further required / not working as needed.