improving dt_circular calculation with investment models
In some models dt_circular is unncessary if model does not use hours outside given data ranges.
However, it is currently calculated for tmp = smax(tt(t), ord(t)); dt_circular(t_full(t))${ ord(t) > tmp } ...
and automatically used in tt_agg_circular
. This is a small cost for models with 20k timesteps, but is annoyingly slow with models with 200+k timesteps. Nothing too severe, but still something to fix.
I tried to fix this once to 3.x, but did it in a way that broke Topi's building model. After few iterations, I returned the 2.x approach and the proper fix still waits.
Maybe dt_circular
would need another if condition based on t_end + t_horizon
? Or t
in active samples?