Difference between the coreal_loss and reconons_loss in forecast_trainer.source_train in engine.py ?

[HQ-LV]

Thank you for the open source code!

What is the difference between the coreal_loss and reconons_loss in forecast_trainer.source_train in engine.py?
coral_loss (line 326) is the difference: (st_s_ - recons (t-s_));
recons loss (line 356) is also the difference: (st_s_recons - st_s_)=(recons (t-s_) - st_s_).
It seems that coral_loss calculated only on randomly nodes with index of s_idx , while recond_loss calculated on all nodes?