Open viatkin opened 2 years ago
Apparently bad parameters cause increasingly strong oscillations of the microbial pool, resulting in negative and impossibly large values when the step size is more than 1 day. 2 solutions to explore: 1) Improve model parametrization using original equations for fluxes from A. Walker. 2) Use differential equation solver with automatically adjusted step size.
Hi Aneesh!
I already added your model to the comparison, however, I noticed that it only works with iterator step (delta_t_val) of 1 day (which makes everything extremely slow). I tried changing it to several days, but every time (even with delta_t_val=2) the model explodes - it starts producing extremely high and extremely low values that eventually cause an overflow. I noticed that in your forward run function (param2res) delta_t_val=1, while in all other models delta_t_val=15 to speed up data assimilation. When I tried changing delta_t_val in param2res, it also caused output to explode. Do you know what might be the reason? We should fix it to make the model run fast.
Thanks!