Discussions/Bugs about the gradient ascent trajectory function

charlesrocabert / gbapy

Growth balance analysis in Python

GNU General Public License v3.0

1 stars 0 forks source link

Discussions/Bugs about the gradient ascent trajectory function #1

Open charlesrocabert opened 3 months ago

charlesrocabert commented 3 months ago

The method model.set_f() is essential to set the $f$ vector before calculating the model state: It should come just after you calculate the new next_f!

next_f = next_f + model.GCC_f[1:] *dt
model.set_f(next_f)
model.calculate()

fumer100 commented 3 months ago

Do I have to setf everytime? I am just setting it when my model is consistent with the next_f. Ok I got it now thank you !

charlesrocabert commented 3 months ago

I think there are two problems with your $dt$ update:

The initial $dt$ is way too large ($dt=20$). You should start with a small value already, e.g. $dt=0.01$,
To reduce the $dt$ value, multiply it by a factor, do not substract a value. For example, $dt = dt*0.5$