Open oyamad opened 6 years ago
The current code solves dynamic programs with
f(s, x)
g(s, x, e)
where
s
N
x
e
Discussion:
I think we should allow for discrete states
yes and yes
Is this different from discrete states?
The current code solves dynamic programs with
f(s, x)
andg(s, x, e)
where
s
is anN
-dimensional continuous state variablex
is a 1-dimensional continuous action variable ande
is a discrete random variable, whose distribution is state independent.Discussion: