irlc.ex02.dp_model.DPModel

class irlc.ex02.dp_model.DPModel(N)

Code for the base DP model class. The lecture notes contains further discussino about the distinction of models/environments/agents which I recommend reading first. Briefly, the role of the DP model is to define (in details) the basic N-step decision problem corresponding to the equations:

x_{k+1} = f_k(x_k, u_k, w_k) cost = g_k(x_k, u_k, w_k). terminal cost = g_N(x_N)

The above corresponds to the first 3 methods below. The rest of the functions specify the available states and actions (see definition of the basic problem in the lecture notes).

__init__(N)

Methods

A(x, k)
Pw(x, u, k)	At step k, given x_k, u_k, compute the set of random noise disturbances w and their probabilities as a dict {..., w_i: pw_i, ...} such that
S(k)
__init__(N)
f(x, u, w, k)
g(x, u, w, k)
gN(x)
w_rnd(x, u, k)	generate random disturbances w ~ P_k(x, u) (useful for simulation)