...gradient.m).¹

https://igi-web.tugraz.at/lehre/MLB/WS10/task1.zip

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... cart-pole.²

https://igi-web.tugraz.at/lehre/MLB/WS10/task2.zip

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Learning³

https://igi-web.tugraz.at/lehre/intern/MDP_Theory.pdf

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... files⁴

https://igi-web.tugraz.at/lehre/intern/MountainCarDemo.zip

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... files⁵

https://igi-web.tugraz.at/lehre/intern/CartPoleDemo.zip

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Toolbox⁶

https://igi-web.tugraz.at/lehre/MLB/WS10/intern/MountainCarDemo.zip

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...swimmer.zip⁷

https://igi-web.tugraz.at/lehre/MLB/WS10/intern/swimmer.zip

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ⁸

This function points to the file evaluate.m

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...\space ⁹

Note that in the lecture slides the gradient $\log \pi(\dot{\mathbf{y}}_t\vert \mathbf{z}_t, x_t; \mathbf{b})$ is given by $\propto \epsilon \Phi(x)^T$ , which is typically a matrix not a vector. The gradient vector can be obtained by transforming the matrix to a vector $[\epsilon_t(1) \Phi(x_t); \epsilon_t(2) \Phi(x_t)]$

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...cannon.zip¹⁰

https://igi-web.tugraz.at/lehre/MLB/WS10/intern/cannon.zip

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...Gibbs.zip¹¹

https://igi-web.tugraz.at/lehre/MLB/WS10/intern/Gibbs.zip

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... endeffector-position¹²

The endeffector position is given by $[x,y]^T = [l_1 sin(q_1) + l_2 sin(q_1 + q_2), l_1 cos(q_1) + l_2 cos(q_1 + q_2)]$ , where $q_i$ denotes the joint angle of the $i$ th joint.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.