A little problem in the Gambler‘s question in chapter4 about reinforcement-learning-an-introduction HOT 8 CLOSED

see line 26 of the python script.

from reinforcement-learning-an-introduction.

@ShangtongZhang
Thank you for your attention.
However, when set state[goal]=1,why don't need add the reward part as 1 when achieve the goal.

where is the "r(s,a,s')" part? shouldn't it also be set to 1 when state[goal] =1.
Still not very clear about that. Am I understand something wrong?

from reinforcement-learning-an-introduction.

1 + gamma * 0 is equal to 0 + gamma * 1 if gamma == 1

from reinforcement-learning-an-introduction.

@ShangtongZhang
Thank you for your answer.
but 1 + gamma * 1 won't happen? r(s,a,s')=1,gamma=1,and v(state=goal)=1?
why is it that when you say r(s,a,s')=1, you defaultly set v(state=goal)=0 or when you say v(state=goal) == 0 , you defaultly set r(s,a,s')=0?
why these two part r(s,a,s')=1 and v(state=goal)=1 cannot exist at the same time when the goal is achieved?
Hopefully i make myself more clear.

from reinforcement-learning-an-introduction.

You may find exercise 4.8 helpful

from reinforcement-learning-an-introduction.

@ShangtongZhang
Have thought about exercise 4.8. Honestly, can't figure it out.
Could you please send me a copy of solutions about exercises?
:-)
Thank you!

from reinforcement-learning-an-introduction.

I guess the pdf solution doesn't exist. You should think harder by yourself.

from reinforcement-learning-an-introduction.

@ShangtongZhang
OK. I think I've got it. Thank you for your patience and lucid code.

from reinforcement-learning-an-introduction.