#td-error

A Q-learning agent with $\gamma = 0.99$ initializes every $Q(s, a)$ optimistically to $1/(1 - \gamma) = 100$, the largest possible return when rewards lie in $[0, 1]$. It takes action $a$ in state $s$, receives reward 0 and lands in $s'$, whose Q-values are all still 100. With learning rate $\alpha = 0.1$, what is $Q(s, a)$ after the update?