# Deciding / Approximating Parity of Small Depth Decision Trees

Let C be a circuit such that:

• C: $\{0,1\}^n$ to $\{0,1\}$

• the top most gate is a parity gate

• all the inputs to the parity gate are small depth decision trees

• there is a total of $2^{ log^k n}$ decision trees.

• each decision tree has depth at most $log^k n$

Question 1:

Can we decide if $C$ is satisfiable faster than $O(2^n)$ ?

Question 2:

For a given epsilon, what is the fastest algorithm for approximating the probability $C$ accepts a uniformly sampled input?

Thanks!

-