# Can $\log^k n$ alternations be simulated in $\mathsf{NC}^k$?

Let $\mathsf{ATISP}(f(n), g(n))$ be the class of languages decided by alternating Turing machines that halt in time $f(n)$ using space $g(n)$. Let $\mathsf{AALTSP}(f(n), g(n))$ be the class of languages decided by alternating Turing machines that halt using $f(n)$ alternations and space $g(n)$.

Ruzzo proved that $\mathsf{NC}^k = \mathsf{ATISP}(\log^k n, \log n)$. He also showed that $\mathsf{NC}^k \subseteq \mathsf{AALTSP}(\log^k n, \log n) \subseteq \mathsf{NC}^{k + 1}$.

Is $\mathsf{NC}^k = \mathsf{AALTSP}(\log^k n, \log n)$?

-

Of course the equalities and inclusions claimed in the question hold only for uniform $\mathsf{NC}^k$. The class $\mathsf{AALTSP}(\log^k n, \log n)$ is the same as uniform $\mathsf{AC}^k$, so the question is the same as whether $\mathsf{NC}^k = \mathsf{AC}^k$.
In particular the case $k=1$ implies that $\mathsf{L}$ is equal to $\mathsf{NL}$ and even $\mathsf{LogCFL}$, since $\mathsf{NC}^1 \subseteq \mathsf{L} \subseteq \mathsf{NL} \subseteq \mathsf{LogCFL} \subseteq \mathsf{AC}^1$.