A Central Limit Theorem for biased random walks on Galton-Watson trees

Let ${\cal T}$ be a rooted Galton-Watson tree with offspring distribution $\{p_k\}$ that has $p_0=0$, mean $m=\sum kp_k>1$ and exponential tails. Consider the $\lambda$-biased random walk $\{X_n\}_{n\geq 0}$ on ${\cal T}$; this is the nearest neighbor random walk which, when at a vertex $v$ with $d_v$ offspring, moves closer to the root with probability $\lambda/(\lambda+d_v)$, and moves to each of the offspring with probability $1/(\lambda+d_v)$. It is known that this walk has an a.s. constant speed $\v=\lim_n |X_n|/n$ (where $|X_n|$ is the distance of $X_n$ from the root), with $\v>0$ for $0<\lambda<m$ and $\v=0$ for $\lambda \ge m$. For all $\lambda \le m$, we prove a quenched CLT for |X_n|-n\v. (For $\lambda>m$ the walk is positive recurrent, and there is no CLT.) The most interesting case by far is $\lambda=m$, where the CLT has the following form: for almost every ${\cal T}$, the ratio $|X_{[nt]}|/\sqrt{n}$ converges in law as $n \to \infty$ to a deterministic multiple of the absolute value of a Brownian motion. Our approach to this case is based on an explicit description of an invariant measure for the walk from the point of view of the particle (previously, such a measure was explicitly known only for $\lambda=1$) and the construction of appropriate harmonic coordinates.