Batched Setting

Berenbrink, Czumaj, Englert, Friedetzky and Nagel (2012) introduced \(\Batched \) setting where balls are allocated in batches of \(b\) balls. More formally, for any balanced allocation process with probability allocation vector \(p\), the setting is defined as follows.

The \(\Batched \) setting:
Parameter: The probability allocation vector \(p^t\) for the process.
Iteration: At each step \(t = 0 \cdot b, 1 \cdot b, 2 \cdot b\ldots \),

  • • Sample \(b\) bins \(i_1, \ldots , i_b\) from \([n]\) according to \(p^t\).

  • • For each \(i \in [n]\), update

    \[ x_i^{t+b} = x_i^t + \sum _{j = 1}^b \mathbf {1}_{i_j = i}. \]

For the \(\TwoChoice \) process, it was shown that this process achieves an \(\Oh (\frac {\log n}{\log \log n})\) for when \(b = n\) and an \(\Oh (\frac {b}{n})\) gap for \(b \geq n \log n\).

Your browser does not support the HTML5 canvas tag.

.

References

  • [1] Petra Berenbrink et al. “Multiple-Choice Balanced Allocation in (Almost) Parallel”. In: Proceedings of 16th International Workshop on Approximation, Randomization, and Combinatorial Optimization (RANDOM’12). Berlin Heidelberg: Springer-Verlag, 2012, pp. 411–422. doi: 10.1007/978-3-642-32512-0_35.