The g-Bounded Process

Alistarh, Brown, Kopinsky, Li and Nadiradze (2018) introduced the \(\GBounded \) process which is an adversarial version of \(\TwoChoice \), where whenever the load difference of the two bins is at most \(g\), it allocates to the heavier bin. More formally, it can be described as follows.

The \(\GBounded \) process:
Parameter: A threshold \(g \in \N \).
Iteration: At each step \(t = 1, 2, \ldots \),

  • • Sample two bins \(i_1\) and \(i_2\) uniformly at random.

  • • If \(|y_{i_1}^t - y_{i_2}^t| \leq g\), then allocate to \(j \in \{i_1, i_2\}\) satisfying \(y_j^t = \max \{ y_{i_1}^t, y_{i_2}^t \}\).

  • • Otherwise, allocate to \(j \in \{i_1, i_2\}\) satisfying \(y_j^t = \min \{ y_{i_1}^t, y_{i_2}^t \}\).

In [2] it was shown that the \(\GBounded \) process achieves a \(\Theta (\frac {g}{\log g} \cdot \log \log n)\) gap.

Your browser does not support the HTML5 canvas tag.

.

References

  • [1] Dan Alistarh et al. “Distributionally Linearizable Data Structures”. In: Proceedings of 30th on Symposium on Parallelism in Algorithms and Architectures (SPAA’18). New York, NY, USA: Association for Computing Machinery, 2018, pp. 133–142. doi: 10.1145/3210377.3210411.

  • [2] Dimitrios Los and Thomas Sauerwald. “Balanced Allocations with the Choice of Noise”. In: 41st Annual ACM-SIGOPT Principles of Distributed Computing (PODC’22). Salerno, Italy: ACM, 2022, 164–175. isbn: 9781450392624. doi: 10.1145/3519270.3538428.