Document

**Published in:** LIPIcs, Volume 254, 40th International Symposium on Theoretical Aspects of Computer Science (STACS 2023)

We study the repeated balls-into-bins process introduced by Becchetti, Clementi, Natale, Pasquale and Posta (2019). This process starts with m balls arbitrarily distributed across n bins. At each round t = 1,2,…, one ball is selected from each non-empty bin, and then placed it into a bin chosen independently and uniformly at random. We prove the following results:
- For any n ⩽ m ⩽ poly(n), we prove a lower bound of Ω(m/n ⋅ log n) on the maximum load. For the special case m = n, this matches the upper bound of 𝒪(log n), as shown in [Luca Becchetti et al., 2019]. It also provides a positive answer to the conjecture in [Luca Becchetti et al., 2019] that for m = n the maximum load is ω(log n/ log log n) at least once in a polynomially large time interval. For m ∈ [ω(n), n log n], our new lower bound disproves the conjecture in [Luca Becchetti et al., 2019] that the maximum load remains 𝒪(log n).
- For any n ⩽ m ⩽ poly(n), we prove an upper bound of 𝒪(m/n ⋅ log n) on the maximum load for all steps of a polynomially large time interval. This matches our lower bound up to multiplicative constants.
- For any m ⩾ n, our analysis also implies an 𝒪(m²/n) waiting time to reach a configuration with a 𝒪(m/n ⋅ log m) maximum load, even for worst-case initial distributions.
- For m ⩾ n, we show that every ball visits every bin in 𝒪(m log m) rounds. For m = n, this improves the previous upper bound of 𝒪(n log² n) in [Luca Becchetti et al., 2019]. We also prove that the upper bound is tight up to multiplicative constants for any n ⩽ m ⩽ poly(n).

Dimitrios Los and Thomas Sauerwald. Tight Bounds for Repeated Balls-Into-Bins. In 40th International Symposium on Theoretical Aspects of Computer Science (STACS 2023). Leibniz International Proceedings in Informatics (LIPIcs), Volume 254, pp. 45:1-45:22, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2023)

Copy BibTex To Clipboard

@InProceedings{los_et_al:LIPIcs.STACS.2023.45, author = {Los, Dimitrios and Sauerwald, Thomas}, title = {{Tight Bounds for Repeated Balls-Into-Bins}}, booktitle = {40th International Symposium on Theoretical Aspects of Computer Science (STACS 2023)}, pages = {45:1--45:22}, series = {Leibniz International Proceedings in Informatics (LIPIcs)}, ISBN = {978-3-95977-266-2}, ISSN = {1868-8969}, year = {2023}, volume = {254}, editor = {Berenbrink, Petra and Bouyer, Patricia and Dawar, Anuj and Kant\'{e}, Mamadou Moustapha}, publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik}, address = {Dagstuhl, Germany}, URL = {https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.STACS.2023.45}, URN = {urn:nbn:de:0030-drops-176975}, doi = {10.4230/LIPIcs.STACS.2023.45}, annote = {Keywords: Repeated balls-into-bins, self-stabilizing systems, balanced allocations, potential functions, random walks} }

Document

**Published in:** LIPIcs, Volume 215, 13th Innovations in Theoretical Computer Science Conference (ITCS 2022)

We consider the allocation of m balls into n bins with incomplete information. In the classical Two-Choice process a ball first queries the load of two randomly chosen bins and is then placed in the least loaded bin. In our setting, each ball also samples two random bins but can only estimate a bin’s load by sending binary queries of the form "Is the load at least the median?" or "Is the load at least 100?".
For the lightly loaded case m = 𝒪(n), Feldheim and Gurel-Gurevich (2021) showed that with one query it is possible to achieve a maximum load of 𝒪(√{log n/log log n}), and they also pose the question whether a maximum load of m/n+𝒪(√{log n/log log n}) is possible for any m = Ω(n). In this work, we resolve this open problem by proving a lower bound of m/n+Ω(√{log n}) for a fixed m = Θ(n √{log n}), and a lower bound of m/n+Ω(log n/log log n) for some m depending on the used strategy.
We complement this negative result by proving a positive result for multiple queries. In particular, we show that with only two binary queries per chosen bin, there is an oblivious strategy which ensures a maximum load of m/n+𝒪(√{log n}) for any m ≥ 1. Further, for any number of k = 𝒪(log log n) binary queries, the upper bound on the maximum load improves to m/n + 𝒪(k(log n)^{1/k}) for any m ≥ 1.
This result for k queries has several interesting consequences: (i) it implies new bounds for the (1+β)-process introduced by Peres, Talwar and Wieder (2015), (ii) it leads to new bounds for the graphical balanced allocation process on dense expander graphs, and (iii) it recovers and generalizes the bound of m/n+𝒪(log log n) on the maximum load achieved by the Two-Choice process, including the heavily loaded case m = Ω(n) which was derived in previous works by Berenbrink et al. (2006) as well as Talwar and Wieder (2014).
One novel aspect of our proofs is the use of multiple super-exponential potential functions, which might be of use in future work.

Dimitrios Los and Thomas Sauerwald. Balanced Allocations with Incomplete Information: The Power of Two Queries. In 13th Innovations in Theoretical Computer Science Conference (ITCS 2022). Leibniz International Proceedings in Informatics (LIPIcs), Volume 215, pp. 103:1-103:23, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2022)

Copy BibTex To Clipboard

@InProceedings{los_et_al:LIPIcs.ITCS.2022.103, author = {Los, Dimitrios and Sauerwald, Thomas}, title = {{Balanced Allocations with Incomplete Information: The Power of Two Queries}}, booktitle = {13th Innovations in Theoretical Computer Science Conference (ITCS 2022)}, pages = {103:1--103:23}, series = {Leibniz International Proceedings in Informatics (LIPIcs)}, ISBN = {978-3-95977-217-4}, ISSN = {1868-8969}, year = {2022}, volume = {215}, editor = {Braverman, Mark}, publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik}, address = {Dagstuhl, Germany}, URL = {https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.ITCS.2022.103}, URN = {urn:nbn:de:0030-drops-156994}, doi = {10.4230/LIPIcs.ITCS.2022.103}, annote = {Keywords: power-of-two-choices, balanced allocations, potential functions, thinning} }