Abstract
This paper addresses the cornerstone family of local problems in distributed computing, and investigates the curious gap between randomized and deterministic solutions under bandwidth restrictions. Our main contribution is in providing tools for derandomizing solutions to local problems, when the n nodes can only send \(O(\log n)\)-bit messages in each round of communication. Our framework mostly follows by the derandomization approach of Luby (J Comput Syst Sci 47(2):250–286, 1993) combined with the power of all to all communication. Our key results are as follows: first, we show that in the congested clique model, which allows all-to-all communication, there is a deterministic maximal independent set algorithm that runs in \(O(\log ^2 {\varDelta })\) rounds, where \({\varDelta }\) is the maximum degree. When \({\varDelta }=O(n^{1/3})\), the bound improves to \(O(\log {\varDelta })\). In addition, we deterministically construct a \((2k-1)\)-spanner with \(O(kn^{1+1/k}\log n)\) edges in \(O(k \log n)\) rounds in the congested clique model.
Similar content being viewed by others
Notes
As standard, with high probability means with probability that is at least \(1-1/n^c\) for a constant c.
Flipping a biased coin with probability \(1/2^i\), is the same as getting a uniformly distributed number y in [1, b] and outputting 1 if and only if \(y \in [1, 2^{b-i}]\).
The term \(M_{t,b}(u,W(v))\) is important as it is what prevents double counting, because the corresponding random variables defined by the neighbors of v are mutually exclusive.
The simplified version that we describe consists of only one phase that includes both phases mentioned in [8]. This is because the number of clusters in the last iteration k is forced to be zero and hence all vertices are unclustered in the last iteration and so they connect to all clusters of the previous iteration.
References
Alon, N., Babai, L., Itai, A.: A fast and simple randomized parallel algorithm for the maximal independent set problem. J. Algorithms 7(4), 567–583 (1986)
Awerbuch, B., Goldberg, A.V., Luby, M., Plotkin, S.A.: Network decomposition and locality in distributed computation. In: FOCS, pp. 364–369 (1989)
Barenboim, L.: Deterministic (\(\Delta + 1\))-coloring in sublinear (in \(\Delta \)) time in static, dynamic and faulty networks. In: PODC, pp. 345–354 (2015)
Barenboim, L., Elkin, M.: Distributed graph coloring: fundamentals and recent developments. Synth. Lect. Distrib. Comput. Theory 4(1), 1–171 (2013)
Barenboim, L., Elkin, M., Gavoille, C.: A fast network-decomposition algorithm and its applications to constant-time distributed computation. In: SIROCCO, pp. 209–223 (2015)
Barenboim, L., Elkin, M., Kuhn, F.: Distributed (\(\Delta + 1\))-coloring in linear (in \(\Delta \)) time. SIAM J. Comput. 43(1), 72–95 (2014)
Barenboim, L., Elkin, M., Pettie, S., Schneider, J.: The locality of distributed symmetry breaking. In: FOCS, pp. 321–330 (2012)
Baswana, S., Sen, S.: A simple and linear time randomized algorithm for computing sparse spanners in weighted graphs. Random Struct. Algorithms 30(4), 532–563 (2007)
Benjamini, I., Gurel-Gurevich, O., Peled, R.: On k-wise independent distributions and Boolean functions. arXiv preprint arXiv:1201.3261 (2012)
Berger, B., Rompel, J.: Simulating (log CN)-wise independence in NC. J. ACM (JACM) 38(4), 1026–1046 (1991)
Brandt, S., Fischer, O., Hirvonen, J., Keller, B., Lempiäinen, T., Rybicki, J., Suomela, J., Uitto, J.: A lower bound for the distributed Lovász local lemma. In: STOC, pp. 479–488 (2016)
Censor-Hillel, K., Kaski, P., Korhonen, J.H., Lenzen, C., Paz, A., Suomela, J.: Algebraic methods in the congested clique. In: PODC, pp. 143–152 (2015)
Chandrasekaran, K., Goyal, N., Haeupler, B.: Deterministic algorithms for the Lovász local lemma. SIAM J. Comput. 42(6), 2132–2155 (2013)
Chang, Y.-J., Fischer, M., Ghaffari, M., Uitto, J., Zheng, Y.: The complexity of (\(\Delta +1\)) coloring in congested clique, massively parallel computation, and centralized local computation. In: Proceedings of the 2019 ACM Symposium on Principles of Distributed Computing, PODC 2019, Toronto, ON, Canada, July 29–August 2, 2019, pp. 471–480 (2019)
Chang, Y.-J., Kopelowitz, T., Pettie, S.: An exponential separation between randomized and deterministic complexity in the LOCAL model. In: FOCS, pp. 615–624 (2016)
Derbel, B., Gavoille, C.: Fast deterministic distributed algorithms for sparse spanners. In: SIROCCO, pp. 100–114 (2006)
Derbel, B. Gavoille, C., Peleg, D.: Deterministic distributed construction of linear stretch spanners in polylogarithmic time. In: DISC, pp. 179–192 (2007)
Derbel, B., Gavoille, C., Peleg, D., Viennot, L.: On the locality of distributed sparse spanner construction. In: PODC, pp. 273–282 (2008)
Derbel, B., Gavoille, C., Peleg, D., Viennot, L.: Local computation of nearly additive spanners. In: DISC, pp. 176–190 (2009)
Derbel, B., Mosbah, M., Zemmari, A.: Sublinear fully distributed partition with applications. Theory Comput. Syst. 47(2), 368–404 (2010)
Dolev, D., Lenzen, C., Peled, S.: “tri, tri again”: finding triangles and small subgraphs in a distributed setting—(extended abstract). In: DISC, pp. 195–209 (2012)
Drucker, A., Kuhn, F., Oshman, R.: On the power of the congested clique model. In: PODC, pp. 367–376 (2014)
Elkin, M.: A near-optimal distributed fully dynamic algorithm for maintaining sparse spanners. In: PODC, pp. 185–194 (2007)
Feuilloley, L., Fraigniaud, P.: Randomized local network computing. In: SPAA, pp. 340–349 (2015)
Le Gall, F.: Further algebraic algorithms in the congested clique model and applications to graph-theoretic problems. In: DISC, pp. 57–70 (2016)
Ghaffari, M.: An improved distributed algorithm for maximal independent set. In: SODA, pp. 270–277 (2016)
Ghaffari, M.: Distributed MIS via all-to-all communication. In: Proceedings of the ACM Symposium on Principles of Distributed Computing, PODC 2017, Washington, DC, USA, July 25–27, 2017, pp. 141–149 (2017)
Ghaffari, M., Parter, M.: MST in log-star rounds of congested clique. In: PODC, pp. 19–28 (2016)
Goldberg, M., Spencer, T.: A new parallel algorithm for the maximal independent set problem. SIAM J. Comput. 18(2), 419–427 (1989)
Han, Y.: A fast derandomization scheme and its applications. SIAM J. Comput. 25(1), 52–82 (1996)
Hegeman, J.W., Pandurangan, G., Pemmaraju, S.V., Sardeshmukh, V.B., Scquizzato, M.: Toward optimal bounds in the congested clique: graph connectivity and MST. In: PODC, pp. 91–100 (2015)
Hegeman, J.W., Pemmaraju, S.V.: Lessons from the congested clique applied to mapreduce. In: SIROCCO, pp. 149–164 (2014)
Hegeman, J.W., Pemmaraju, S.V., Sardeshmukh, V.: Near-constant-time distributed algorithms on a congested clique. In: DISC, pp. 514–530 (2014)
Henzinger, M., Krinninger, S., Nanongkai, D.: A deterministic almost-tight distributed algorithm for approximating single-source shortest paths. In: STOC, pp. 489–498 (2016)
Holzer, S., Pinsker, N.: Approximation of distances and shortest paths in the broadcast congest clique. In: OPODIS, pp. 6:1–6:16 (2015)
Israeli, A., Itai, A.: A fast and simple randomized parallel algorithm for maximal matching. Inf. Process. Lett. 22(2), 77–80 (1986)
Jurdzinski, T., Nowicki, K.: MST in O(1) rounds of congested clique. In: SODA, pp. 2620–2632. SIAM (2018)
Karp, R.M., Wigderson, A.: A fast parallel algorithm for the maximal independent set problem. In: STOC, pp. 266–272 (1984)
Kuhn, F., Moscibroda, T., Wattenhofer, R.: Local computation: lower and upper bounds. J. ACM 63(2), 17 (2016)
Lenzen, C.: Optimal deterministic routing and sorting on the congested clique. In: PODC, pp. 42–50 (2013)
Lenzen, C., Wattenhofer, R.: Tight bounds for parallel randomized load balancing: extended abstract. In: STOC, pp. 11–20 (2011)
Linial, N.: Locality in distributed graph algorithms. SIAM J. Comput. 21(1), 193–201 (1992)
Lotker, Z., Pavlov, E., Patt-Shamir, B., Peleg, D.: MST construction in O(\(\log log n\)) communication rounds. In: SPAA, pp. 94–100 (2003)
Luby, M.: A simple parallel algorithm for the maximal independent set problem. SIAM J. Comput. 15(4), 1036–1053 (1986)
Luby, M.: Removing randomness in parallel computation without a processor penalty. J. Comput. Syst. Sci. 47(2), 250–286 (1993)
Motwani, R., Naor, J., Naor, M.: The probabilistic method yields deterministic parallel algorithms. J. Comput. Syst. Sci. 49(3), 478–516 (1994)
Nanongkai, D.: Distributed approximation algorithms for weighted shortest paths. In: STOC, pp. 565–573 (2014)
Naor, M., Stockmeyer, L.J.: What can be computed locally? SIAM J. Comput. 24(6), 1259–1277 (1995)
Panconesi, A., Srinivasan, A.: Improved distributed algorithms for coloring and network decomposition problems. In: STOC, pp. 581–592 (1992)
Pantziou, G., Spirakis, P., Zaroliagis, C.: Fast parallel approximations of the maximum weighted cut problem through derandomization. In: FSTTCS, pp. 20–29 (1989)
Parter, M.: (\(\Delta + 1\)) coloring in the congested clique model. In: 45th International Colloquium on Automata, Languages, and Programming, ICALP 2018, July 9–13, 2018, Prague, Czech Republic, pp. 160:1–160:14 (2018)
Parter, M., Su, H.-H.: Randomized (\(\Delta + 1\))-coloring in \(O(\log *\Delta )\) congested clique rounds. In: 32nd International Symposium on Distributed Computing, DISC 2018, New Orleans, LA, USA, October 15–19, 2018, pp. 39:1–39:18 (2018)
Patt-Shamir, B., Teplitsky, M.: The round complexity of distributed sorting: extended abstract. In: PODC, pp. 249–256 (2011)
Pemmaraju, S.V.: Equitable coloring extends Chernoff-hoeffding bounds. In: Approximation, Randomization and Combinatorial Optimization: Algorithms and Techniques, 4th International Workshop on Approximation Algorithms for Combinatorial Optimization Problems, APPROX 2001 and 5th International Workshop on Randomization and Approximation Techniques in Computer Science, RANDOM 2001 Berkeley, CA, USA, August 18–20, 2001, Proceedings, pp. 285–296 (2001). https://doi.org/10.1007/3-540-44666-4_31
Pettie, S.: Distributed algorithms for ultrasparse spanners and linear size skeletons. Distrib. Comput. 22(3), 147–166 (2010)
Roditty, L., Thorup, M., Zwick, U. Deterministic constructions of approximate distance oracles and spanners. In: ICALP, pp. 261–272 (2005)
Schmidt, J.P., Siegel, A., Srinivasan, A.: Chernoff-hoeffding bounds for applications with limited independence. SIAM J. Discrete Math. 8(2), 223–250 (1995)
Srivastav, A., Kliemann, L.: Parallel algorithms via the probabilistic method. In: Rajasekaran, S. (ed.) Handbook of Parallel Computing: Models. Algorithms and Applications. Chapman and Hall/CRC, Boca Raton (2007)
Suomela, J.: Survey of local algorithms. ACM Comput. Surv. 45(2), 24 (2013)
Vadhan, S.P.: Pseudorandomness. Found. Trends Theor. Comput. Sci. 7(1–3), 1–336 (2012)
Acknowledgements
We are very grateful to Mohsen Ghaffari for many helpful discussions and useful observations involving the derandomization of his MIS algorithm.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Keren Censor-Hillel: Supported in part by the Israel Science Foundation (Grant 1696/14). This project has received funding from the European Union’s Horizon 2020 Research And Innovation Program under grant Agreement No. 755839. Gregory Schwartzman: This work was supported by JSPS KAKENHI Grant Numbers 19K20216 and JP18H05291.
Pseudocode of the deterministic MIS algorithm
Pseudocode of the deterministic MIS algorithm
Let \({\mathcal {H}}={\mathcal {H}}_{\gamma ,\beta }\) with \(\gamma ={\varTheta }(\log n)\) and \(\beta ={\varTheta }(\log {\varDelta })\) be given by Lemma 1. Let \({\mathcal {H}}(Y_i) \subseteq {\mathcal {H}}_{\gamma ,\beta }\) be the collection of all hash functions that agree with the partial seed \(Y_i\). Each function \(h \in {\mathcal {H}}(Y_i)\) corresponds to a deterministic MIS algorithm.
For a hash function \(h \in {\mathcal {H}}\) and value \(p=1/2^i\) representing the probability of a node to be marked, define \(m_h(v,p)=1\) if \(h(ID(v)) \in [1,2^{\beta -i}]\) and \(m_h(v,p)=0\) otherwise. That is, marking a node v with probability p is simulated deterministically by computing \(m_h(v,p)\), since we output 1 only if the value h(ID(v)) appears in the top \(1/2^i\) fraction of the range \([1, \beta ]\). For \(p=1/2^i\) and \(p'=1/2^{i'}\), define \(m_h(v,u, p,p')=1\) if \(h(ID(v)) \in [1,2^{\beta -i}]\) and \(h(ID(u)) \in [1, 2^{\beta -i'}]\) and \(m_h(v,u, p,p')=0\) otherwise. That is \(m_h(v,u, p,p')=1\) is the deterministic simulation of having both v and u being marked when v, u are marked with probability \(p,p'\). Let \(\alpha \in (0,1]\) be the constant such that every golden node is removed with probability at least \(\alpha \) when using pairwise independence. Algorithm 5 gives the pseudocode of our deterministic MIS algorithm.
Rights and permissions
About this article
Cite this article
Censor-Hillel, K., Parter, M. & Schwartzman, G. Derandomizing local distributed algorithms under bandwidth restrictions. Distrib. Comput. 33, 349–366 (2020). https://doi.org/10.1007/s00446-020-00376-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00446-020-00376-1