An official website of the United States government
We consider the maximum matching problem in the semi-streaming model formalized by Feigenbaum, Kannan, McGregor, Suri, and Zhang that is inspired by giant graphs of today. As our main result, we give a two-pass (1/2 + 1/16)-approximation algorithm for triangle-free graphs and a two-pass (1/2 + 1/32)-approximation algorithm for general graphs; these improve the approximation ratios of 1/2 + 1/52 for bipartite graphs and 1/2 + 1/140 for general graphs by Konrad, Magniez, and Mathieu. In three passes, we are able to achieve approximation ratios of 1/2 + 1/10 for triangle-free graphs and 1/2 + 1/19.753 for general graphs. We also give a multi-pass algorithm where we bound the number of passes precisely—we give a (2/3 − ε)- approximation algorithm that uses 2/(3ε) passes for triangle-free graphs and 4/(3ε) passes for general graphs. Our algorithms are simple and combinatorial, use O(n log n) space, and (can be implemented to) have O(1) update time per edge. For general graphs, our multi-pass algorithm improves the best known deterministic algorithms in terms of the number of passes: * Ahn and Guha give a (2/3−ε)-approximation algorithm that uses O(log(1/ε)/ε2) passes, whereas our (2/3 − ε)-approximation algorithm uses 4/(3ε) passes; * they also give a (1 − ε)-approximation algorithm that uses O(log n · poly(1/ε)) passes, where n is the number of vertices of the input graph; although our algorithm is (2/3−ε)-approximation, our number of passes do not depend on n. Earlier multi-pass algorithms either have a large constant inside big-O notation for the number of passes or the constant cannot be determined due to the involved analysis, so our multi-pass algorithm should use much fewer passes for approximation ratios bounded slightly below 2/3.
more »
« less
This content will become publicly available on May 10, 2025
Streaming Algorithms with Few State Changes
In this paper, we study streaming algorithms that minimize the number of changes made to their internal state (i.e., memory contents). While the design of streaming algorithms typically focuses on minimizing space and update time, these metrics fail to capture the asymmetric costs, inherent in modern hardware and database systems, of reading versus writing to memory. In fact, most streaming algorithms write to their memory on every update, which is undesirable when writing is significantly more expensive than reading. This raises the question of whether streaming algorithms with small space and number of memory writes are possible. We first demonstrate that, for the fundamental Fpmoment estimation problem with p ≥ 1, any streaming algorithm that achieves a constant factor approximation must make Ω(n1-1/p) internal state changes, regardless of how much space it uses. Perhaps surprisingly, we show that this lower bound can be matched by an algorithm which also has near-optimal space complexity. Specifically, we give a (1+ε)-approximation algorithm for Fpmoment estimation that use a near-optimal ~Oε(n1-1/p) number of state changes, while simultaneously achieving near-optimal space, i.e., for p∈[1,2), our algorithm uses poly(log n,1/ε) bits of space for, while for p>2, the algorithm uses ~Oε(n1-1/p) space. We similarly design streaming algorithms that are simultaneously near-optimal in both space complexity and the number of state changes for the heavy-hitters problem, sparse support recovery, and entropy estimation. Our results demonstrate that an optimal number of state changes can be achieved without sacrificing space complexity.
more »
« less
- Award ID(s):
- 2335411
- PAR ID:
- 10560522
- Publisher / Repository:
- Proc. ACM Manag. Data
- Date Published:
- Journal Name:
- Proceedings of the ACM on Management of Data
- Volume:
- 2
- Issue:
- 2
- ISSN:
- 2836-6573
- Page Range / eLocation ID:
- 1 to 28
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Given a family of sets (S1, S2,... SM) over a universe Ω, estimating the size of their union in the data streaming model is a fundamental computational problem with a wide variety of applications. The holy grail in the field of streaming is to seek design of algorithms that achieve (ε, δ)-approximation with poly(log |Ω|, ε-1, log δ-1) space and update time complexity. Earlier investigations achieve algorithms with desired space and update time complexity for restricted cases such as singletons (Distinct Elements problem), one-dimensional ranges, arithmetic progressions, and sub-cubes. However, techniques used in these works fail for many other simple structured sets. A prominent example is that of Klee's Measure Problem (KMP), wherein every set Si is represented by an axis-parallel rectangle in d-dimensional spaces. Despite extensive prior work, the best-known streaming algorithms for many of these cases depend on the size of the stream, and therefore the problem of whether there exists a streaming algorithm for estimations of size of the union of sets with poly(log |Ω|, ε-1, log δ-1) space and update time complexity has remained open. In this work, we focus on certain general families of sets called Delphic families (which allows efficient membership, sampling, and cardinality queries). Such families of sets capture several well-known problems, including KMP, test coverage, and hypervolume estimation. The primary contribution of our work is to resolve the above-mentioned open problem for streams over Delphic families. In particular, we design the first streaming algorithm for estimating |⋃i=1M Si| with poly(log |Ω|, ε-1, log δ-1) space and update time complexity (independent of M, the length of the stream) when each Si is a member from a Delphic family of sets. We further generalize our results to larger families of sets, called approximate-Delphic families, for which the size of a set can be known approximately but not exactly. Our results resolve two of the open problems listed in Meel, Vinodchandran, Chakraborty (PODS-21).more » « less
-
There has been a flurry of recent literature studying streaming algorithms for which the input stream is chosen adaptively by a black-box adversary who observes the output of the streaming algorithm at each time step. However, these algorithms fail when the adversary has access to the internal state of the algorithm, rather than just the output of the algorithm. We study streaming algorithms in the white-box adversarial model, where the stream is chosen adaptively by an adversary who observes the entire internal state of the algorithm at each time step. We show that nontrivial algorithms are still possible. We first give a randomized algorithm for the L1-heavy hitters problem that outperforms the optimal deterministic Misra-Gries algorithm on long streams. If the white-box adversary is computationally bounded, we use cryptographic techniques to reduce the memory of our L1-heavy hitters algorithm even further and to design a number of additional algorithms for graph, string, and linear algebra problems. The existence of such algorithms is surprising, as the streaming algorithm does not even have a secret key in this model, i.e., its state is entirely known to the adversary. One algorithm we design is for estimating the number of distinct elements in a stream with insertions and deletions achieving a multiplicative approximation and sublinear space; such an algorithm is impossible for deterministic algorithms. We also give a general technique that translates any two-player deterministic communication lower bound to a lower bound for randomized algorithms robust to a white-box adversary. In particular, our results show that for all p ≥ 0, there exists a constant Cp > 1 such that any Cp-approximation algorithm for Fp moment estimation in insertion-only streams with a white-box adversary requires Ω(n) space for a universe of size n. Similarly, there is a constant C > 1 such that any C-approximation algorithm in an insertion-only stream for matrix rank requires Ω(n) space with a white-box adversary. These results do not contradict our upper bounds since they assume the adversary has unbounded computational power. Our algorithmic results based on cryptography thus show a separation between computationally bounded and unbounded adversaries. Finally, we prove a lower bound of Ω(log n) bits for the fundamental problem of deterministic approximate counting in a stream of 0’s and 1’s, which holds even if we know how many total stream updates we have seen so far at each point in the stream. Such a lower bound for approximate counting with additional information was previously unknown, and in our context, it shows a separation between multiplayer deterministic maximum communication and the white-box space complexity of a streaming algorithmmore » « less
-
Barman, Siddharth; Lasota, Sławomir (Ed.)We consider the problem of estimating the size of a maximum matching in low-arboricity graphs in the dynamic graph stream model. In this setting, an algorithm with limited memory makes multiple passes over a stream of edge insertions and deletions, resulting in a low-arboricity graph. Let n be the number of vertices of the input graph, and α be its arboricity. We give the following results. 1) As our main result, we give a three-pass streaming algorithm that produces an (α + 2)(1 + ε)-approximation and uses space O(ε^{-2}⋅α²⋅n^{1/2}⋅log n). This result should be contrasted with the Ω(α^{-5/2}⋅n^{1/2}) space lower bound established by [Assadi et al., SODA'17] for one-pass algorithms, showing that, for graphs of constant arboricity, the one-pass space lower bound can be achieved in three passes (up to poly-logarithmic factors). Furthermore, we obtain a two-pass algorithm that uses space O(ε^{-2}⋅α²⋅n^{3/5}⋅log n). 2) We also give a (1+ε)-approximation multi-pass algorithm, where the space used is parameterized by an upper bound on the size of a largest matching. For example, using O(log log n) passes, the space required is O(ε^{-1}⋅α²⋅k⋅log n), where k denotes an upper bound on the size of a largest matching. Finally, we define a notion of arboricity in the context of matrices. This is a natural measure of the sparsity of a matrix that is more nuanced than simply bounding the total number of nonzero entries, but less restrictive than bounding the number of nonzero entries in each row and column. For such matrices, we exploit our results on estimating matching size to present upper bounds for the problem of rank estimation in the dynamic data stream model.more » « less
-
Chan, Timothy; Fischer, Johannes; Iacono, John; Herman, Grzegorz (Ed.)The maximum coverage problem is to select k sets, from a collection of m sets, such that the cardinality of their union, in a universe of size n, is maximized. We consider (1-1/e-ε)-approximation algorithms for this NP-hard problem in three standard data stream models. 1) Dynamic Model. The stream consists of a sequence of sets being inserted and deleted. Our multi-pass algorithm uses ε^{-2} k ⋅ polylog(n,m) space. The best previous result (Assadi and Khanna, SODA 2018) used (n +ε^{-4} k) polylog(n,m) space. While both algorithms use O(ε^{-1} log m) passes, our analysis shows that, when ε ≤ 1/log log m, it is possible to reduce the number of passes by a 1/log log m factor without incurring additional space. 2) Random Order Model. In this model, there are no deletions, and the sets forming the instance are uniformly randomly permuted to form the input stream. We show that a single pass and k polylog(n,m) space suffices for arbitrary small constant ε. The best previous result, by Warneke et al. (ESA 2023), used k² polylog(n,m) space. 3) Insert-Only Model. Lastly, our results, along with numerous previous results, use a sub-sampling technique introduced by McGregor and Vu (ICDT 2017) to sparsify the input instance. We explain how this technique and others used in the paper can be implemented such that the amortized update time of our algorithm is polylogarithmic. This also implies an improvement of the state-of-the-art insert only algorithms in terms of the update time: polylog(m,n) update time suffices, whereas the best previous result by Jaud et al. (SEA 2023) required update time that was linear in k.more » « less
