Graph Sparsification for Derandomizing Massively Parallel Computation with Low Space

  title={Graph Sparsification for Derandomizing Massively Parallel Computation with Low Space},
  author={Artur Czumaj and Peter Davies and Merav Parter},
  journal={Proceedings of the 32nd ACM Symposium on Parallelism in Algorithms and Architectures},
Massively Parallel Computation (MPC) is an emerging model which distills core aspects of distributed and parallel computation. It was developed as a tool to solve (typically graph) problems in systems where input is distributed over many machines with limited space. Recent work has focused on the regime in which machines have sublinear (in n, number of nodes in the input graph) space, with randomized algorithms presented for the fundamental problems of Maximal Matching and Maximal Independent… 

Deterministic massively parallel connectivity

A deterministic MPC low local space algorithm that in O(logD + loglogn) rounds determines connected components of the input graph is presented, which matches the complexity of state of the art randomized algorithms for this task.

Optimal Deterministic Massively Parallel Connectivity on Forests

This work gives an algorithm that identifies connected components in O(logD) deterministic rounds and yields a deterministic forest-rooting algorithm with the same runtime and memory bounds as the above algorithm, which can be verified by checking the O(1)-radius neighborhood of each node.

Massively Parallel Computation in a Heterogeneous Regime

It is shown that even a single large machine suffices to circumvent most of the conditional hardness results for the sublinear regime, and it is observed that the best known near-linear MPC algorithms for several other graph problems can be transformed to work in the heterogeneous MPC model with a single near- linear machine.

Improved Deterministic Connectivity in Massively Parallel Computation

A deterministic connectivity algorithm that matches all the parameters of the randomized algorithm and reduces the local computation time to nearly linear is presented, which is the first to have efficient local computation in low-memory MPC.

Exponential Speedup Over Locality in MPC with Optimal Memory

This work provides a method that, given the complexity of an LCL problem in the LOCAL model, automatically provides an exponentially faster algorithm for the low-space MPC setting that uses optimal global memory, that is, truly linear.

Massively Parallel Correlation Clustering in Bounded Arboricity Graphs

This paper studies the problem of correlation clustering in bounded arboricity graphs with respect to the Massively Parallel Computation (MPC) model, and develops a 3-approximation algorithm that runs in O (log λ · log logn) MPC rounds in the sublinear memory regime.

Improved Deterministic (Δ+1) Coloring in Low-Space MPC

The Chang-Li-Pettie algorithm runs in T_local =poly(loglog n) rounds, which sets the state-of-the-art randomized round complexity for the problem in the local model, and employs a combination of tools, notably pseudorandom generators (PRG) and bounded-independence hash functions.

Brief Announcement: A Randomness-efficient Massively Parallel Algorithm for Connectivity

The Connectivity algorithm is an instantiation of a general method for converting randomized algorithms in the PRAM model to highly randomness-efficient MPC algorithms and achieves a super-polynomial saving in randomness complexity.

Sample-and-Gather: Fast Ruling Set Algorithms in the Low-Memory MPC Model

It is shown that a $\beta$-ruling set can be computed in the low-memory MPC model with $O(n^\eps)$ memory-per-machine in $\tilde{O}(\beta \cdot \log^{1/(2^{\beta+1}-2)} \Delta)$ rounds, whp.

Deterministic graph coloring in the streaming model

It is proved that there is no deterministic single-pass semi-streaming algorithm that given a graph G with maximum degree Δ, can output a proper coloring of G using any number of colors which is sub-exponential in Δ.



Massively Parallel Computation of Matching and MIS in Sparse Graphs

This work considers maximal matching and maximal independent set problems in the MPC model, and a degree reduction technique is used that reduces these problems in graphs with arboricity λ to the corresponding problems inGraphs with maximum degree poly(λ, log n) in O(log2log n) rounds, giving rise to O(√ log λ ⋅ loglog λ + log 2 log n)-round algorithms.

Near-Optimal Massively Parallel Graph Connectivity

This paper presents an algorithm that for graphs with diameter D in the wide range [log^ε n, n], takes O(log D) rounds to identify the connected components and takes O (log log n) rounds for all other graphs and uses an optimal total space of O(m).

Massively Parallel Algorithms for Finding Well-Connected Components in Sparse Graphs

An MPC algorithm that finds all connected components with spectral gap at least λ in a graph in O(log log n + log(1/λ)) MPC rounds and nδ memory per machine for any constant δ ∈ (0,1).

Sparsifying Distributed Algorithms with Ramifications in Massively Parallel Computation and Centralized Local Computation

We introduce a method for sparsifying distributed algorithms and exhibit how it leads to improvements that go past known barriers in two algorithmic settings of large-scale graph processing:

Round Compression for Parallel Graph Algorithms in Strongly Sublinear Space

The first algorithm is given that breaks this logarithmic barrier and runs in $\tilde O(\sqrt{\log n})$ rounds, as long as the total space is at least slightly superlinear in the number of vertices.

Parallel Graph Connectivity in Log Diameter Rounds

This work gives fully scalable, faster algorithms for the connectivity problem, by parameterizing the time complexity as a function of the diameter of the graph, and describes a general technique called double exponential speed problem size reduction.

Exponentially Faster Massively Parallel Maximal Matching

It is shown that this algorithm drastically reduces the vertex degrees, which leads to an O(log log ) round algorithm for maximal matching with O(n) space (or even mildly sublinear in n using standard techniques), and is the best possible approximation factor under standard assumptions.

Shuffles and Circuits (On Lower Bounds for Modern Parallel Computation)

An abstract model of massively parallel computation, where essentially the only restrictions are that the “fan-in” of each machine is limited to s bits, where s is smaller than the input size n, and that computation proceeds in synchronized rounds.

Shuffles and Circuits: (On Lower Bounds for Modern Parallel Computation)

An abstract model of massively parallel computation, where essentially the only restrictions are that the "fan-in" of each machine is limited to s bits, and that computation proceeds in synchronized rounds, which is proved to be the best one could hope for.

Round compression for parallel matching algorithms

The above O(logn) round complexity bound is broken even in the case of slightly sublinear memory per machine, and the improvement here is almost exponential: the best MPC round complexity matches what one can already get in the PRAM model, without the need to take advantage of the extra local computation power.