arXiv subjects

Peter Sanders

Publications and source records attributed to Peter Sanders.

At least 91 records · Page 5Linked to original sources

arXiv2018

Relational Equivalence Proofs Between Imperative and MapReduce Algorithms

MapReduce frameworks are widely used for the implementation of distributed algorithms. However, translating imperative algorithms into these frameworks requires significant structural changes to the algorithm. As the costs of running faulty algorithms at scale can be severe, it is highly desirable to verify the correctness of the translation, i.e., to prove that the MapReduce version is equivalent to the imperative original. We present a novel approach for proving equivalence between imperative and MapReduce algorithms based on partitioning the equivalence proof into a sequence of equivalence proofs between intermediate programs with smaller differences. Our approach is based on the insight that two kinds of sub-proofs are required: (1) uniform transformations changing the controlflow structure that are mostly independent of the particular context in which they are applied; and (2) context-dependent transformations that are not uniform but that preserve the overall structure and can be proved correct using coupling invariants. We demonstrate the feasibility of our approach by evaluating it on two prototypical algorithms commonly used as examples in MapReduce frameworks: k-means and PageRank. To carry out the proofs, we use the interactive theorem prover Coq with partial proof automation. The results show that our approach and its prototypical implementation based on Coq enables equivalence proofs of non-trivial algorithms and could be automated to a large degree.

cs.LO↗

arXiv2017

Fast OLAP Query Execution in Main Memory on Large Data in a Cluster

Main memory column-stores have proven to be efficient for processing analytical queries. Still, there has been much less work in the context of clusters. Using only a single machine poses several restrictions: Processing power and data volume are bounded to the number of cores and main memory fitting on one tightly coupled system. To enable the processing of larger data sets, switching to a cluster becomes necessary. In this work, we explore techniques for efficient execution of analytical SQL queries on large amounts of data in a parallel database cluster while making maximal use of the available hardware. This includes precompiled query plans for efficient CPU utilization, full parallelization on single nodes and across the cluster, and efficient inter-node communication. We implement all features in a prototype for running a subset of TPC-H benchmark queries. We evaluate our implementation using a 128 node cluster running TPC-H queries with 30 000 gigabyte of uncompressed data.

cs.DB↗

arXiv2017

In-place Parallel Super Scalar Samplesort (IPS$^4$o)

We present a sorting algorithm that works in-place, executes in parallel, is cache-efficient, avoids branch-mispredictions, and performs work O(n log n) for arbitrary inputs with high probability. The main algorithmic contributions are new ways to make distribution-based algorithms in-place: On the practical side, by using coarse-grained block-based permutations, and on the theoretical side, we show how to eliminate the recursion stack. Extensive experiments show that our algorithm IPS$^4$o scales well on a variety of multi-core machines. We outperform our closest in-place competitor by a factor of up to 3. Even as a sequential algorithm, we are up to 1.5 times faster than the closest sequential competitor, BlockQuicksort.

cs.DC↗

arXiv2017

Dynamic Space Efficient Hashing

We consider space efficient hash tables that can grow and shrink dynamically and are always highly space efficient, i.e., their space consumption is always close to the lower bound even while growing and when taking into account storage that is only needed temporarily. None of the traditionally used hash tables have this property. We show how known approaches like linear probing and bucket cuckoo hashing can be adapted to this scenario by subdividing them into many subtables or using virtual memory overcommitting. However, these rather straightforward solutions suffer from slow amortized insertion times due to frequent reallocation in small increments. Our main result is DySECT ({\bf Dy}namic {\bf S}pace {\bf E}fficient {\bf C}uckoo {\bf T}able) which avoids these problems. DySECT consists of many subtables which grow by doubling their size. The resulting inhomogeneity in subtable sizes is equalized by the flexibility available in bucket cuckoo hashing where each element can go to several buckets each of which containing several cells. Experiments indicate that DySECT works well with load factors up to 98\%. With up to 2.7 times better performance than the next best solution.

cs.DS↗

arXiv2017

Engineering DFS-Based Graph Algorithms

Depth-first search (DFS) is the basis for many efficient graph algorithms. We introduce general techniques for the efficient implementation of DFS-based graph algorithms and exemplify them on three algorithms for computing strongly connected components. The techniques lead to speed-ups by a factor of two to three compared to the implementations provided by LEDA and BOOST. We have obtained similar speed-ups for biconnected components algorithms. We also compare the graph data types of LEDA and BOOST.

cs.DS↗

arXiv2017

3D Cell Nuclei Segmentation with Balanced Graph Partitioning

Cell nuclei segmentation is one of the most important tasks in the analysis of biomedical images. With ever-growing sizes and amounts of three-dimensional images to be processed, there is a need for better and faster segmentation methods. Graph-based image segmentation has seen a rise in popularity in recent years, but is seen as very costly with regard to computational demand. We propose a new segmentation algorithm which overcomes these limitations. Our method uses recursive balanced graph partitioning to segment foreground components of a fast and efficient binarization. We construct a model for the cell nuclei to guide the partitioning process. Our algorithm is compared to other state-of-the-art segmentation algorithms in an experimental evaluation on two sets of realistically simulated inputs. Our method is faster, has similar or better quality and an acceptable memory overhead.

cs.CV↗

arXiv2017

Distributed Evolutionary k-way Node Separators

Computing high quality node separators in large graphs is necessary for a variety of applications, ranging from divide-and-conquer algorithms to VLSI design. In this work, we present a novel distributed evolutionary algorithm tackling the k-way node separator problem. A key component of our contribution includes new k-way local search algorithms based on maximum flows. We combine our local search with a multilevel approach to compute an initial population for our evolutionary algorithm, and further show how to modify the coarsening stage of our multilevel algorithm to create effective combine and mutation operations. Lastly, we combine these techniques with a scalable communication protocol, producing a system that is able to compute high quality solutions in a short amount of time. Our experiments against competing algorithms show that our advanced evolutionary algorithm computes the best result on 94% of the chosen benchmark instances.

cs.NE↗

arXiv2016

Engineering a Distributed Full-Text Index

We present a distributed full-text index for big data applications in a distributed environment. Our index can answer different types of pattern matching queries (existential, counting and enumeration). We perform experiments on inputs up to 100 GiB using up to 512 processors, and compare our index with the distributed suffix array by Arroyuelo et al. [Parall. Comput. 40(9): 471--495, 2014]. The result is that our index answers counting queries up to 5.5 times faster than the distributed suffix array, while using about the same space. We also provide a succinct variant of our index that uses only one third of the memory compared with our non-succinct variant, at the expense of only 20% slower query times.

cs.DS↗

arXiv2016

Concurrent Hash Tables: Fast and General?(!)

Concurrent hash tables are one of the most important concurrent data structures with numerous applications. Since hash table accesses can dominate the execution time of the overall application, we need implementations that achieve good speedup. Unfortunately, currently available concurrent hashing libraries turn out to be far away from this requirement in particular when contention on some elements occurs. Our starting point for better performing data structures is a fast and simple lock-free concurrent hash table based on linear probing that is limited to word-sized key-value types and does not support dynamic size adaptation. We explain how to lift these limitations in a provably scalable way and demonstrate that dynamic growing has a performance overhead comparable to the same generalization in sequential hash tables. We perform extensive experiments comparing the performance of our implementations with six of the most widely used concurrent hash tables. Ours are considerably faster than the best algorithms with similar restrictions and an order of magnitude faster than the best more general tables. In some extreme cases, the difference even approaches four orders of magnitude.

cs.DS↗

arXiv2016

Thrill: High-Performance Algorithmic Distributed Batch Data Processing with C++

We present the design and a first performance evaluation of Thrill -- a prototype of a general purpose big data processing framework with a convenient data-flow style programming interface. Thrill is somewhat similar to Apache Spark and Apache Flink with at least two main differences. First, Thrill is based on C++ which enables performance advantages due to direct native code compilation, a more cache-friendly memory layout, and explicit memory management. In particular, Thrill uses template meta-programming to compile chains of subsequent local operations into a single binary routine without intermediate buffering and with minimal indirections. Second, Thrill uses arrays rather than multisets as its primary data structure which enables additional operations like sorting, prefix sums, window scans, or combining corresponding fields of several arrays (zipping). We compare Thrill with Apache Spark and Apache Flink using five kernels from the HiBench suite. Thrill is consistently faster and often several times faster than the other frameworks. At the same time, the source codes have a similar level of simplicity and abstraction

cs.DC↗

arXiv2016

Fast Parallel Operations on Search Trees

Using (a,b)-trees as an example, we show how to perform a parallel split with logarithmic latency and parallel join, bulk updates, intersection, union (or merge), and (symmetric) set difference with logarithmic latency and with information theoretically optimal work. We present both asymptotically optimal solutions and simplified versions that perform well in practice - they are several times faster than previous implementations.

cs.DS↗

arXiv2016

Generating Semi-Synthetic Validation Benchmarks for Embryomics

Systematic validation is an essential part of algorithm development. The enormous dataset sizes and the complexity observed in many recent time-resolved 3D fluorescence microscopy imaging experiments, however, prohibit a comprehensive manual ground truth generation. Moreover, existing simulated benchmarks in this field are often too simple or too specialized to sufficiently validate the observed image analysis problems. We present a new semi-synthetic approach to generate realistic 3D+t benchmarks that combines challenging cellular movement dynamics of real embryos with simulated fluorescent nuclei and artificial image distortions including various parametrizable options like cell numbers, acquisition deficiencies or multiview simulations. We successfully applied the approach to simulate the development of a zebrafish embryo with thousands of cells over 14 hours of its early existence.

cs.CV↗

arXiv2016

Scalable Generation of Scale-free Graphs

We explain how massive instances of scale-free graphs following the Barabasi-Albert model can be generated very quickly in an embarrassingly parallel way. This makes this popular model available for studying big data graph problems. As a demonstration, we generated a Petaedge graph in less than an hour.

cs.DS↗

arXiv2016

Accelerating Local Search for the Maximum Independent Set Problem

Computing high-quality independent sets quickly is an important problem in combinatorial optimization. Several recent algorithms have shown that kernelization techniques can be used to find exact maximum independent sets in medium-sized sparse graphs, as well as high-quality independent sets in huge sparse graphs that are intractable for exact (exponential-time) algorithms. However, a major drawback of these algorithms is that they require significant preprocessing overhead, and therefore cannot be used to find a high-quality independent set quickly. In this paper, we show that performing simple kernelization techniques in an online fashion significantly boosts the performance of local search, and is much faster than pre-computing a kernel using advanced techniques. In addition, we show that cutting high-degree vertices can boost local search performance even further, especially on huge (sparse) complex networks. Our experiments show that we can drastically speed up the computation of large independent sets compared to other state-of-the-art algorithms, while also producing results that are very close to the best known solutions.

cs.DS↗

arXiv2015

k-way Hypergraph Partitioning via n-Level Recursive Bisection

We develop a multilevel algorithm for hypergraph partitioning that contracts the vertices one at a time. Using several caching and lazy-evaluation techniques during coarsening and refinement, we reduce the running time by up to two-orders of magnitude compared to a naive $n$-level algorithm that would be adequate for ordinary graph partitioning. The overall performance is even better than the widely used hMetis hypergraph partitioner that uses a classical multilevel algorithm with few levels. Aided by a portfolio-based approach to initial partitioning and adaptive budgeting of imbalance within recursive bipartitioning, we achieve very high quality. We assembled a large benchmark set with 310 hypergraphs stemming from application areas such VLSI, SAT solving, social networks, and scientific computing. We achieve significantly smaller cuts than hMetis and PaToH, while being faster than hMetis. Considerably larger improvements are observed for some instance classes like social networks, for bipartitioning, and for partitions with an allowed imbalance of 10%. The algorithm presented in this work forms the basis of our hypergraph partitioning framework KaHyPar (Karlsruhe Hypergraph Partitioning).

cs.DS↗

arXiv2015

Communication Efficient Algorithms for Top-k Selection Problems

We present scalable parallel algorithms with sublinear per-processor communication volume and low latency for several fundamental problems related to finding the most relevant elements in a set, for various notions of relevance: We begin with the classical selection problem with unsorted input. We present generalizations with locally sorted inputs, dynamic content (bulk-parallel priority queues), and multiple criteria. Then we move on to finding frequent objects and top-k sum aggregation. Since it is unavoidable that the output of these algorithms might be unevenly distributed over the processors, we also explain how to redistribute this data with minimal communication.

cs.DS↗

arXiv2015

Advanced Multilevel Node Separator Algorithms

A node separator of a graph is a subset S of the nodes such that removing S and its incident edges divides the graph into two disconnected components of about equal size. In this work, we introduce novel algorithms to find small node separators in large graphs. With focus on solution quality, we introduce novel flow-based local search algorithms which are integrated in a multilevel framework. In addition, we transfer techniques successfully used in the graph partitioning field. This includes the usage of edge ratings tailored to our problem to guide the graph coarsening algorithm as well as highly localized local search and iterated multilevel cycles to improve solution quality even further. Experiments indicate that flow-based local search algorithms on its own in a multilevel framework are already highly competitive in terms of separator quality. Adding additional local search algorithms further improves solution quality. Our strongest configuration almost always outperforms competing systems while on average computing 10% and 62% smaller separators than Metis and Scotch, respectively.

cs.DS↗

arXiv2015

Finding Near-Optimal Independent Sets at Scale

The independent set problem is NP-hard and particularly difficult to solve in large sparse graphs. In this work, we develop an advanced evolutionary algorithm, which incorporates kernelization techniques to compute large independent sets in huge sparse networks. A recent exact algorithm has shown that large networks can be solved exactly by employing a branch-and-reduce technique that recursively kernelizes the graph and performs branching. However, one major drawback of their algorithm is that, for huge graphs, branching still can take exponential time. To avoid this problem, we recursively choose vertices that are likely to be in a large independent set (using an evolutionary approach), then further kernelize the graph. We show that identifying and removing vertices likely to be in large independent sets opens up the reduction space---which not only speeds up the computation of large independent sets drastically, but also enables us to compute high-quality independent sets on much larger instances than previously reported in the literature.

cs.DS↗