Distributed Gröbner Bases Computation with MPJ

Distributed Gröbner Bases Computation with MPJ
复制标题

使用 MPJ 进行分布式 Grobner 基础计算

DOI:
10.1109/waina.2013.101
复制
发表时间:
2013
期刊:
2013 27th International Conference on Advanced Information Networking and Applications Workshops
影响因子:
--
通讯作者:
Heinz Kredel
Heinz Kredel
中科院分区:
--
文献类型:
--
作者:
Heinz Kredel

文献摘要

被引文献

相似文献

Groebner基是计算机代数的一个重要组成部分,它在具有多cpu计算节点的分布式内存并行计算机上的计算越来越成功。在以往工作的基础上,我们提出并比较了使用MPJ实现的MPI消息传递标准的新程序。MPJ是针对Java的MPI规范,具有高效的实现。此外,新版本在主节点和工作节点之间使用一个控制通信通道。在混合版本中,多项式在节点上的线程之间共享,而在纯分布式版本中,多项式在每个进程上复制。多项式在一个单独的分布式数据结构中异步传输到算法的控制流。实现是通用的,适用于所有实现的字段。我们提出了第一个有希望的性能测量。
Groebner bases, an important building block in computer algebra, are computed on distributed memory parallel computers with multi-CPU compute nodes with increasing success. Based on previous efforts we present and compare new programs using the MPI message passing standard implemented using MPJ. MPJ is a specification of MPI for Java with efficient implementations. Also the new versions use one control communication channel between the master node and the worker nodes. In the hybrid version the polynomials are shared between the threads on a node and in the pure distributed version the polynomials are replicated on every process. The polynomials are transported asynchronous to the control-flow of the algorithm in a separate distributed data structure. The implementation is generic and works for all implemented fields. We present first promising performance measurements.