Revision of DASHMM: Dynamic Adaptive System for Hierarchical Multipole Methods

Journal Article · Mon Jan 01 00:00:00 EST 2018 · Communications in Computational Physics

DOI:https://doi.org/10.4208/cicp.230517.300617sw· OSTI ID:1524174

DeBuhr, J. ^[1]; Zhang, B. ^[1]; Sterling, T. ^[1]

---

1. Indiana Univ., Bloomington, IN (United States). Center for Research in Extreme Scale Techologies, School of Informatics and Computing

We present version 1.2.0 of DASHMM, a general library implementing hierarchical multipole methods using the asynchronous multi-tasking HPX-5 runtime system. Compared with the previous release [10], this new version: (1) enables execution in both shared and distributed memory architectures; (2) extends DASHMM’s infrastructure to support advanced multipole methods [18]; and (3) provides built-in implementations of both the Yukawa [15] potential and Helmholtz [16] potential in the low frequency regime. These additions have not impacted the user interface, which remains simple and extensible.

• Research Organization: Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States). National Energy Research Scientific Computing Center (NERSC)
• Sponsoring Organization: USDOE Office of Science (SC)
• DOE Contract Number: AC02-05CH11231
• OSTI ID: 1524174
• Journal Information: Communications in Computational Physics, Vol. 23, Issue 1; ISSN 1815-2406
• Country of Publication: United States
• Language: English