Publication: DSJM: A Software Toolkit for Direct Determination of Sparse Jacobian Matrices
Introduction
Applications
Tools
Research Groups
Workshops
Publications
   List Publications
   Advanced Search
   Info
   Add Publications
My Account
About
Impress

DSJM: A Software Toolkit for Direct Determination of Sparse Jacobian Matrices

- Part of a collection -
 

Author(s)
Mahmudul Hasan , Shahadat Hossain , Ahamad Imtiaz Khan , Nasrin Hakim Mithila , Ashraful Huq Suny

Published in
Mathematical Software -- ICMS 2016, 5th International Conference, Berlin, Germany, July 11--14, 2016,

Editor(s)
Gert-Martin Greuel, Thorsten Koch, Peter Paule, Andrew Sommese

Year
2016

Publisher
Springer International Publishing

Abstract
We describe the main design features of DSJM (Determine Sparse Jacobian Matrices), a software toolkit written in standard C++ that enables direct determination of sparse Jacobian matrices. Our design exploits the recently proposed unifying framework ``pattern graph″ and employs cache-friendly array-based sparse data structures. The DSJM implements a greedy grouping (coloring) algorithm and several ordering heuristics. In our numerical testing on a suite of large-scale test instances DSJM consistently produced better timing and partitions compared with a similar software.

AD Theory and Techniques
graph coloring, Sparsity

BibTeX
@INPROCEEDINGS{
         Hasan2016DAS,
       author = "Hasan, Mahmudul and Hossain, Shahadat and Khan, Ahamad Imtiaz and Mithila, Nasrin
         Hakim and Suny, Ashraful Huq",
       editor = "Greuel, Gert-Martin and Koch, Thorsten and Paule, Peter and Sommese, Andrew",
       title = "{DSJM}: A Software Toolkit for Direct Determination of Sparse {J}acobian Matrices",
       booktitle = "Mathematical Software -- {ICMS} 2016, 5th International Conference, Berlin,
         Germany, July 11--14, 2016,",
       year = "2016",
       volume = "9725",
       series = "Lecture Notes in Computer Science",
       publisher = "Springer International Publishing",
       address = "Cham",
       pages = "275--283",
       abstract = "We describe the main design features of DSJM (Determine Sparse Jacobian Matrices),
         a software toolkit written in standard C++ that enables direct determination of sparse Jacobian
         matrices. Our design exploits the recently proposed unifying framework ``pattern graph''
         and employs cache-friendly array-based sparse data structures. The DSJM implements a greedy grouping
         (coloring) algorithm and several ordering heuristics. In our numerical testing on a suite of
         large-scale test instances DSJM consistently produced better timing and partitions compared with a
         similar software.",
       isbn = "978-3-319-42432-3",
       doi = "10.1007/978-3-319-42432-3_34",
       url = "https://doi.org/10.1007/978-3-319-42432-3_34",
       ad_theotech = "graph coloring, Sparsity"
}


back
  

Contact:
autodiff.org
Username:
Password:
(lost password)