Title :
Algorithm Research for Vector-Linked List Sparse Matrix Multiplication
Author :
Hu, Jikuan ; Wang, Weiqing
Author_Institution :
Dept. of Inf. Manage., Southwest Univ., Chongqing, China
Abstract :
The storage of non-zero elements of sparse matrix in the linked lists can dispense with transpose operation in the multiplication process, because multiplication shall be conducted only for elements in the two sparse matrices whose line number and column number are the same and whose values are not zero. As a result, the number of multiplications will be reduced. In the meantime, the use of linked list to store the elements of matrix will cut back the times of scanning the matrix elements. The theoretical analysis and experimental results show that algorithm of Vector-linked list sparse matrix multiplication has greater time efficiency and needs less storage space. It will lay the foundation for research in the field of image processing.
Keywords :
matrix multiplication; sparse matrices; image processing; multiplication process; sparse matrix multiplication; transpose operation; vector-linked list; Algorithm design and analysis; Focusing; Functional analysis; Graphics; Image coding; Image processing; Information management; Sparse matrices; Wearable computers; linked list; sparse factor; sparse matrix; vector;
Conference_Titel :
Wearable Computing Systems (APWCS), 2010 Asia-Pacific Conference on
Conference_Location :
Shenzhen
Print_ISBN :
978-1-4244-6467-8
Electronic_ISBN :
978-1-4244-6468-5
DOI :
10.1109/APWCS.2010.36