Browse Prior Art Database

Storage of Sparsely Populated Data Matrices

IP.com Disclosure Number: IPCOM000073787D
Original Publication Date: 1971-Feb-01
Included in the Prior Art Database: 2005-Feb-23
Document File: 2 page(s) / 26K

Publishing Venue

IBM

Related People

Manifold, GW: AUTHOR

Abstract

The efficient core storage of sparsely populated n-dimensional data arrays is a well known programming problem and numerous solutions have been proposed. With the present method, the storage area required for data is greatly reduced by employing an intermediate matrix of subscripts (termed pointers) which index into a compact, single-dimension data array. As compared with previous techniques, this method facilitates easy access to each data element in the n-dimensional array as a function of its n coordinates. The amount of core saved depends on the relative sparsity of the data elements within the array, as well as the ratio between the core required to store one data element and that required to store one pointer.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 64% of the total text.

Page 1 of 2

Storage of Sparsely Populated Data Matrices

The efficient core storage of sparsely populated n-dimensional data arrays is a well known programming problem and numerous solutions have been proposed. With the present method, the storage area required for data is greatly reduced by employing an intermediate matrix of subscripts (termed pointers) which index into a compact, single-dimension data array. As compared with previous techniques, this method facilitates easy access to each data element in the n-dimensional array as a function of its n coordinates. The amount of core saved depends on the relative sparsity of the data elements within the array, as well as the ratio between the core required to store one data element and that required to store one pointer.

For example, a reporting system is to generate two sets of management reports for each organization within a facility. One set of reports is to be summarized by department by Standard Expense Account (SEC) and the other set is to be tiered by SEC by department. A maximum of 20 departments per organization is allotted, and 200 different SEC's can be encountered within each organization. It is expected, however, that only 25O SEC/department combinations will be encountered within an organization. Assume that 4 bytes are required for each pointer entry and 40 bytes of data must be stored for each SEC/department combination. Define: 'n' as the dimension of the array to be stored; 'a(i)' as the number of elements...