Automatic data decomposition for message-passing machines

Author(s):  
Mirela Damian-Iordache ◽  
Sriram V. Pemmaraju
1995 ◽  
Vol 05 (04) ◽  
pp. 539-550 ◽  
Author(s):  
QI NING ◽  
VINCENT VAN DONGEN ◽  
GUANG R. GAO

In this paper, we develop an automatic compile-time computation and data decomposition technique for distributed-memory machines. Our method handles complex programs containing perfect and non-perfect loop nests with or without loop-carried dependences. Applying our algorithms, a program will be divided into collections (called clusters) of loop nests, such that data redistributions are allowed only between the clusters. Within each cluster of loop nests, decomposition and data locality constraints are formulated as a system of homogeneous linear equations which is solved by polynomial time algorithms. Our algorithm can selectively relax data locality constraints within a cluster to achieve a balance between parallelism and data locality. Such relaxations are guided by exploiting the hierarchical program nesting structures from outer to inner nesting levels to keep the communications at a outer-most level possible.


Author(s):  
B. Ralph ◽  
A.R. Jones

In all fields of microscopy there is an increasing interest in the quantification of microstructure. This interest may stem from a desire to establish quality control parameters or may have a more fundamental requirement involving the derivation of parameters which partially or completely define the three dimensional nature of the microstructure. This latter categorey of study may arise from an interest in the evolution of microstructure or from a desire to generate detailed property/microstructure relationships. In the more fundamental studies some convolution of two-dimensional data into the third dimension (stereological analysis) will be necessary.In some cases the two-dimensional data may be acquired relatively easily without recourse to automatic data collection and further, it may prove possible to perform the data reduction and analysis relatively easily. In such cases the only recourse to machines may well be in establishing the statistical confidence of the resultant data. Such relatively straightforward studies tend to result from acquiring data on the whole assemblage of features making up the microstructure. In this field data mode, when parameters such as phase volume fraction, mean size etc. are sought, the main case for resorting to automation is in order to perform repetitive analyses since each analysis is relatively easily performed.


1978 ◽  
Vol 17 (01) ◽  
pp. 36-40 ◽  
Author(s):  
J.-P. Durbec ◽  
Jaqueline Cornée ◽  
P. Berthezene

The practice of systematic examinations in hospitals and the increasing development of automatic data processing permits the storing of a great deal of information about a large number of patients belonging to different diagnosis groups.To predict or to characterize these diagnosis groups some descriptors are particularly useful, others carry no information. Data screening based on the properties of mutual information and on the log cross products ratios in contingency tables is developed. The most useful descriptors are selected. For each one the characterized groups are specified.This approach has been performed on a set of binary (presence—absence) radiological variables. Four diagnoses groups are concerned: cancer of pancreas, chronic calcifying pancreatitis, non-calcifying pancreatitis and probable pancreatitis. Only twenty of the three hundred and forty initial radiological variables are selected. The presence of each corresponding sign is associated with one or more diagnosis groups.


Sign in / Sign up

Export Citation Format

Share Document