codistributor1d

1-D distribution scheme for codistributed array

Description

Use a codistributor1d object to define the one-dimensional distribution scheme for a codistributed array. The one-dimensional codistributor distributes arrays along a single specified dimension, the distribution dimension, in a noncyclic, partitioned manner.

Creation

Syntax

codist = codistributor1d

codist = codistributor1d(Dimension)

codist = codistributor1d(Dimension,Partition)

codist = codistributor1d(Dimension,Partition,gsize)

Description

codist = codistributor1d returns a codistributor1d object using the default dimension and partition. The default dimension is the last nonsingleton dimension of the codistributed array. The default partition distributes the array along the default dimension as evenly as possible.

codist = codistributor1d(Dimension) returns a 1-D codistributor object for distribution along the dimension specified by the Dimension property. For example, if Dimension is 1, the function distributes the object along rows.

codist = codistributor1d(Dimension,Partition) also returns a 1-D codistributor object for distribution according to the partition vector specified by the Partition property. For example, C1 = codistributor1d(1,[1,2,3,4]) distributes an array of 10 rows to four workers, with one row to the first worker, two rows to the second worker, three rows to the third worker, and four rows to the fourth worker.

The resulting codistributor of any of the above syntaxes is incomplete because its global size is not specified. Use a codistributor constructed this way as an argument to other functions as a template codistributor when creating codistributed arrays.

codist = codistributor1d(Dimension,Partition,gsize) returns a codistributor object with the global size gsize.

You can use the resulting codistributor object to build a codistributed array from its local parts with codistributed.build. To use a default dimension, specify codistributor1d.unsetDimension for the Dimension property; the function derives the distribution dimension from gsize and selects the last nonsingleton dimension as the default dimension. Similarly, to use a default partition, specify codistributor1d.unsetPartition for the Partition property; the function derives the default partition from the global size and distribution dimension.

The local part on worker workerIndex of a codistributed array using such a codistributor is of size gsize in all dimensions except dimension, where the size is part(workerIndex). The local part has the same class and attributes as the overall codistributed array. The overall global array can be reconstructed by concatenating the various local parts along dimension dimension.

Input Arguments

gsize — Global size of codistributed array

integer

Global size of the codistributed array, specified as an integer.

Properties

Dimension — Distribution dimension

scalar integer

Distribution dimension, specified as a scalar integer. The distribution dimension specifies the dimension over which you distribute the codistributed array.

Partition — Partitioning vector

integer row vector

Partitioning vector, specified as an integer row vector. The partitioning vector specifies the distribution of the codistributed array to the workers.

Object Functions

codistributed.cell	Create codistributed cell array
codistributed.colon	Distributed colon operation
codistributed.spalloc	Allocate space for sparse codistributed matrix
codistributed.speye	Create codistributed sparse identity matrix
codistributed.sprand	Create codistributed sparse array of uniformly distributed pseudo-random values
codistributed.sprandn	Create codistributed sparse array of normally distributed pseudo-random values
eye	Create codistributed identity matrix
false	Create codistributed array of logical 0 (false)
globalIndices	Global indices for local part of codistributed array
Inf	Create codistributed array of all Inf values
isComplete	True if codistributor object is complete
NaN	Create codistributed array of all NaN values
ones	Create codistributed array of all ones
rand	Create codistributed array of uniformly distributed random numbers
randn	Create codistributed array of normally distributed random numbers
sparse	Create codistributed sparse matrix
true	Create codistributed array of logical 1 (true)
zeros	Create codistributed array of all zeros

Examples

Create Codistributed Arrays with Distribution Scheme

Use a codistributor1d object to create an N-by-N matrix of ones, distributed by rows.

N = 1000;
spmd
    codistr = codistributor1d(1); % 1st dimension (rows)
    C = ones(N,codistr);
end

Fully Specify Codistributor1d Object and Create Codistributed Array

Use a fully specified codistributor1d object to create a N-by-N codistributed matrix from its local parts. Then visualize which elements are stored on worker 2.

Start with full sized array on each worker then set myLocalSize to default part of whole array.

N = 1000;
spmd
    codistr = codistributor1d( ...
                    codistributor1d.unsetDimension, ...
                    codistributor1d.unsetPartition, ...
                    [N N]);
    myLocalSize = [N N]; % 
    myLocalSize(codistr.Dimension) = codistr.Partition(spmdIndex);
    myLocalPart = spmdIndex*ones(myLocalSize);
    D = codistributed.build(myLocalPart,codistr);
end
spy(D==2);

Version History

Introduced in R2009b