Package org.apache.sysds.runtime.compress.cost

Class DistinctCostEstimator

java.lang.Object

org.apache.sysds.runtime.compress.cost.ACostEstimate

org.apache.sysds.runtime.compress.cost.DistinctCostEstimator

All Implemented Interfaces:

Serializable

public class DistinctCostEstimator
extends ACostEstimate

A Cost based estimator that based the compression and co-coding cost on the number of distinct elements in the column groups. The base cost of the uncompressed representation is the number of cells in the matrix that has a value. Aka nonzero values.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor	Description
DistinctCostEstimator(int nRows, CompressionSettings cs, double sparsity)

Method Summary

Modifier and Type	Method	Description
double	getCost(AColGroup cg, int nRows)	Get the cost of a compressed columnGroup.
double	getCost(MatrixBlock mb)	Get the cost of a matrix block.
boolean	shouldSparsify()	Ask the cost estimator if it is a good idea to try to sparsify a column group.

Methods inherited from class org.apache.sysds.runtime.compress.cost.ACostEstimate

getCost, getCost, getCost, getCost, toString

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

DistinctCostEstimator

public DistinctCostEstimator(int nRows,
                             CompressionSettings cs,
                             double sparsity)

Method Detail

getCost

public double getCost(MatrixBlock mb)

Description copied from class: ACostEstimate

Get the cost of a matrix block.

Specified by:

getCost in class ACostEstimate

Parameters:

mb - A MatrixBlock

Returns:

The cost subject to the internal cost functions

getCost

public double getCost(AColGroup cg,
                      int nRows)

Description copied from class: ACostEstimate

Get the cost of a compressed columnGroup.

Specified by:

getCost in class ACostEstimate

Parameters:

cg - A ColumnGroup

nRows - The number of rows in the column group

Returns:

The cost subject to the internal cost functions

shouldSparsify

public boolean shouldSparsify()

Description copied from class: ACostEstimate

Ask the cost estimator if it is a good idea to try to sparsify a column group. It is the same as asking if it is a good idea to make FOR on top of the column group.

Specified by:

shouldSparsify in class ACostEstimate

Returns:

true if yes