magpie.data.utilities.normalizers

Class RegularSolutionNormalizer

java.lang.Object

magpie.data.utilities.normalizers.BaseDatasetNormalizer

magpie.data.utilities.normalizers.RegularSolutionNormalizer

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable, Commandable, Options, Printable

public class RegularSolutionNormalizer
extends BaseDatasetNormalizer

Normalize data using regular solution model. Goal is to transform class variable based on stoichiometry of material by fitting a regular solution model to the class variable given associated composition.

Let y be the variable to be normalized, x be the composition of the material where x[Z] is the fraction of element with atomic number Z, and y' be the normalized class variable.

y = y' * sum[ x[i] * x[j] ] for j>i ]

This model is designed to normalize class values where the class must be equal to zero for unary compositions, and increases in magnitude when the amount of mixing is large (e.g., formation energy).

Note: You should filter out unary compositions before using this filter. This normalization procedure would lead to division by zero for unary compounds. For numerical stability, we multiple them by -1 during the normalization procedure, but we recommend you filter out unary entries before running this normalizer.

Usage: *No Options*

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
RegularSolutionNormalizer()

Method Summary

Modifier and Type	Method and Description
static double	computeMixingTerm(CompositionEntry entry) Compute the mixing term for a solution model.
protected void	normalizeAttributes(Dataset Data) Perform the actual normalization on the attributes
protected void	normalizeClassVariable(Dataset Data) Perform the actual normalization on the class variable.
java.lang.String	printUsage() Print out required format for options.
protected void	restoreAttributes(Dataset Data) Perform the actual restoration on the attributes
protected void	restoreClassVariable(Dataset Data) Perform the actual restoration on the class variables (measured and predicted!)
void	setOptions(java.util.List<java.lang.Object> Options) Set any options for this object.
protected void	trainOnAttributes(Dataset Data) Perform the actual training work for attributes.
protected void	trainOnMeasuredClass(Dataset Data) Perform the actual training work for the class variable.

Methods inherited from class magpie.data.utilities.normalizers.BaseDatasetNormalizer

about, clone, isTrained, normalize, printCommand, printDescription, restore, runCommand, setToNormalizeAttributes, setToNormalizeClass, test, train, willNormalizeAttributes, willNormalizeClass

Methods inherited from class java.lang.Object

equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

RegularSolutionNormalizer

public RegularSolutionNormalizer()

Method Detail

setOptions

public void setOptions(java.util.List<java.lang.Object> Options)
                throws java.lang.Exception

Description copied from interface: Options

Set any options for this object. If parsing the options fails, return an Exception with the usage as its message.

Parameters:

Options - Array of options as Objects - can be null

Throws:

java.lang.Exception - if problem with inputs

printUsage

public java.lang.String printUsage()

Description copied from interface: Options

Print out required format for options. For consistency, should start with "Usage: ".

Returns:

Usage for this function

trainOnAttributes

protected void trainOnAttributes(Dataset Data)

Description copied from class: BaseDatasetNormalizer

Perform the actual training work for attributes. Must establish the parameters necessary to normalize attributes and return them to their original range.

Specified by:

trainOnAttributes in class BaseDatasetNormalizer

Parameters:

Data - Training set

trainOnMeasuredClass

protected void trainOnMeasuredClass(Dataset Data)

Description copied from class: BaseDatasetNormalizer

Perform the actual training work for the class variable. Must establish the parameters necessary to normalize attributes and return them to their original range.

NOTE: Use the measured class variable to train normalizer.

Specified by:

trainOnMeasuredClass in class BaseDatasetNormalizer

Parameters:

Data - Training set

normalizeAttributes

protected void normalizeAttributes(Dataset Data)

Description copied from class: BaseDatasetNormalizer

Perform the actual normalization on the attributes

Specified by:

normalizeAttributes in class BaseDatasetNormalizer

Parameters:

Data - Dataset to be transformed

restoreAttributes

protected void restoreAttributes(Dataset Data)

Description copied from class: BaseDatasetNormalizer

Perform the actual restoration on the attributes

Specified by:

restoreAttributes in class BaseDatasetNormalizer

Parameters:

Data - Dataset to be transformed

normalizeClassVariable

protected void normalizeClassVariable(Dataset Data)

Description copied from class: BaseDatasetNormalizer

Perform the actual normalization on the class variable. Predicted and measured class variables must be transformed the same way

Specified by:

normalizeClassVariable in class BaseDatasetNormalizer

Parameters:

Data - Dataset to be transformed

restoreClassVariable

protected void restoreClassVariable(Dataset Data)

Description copied from class: BaseDatasetNormalizer

Perform the actual restoration on the class variables (measured and predicted!)

Specified by:

restoreClassVariable in class BaseDatasetNormalizer

Parameters:

Data - Dataset to be transformed

computeMixingTerm

public static double computeMixingTerm(CompositionEntry entry)

Compute the mixing term for a solution model. This is defined as

sum[ x[i] * x[j] ] for i where x[Z] is the fraction of element Z.

Note: If number of elements is 1, returns -1.

Parameters:

entry - Entry for which to compute mixing term

Returns:

Value of mixing term