org.ejml.simple.ops.SimpleOperations_SPARSE Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of ejml-simple Show documentation
Show all versions of ejml-simple Show documentation
A fast and easy to use dense and sparse matrix linear algebra library written in Java.
/*
* Copyright (c) 2009-2019, Peter Abeles. All Rights Reserved.
*
* This file is part of Efficient Java Matrix Library (EJML).
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.ejml.simple.ops;
import org.ejml.data.*;
import org.ejml.ops.MatrixIO;
import org.ejml.simple.ConvertToDenseException;
import org.ejml.simple.ConvertToImaginaryException;
import org.ejml.simple.SimpleSparseOperations;
import org.ejml.sparse.csc.CommonOps_DSCC;
import org.ejml.sparse.csc.MatrixFeatures_DSCC;
import org.ejml.sparse.csc.NormOps_DSCC;
import java.io.PrintStream;
/**
* @author Peter Abeles
*/
public class SimpleOperations_SPARSE implements SimpleSparseOperations {
// Workspace variables
public transient IGrowArray gw = new IGrowArray();
public transient DGrowArray gx = new DGrowArray();
@Override
public void set(DMatrixSparseCSC A, int row, int column, /**/double value) {
A.set(row,column, (double)value);
}
@Override
public void set(DMatrixSparseCSC A, int row, int column, /**/double real, /**/double imaginary) {
throw new ConvertToImaginaryException();
}
@Override
public double get(DMatrixSparseCSC A, int row, int column) {
return A.get(row,column);
}
@Override
public void get(DMatrixSparseCSC A, int row, int column, Complex_F64 value) {
value.real = A.get(row,column);
value.imaginary = 0;
}
@Override
public void fill(DMatrixSparseCSC A, double value) {
if( value == 0 ) {
A.zero();
} else {
throw new ConvertToDenseException();
}
}
@Override
public void transpose(DMatrixSparseCSC input, DMatrixSparseCSC output) {
CommonOps_DSCC.transpose(input,output,gw);
}
@Override
public void mult(DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output) {
CommonOps_DSCC.mult(A,B,output);
}
@Override
public void multTransA(DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output) {
CommonOps_DSCC.multTransA(A,B,output,gw,gx);
}
@Override
public void extractDiag(DMatrixSparseCSC input, DMatrixRMaj output) {
CommonOps_DSCC.extractDiag(input,output);
}
@Override
public void multTransA(DMatrixSparseCSC A, DMatrixRMaj B, DMatrixRMaj output) {
CommonOps_DSCC.multTransA(A,B,output);
}
public void mult(DMatrixSparseCSC A , DMatrixRMaj B , DMatrixRMaj output ) {
CommonOps_DSCC.mult(A,B,output);
}
@Override
public void kron(DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output) {
// CommonOps_DSCC.kron(A,B,output);
throw new RuntimeException("Unsupported");
}
@Override
public void plus(DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output) {
CommonOps_DSCC.add(1,A,1,B,output, null, null);
}
@Override
public void minus(DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output) {
CommonOps_DSCC.add(1,A,-1,B,output, null, null);
}
@Override
public void minus(DMatrixSparseCSC A, /**/double b, DMatrixSparseCSC output) {
throw new ConvertToDenseException();
}
@Override
public void plus(DMatrixSparseCSC A, /**/double b, DMatrixSparseCSC output) {
throw new ConvertToDenseException();
}
@Override
public void plus(DMatrixSparseCSC A, /**/double beta, DMatrixSparseCSC b, DMatrixSparseCSC output) {
CommonOps_DSCC.add(1, A, (double)beta, b, output,gw,gx);
}
@Override
public void plus( /**/double alpha, DMatrixSparseCSC A, /**/double beta, DMatrixSparseCSC b, DMatrixSparseCSC output) {
CommonOps_DSCC.add( (double)alpha, A, (double)beta, b, output,gw,gx);
}
@Override
public /**/double dot(DMatrixSparseCSC A, DMatrixSparseCSC v) {
return CommonOps_DSCC.dotInnerColumns(A,0, v,0,gw,gx);
}
@Override
public void scale(DMatrixSparseCSC A, /**/double val, DMatrixSparseCSC output) {
CommonOps_DSCC.scale( (double)val, A,output);
}
@Override
public void divide(DMatrixSparseCSC A, /**/double val, DMatrixSparseCSC output) {
CommonOps_DSCC.divide( A, (double)val, output);
}
@Override
public boolean invert(DMatrixSparseCSC A, DMatrixSparseCSC output) {
return solve(A,output,CommonOps_DSCC.identity(A.numRows,A.numCols));
}
@Override
public void setIdentity(DMatrixSparseCSC A) {
CommonOps_DSCC.setIdentity(A);
}
@Override
public void pseudoInverse(DMatrixSparseCSC A, DMatrixSparseCSC output) {
throw new RuntimeException("Unsupported");
}
@Override
public boolean solve(DMatrixSparseCSC A, DMatrixSparseCSC X, DMatrixSparseCSC B) {
return CommonOps_DSCC.solve(A, X, B);
}
public boolean solve(DMatrixSparseCSC A, DMatrixRMaj X, DMatrixRMaj B) {
return CommonOps_DSCC.solve(A, X, B);
}
@Override
public void zero(DMatrixSparseCSC A) {
A.zero();
}
@Override
public /**/double normF(DMatrixSparseCSC A) {
return NormOps_DSCC.normF(A);
}
@Override
public /**/double conditionP2(DMatrixSparseCSC A) {
throw new RuntimeException("Unsupported");
}
@Override
public /**/double determinant(DMatrixSparseCSC A) {
return CommonOps_DSCC.det(A);
}
@Override
public /**/double trace(DMatrixSparseCSC A) {
return CommonOps_DSCC.trace(A);
}
@Override
public void setRow(DMatrixSparseCSC A, int row, int startColumn, /**/double... values) {
// TODO Update with a more efficient algorithm
for (int i = 0; i < values.length; i++) {
A.set(row, startColumn + i, (double)values[i]);
}
// check to see if value are zero, if so ignore them
// Do a pass through the matrix and see how many elements need to be added
// see if the existing storage is enough
// If it is enough ...
// starting from the tail, move a chunk, insert, move the next chunk, ...etc
// If not enough, create new arrays and construct it
}
@Override
public void setColumn(DMatrixSparseCSC A, int column, int startRow, /**/double... values) {
// TODO Update with a more efficient algorithm
for (int i = 0; i < values.length; i++) {
A.set(startRow + i, column, (double)values[i]);
}
}
@Override
public void extract(DMatrixSparseCSC src, int srcY0, int srcY1, int srcX0, int srcX1, DMatrixSparseCSC dst, int dstY0, int dstX0) {
CommonOps_DSCC.extract(src,srcY0,srcY1,srcX0,srcX1,dst,dstY0,dstX0);
}
@Override
public DMatrixSparseCSC diag(DMatrixSparseCSC A) {
DMatrixSparseCSC output;
if (MatrixFeatures_DSCC.isVector(A)) {
int N = Math.max(A.numCols,A.numRows);
output = new DMatrixSparseCSC(N,N);
CommonOps_DSCC.diag(output,A.nz_values,0,N);
} else {
int N = Math.min(A.numCols,A.numRows);
output = new DMatrixSparseCSC(N,1);
CommonOps_DSCC.extractDiag(A,output);
}
return output;
}
@Override
public boolean hasUncountable(DMatrixSparseCSC M) {
return MatrixFeatures_DSCC.hasUncountable(M);
}
@Override
public void changeSign(DMatrixSparseCSC a) {
CommonOps_DSCC.changeSign(a,a);
}
@Override
public /**/double elementMaxAbs(DMatrixSparseCSC A) {
return CommonOps_DSCC.elementMaxAbs(A);
}
@Override
public /**/double elementMinAbs(DMatrixSparseCSC A) {
return CommonOps_DSCC.elementMinAbs(A);
}
@Override
public /**/double elementSum(DMatrixSparseCSC A) {
return CommonOps_DSCC.elementSum(A);
}
@Override
public void elementMult(DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output) {
CommonOps_DSCC.elementMult(A,B,output,null,null);
}
@Override
public void elementDiv(DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output) {
throw new ConvertToDenseException();
}
@Override
public void elementPower(DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output) {
throw new ConvertToDenseException();
}
@Override
public void elementPower(DMatrixSparseCSC A, /**/double b, DMatrixSparseCSC output) {
throw new ConvertToDenseException();
}
@Override
public void elementExp(DMatrixSparseCSC A, DMatrixSparseCSC output) {
throw new ConvertToDenseException();
}
@Override
public void elementLog(DMatrixSparseCSC A, DMatrixSparseCSC output) {
throw new ConvertToDenseException();
}
@Override
public boolean isIdentical(DMatrixSparseCSC A, DMatrixSparseCSC B, double tol) {
return MatrixFeatures_DSCC.isEqualsSort(A, B, tol);
}
@Override
public void print(PrintStream out, Matrix mat, String format ) {
MatrixIO.print(out, (DMatrixSparseCSC)mat, format);
}
}