org.ejml.simple.ops.SimpleOperations_DSCC 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
A fast and easy to use dense and sparse matrix linear algebra library written in Java.
The newest version!
/*
 * Copyright (c) 2023, 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.concurrency.EjmlConcurrency;
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.CommonOps_MT_DSCC;
import org.ejml.sparse.csc.MatrixFeatures_DSCC;
import org.ejml.sparse.csc.NormOps_DSCC;
import org.ejml.sparse.csc.mult.Workspace_MT_DSCC;
import pabeles.concurrency.GrowArray;

import java.io.PrintStream;

import static org.ejml.concurrency.EjmlConcurrency.useConcurrent;

/**
 * Implementation of {@link org.ejml.simple.SimpleOperations} for {@link DMatrixSparseCSC}.
 *
 * @author Peter Abeles
 */
public class SimpleOperations_DSCC implements SimpleSparseOperations {

    // Workspace variables
    public transient IGrowArray gw = new IGrowArray();
    public transient DGrowArray gx = new DGrowArray();

    // Workspace for concurrent algorithms
    public transient GrowArray workspaceMT = new GrowArray<>(Workspace_MT_DSCC::new);
    public transient GrowArray workspaceA = new GrowArray<>(DGrowArray::new);

    @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 /**/double getReal( DMatrixSparseCSC A, int row, int column ) {
        return A.get(row, column);
    }

    @Override public /**/double getImaginary( DMatrixSparseCSC A, int row, int column ) {
        return 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 ) {
        if (useConcurrent(A) || useConcurrent(B)) {
            CommonOps_MT_DSCC.mult(A, B, output, workspaceMT);
        } else {
            CommonOps_DSCC.mult(A, B, output);
        }
    }

    @Override public void multTransA( DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output ) {
        var At = new DMatrixSparseCSC(1, 1);
        CommonOps_DSCC.transpose(A, At, gw);

        if (useConcurrent(A) || useConcurrent(B)) {
            CommonOps_MT_DSCC.mult(A, B, output, workspaceMT);
        } else {
            CommonOps_DSCC.mult(At, 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 ) {
        if (useConcurrent(A) || useConcurrent(B)) {
            CommonOps_MT_DSCC.multTransA(A, B, output, workspaceA);
        } else {
            CommonOps_DSCC.multTransA(A, B, output, null);
        }
    }

    @Override public void mult( DMatrixSparseCSC A, DMatrixRMaj B, DMatrixRMaj output ) {
        if (useConcurrent(A)) {
            CommonOps_MT_DSCC.mult(A, B, output, workspaceA);
        } else {
            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 for DSCC. Make a feature request if you need this!");
    }

    @Override public void plus( DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output ) {
        if (EjmlConcurrency.useConcurrent(A)) {
            CommonOps_MT_DSCC.add(1, A, 1, B, output, workspaceMT);
        } else {
            CommonOps_DSCC.add(1, A, 1, B, output, null, null);
        }
    }

    @Override public void minus( DMatrixSparseCSC A, DMatrixSparseCSC B, DMatrixSparseCSC output ) {
        if (EjmlConcurrency.useConcurrent(A)) {
            CommonOps_MT_DSCC.add(1, A, -1, B, output, workspaceMT);
        } else {
            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 ) {
        if (useConcurrent(A) || useConcurrent(b)) {
            CommonOps_MT_DSCC.add(1, A, (double)beta, b, output, workspaceMT);
        } else {
            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 ) {
        if (useConcurrent(A) || useConcurrent(b)) {
            CommonOps_MT_DSCC.add((double)alpha, A, (double)beta, b, output, workspaceMT);
        } else {
            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 /**/double[] getRow( DMatrixSparseCSC A, int row, int col0, int col1 ) {
        var v = new /**/double[col1 - col0];

        // Exhaustively search every column in the allowed range for rows that match the target
        // If a match is found copy it's value
        for (int col = col0; col < col1; col++) {
            int rowIdx0 = A.col_idx[col];
            int rowIdx1 = A.col_idx[col + 1];

            for (int i = rowIdx0; i < rowIdx1; i++) {
                if (row != A.nz_rows[i])
                    continue;
                v[col - col0] = A.nz_values[i];
            }
        }

        return v;
    }

    @Override public /**/double[] getColumn( DMatrixSparseCSC A, int col, int row0, int row1 ) {
        var v = new /**/double[row1 - row0];

        // Go through the target column and find all row elements within the allowed range
        int rowIdx0 = A.col_idx[col];
        int rowIdx1 = A.col_idx[col + 1];

        for (int i = rowIdx0; i < rowIdx1; i++) {
            int row = A.nz_rows[i];
            if (row < row0 || row >= row1)
                continue;
            v[row - row0] = A.nz_values[i];
        }

        return v;
    }

    @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 elementMax( DMatrixSparseCSC A ) {
        return CommonOps_DSCC.elementMax(A);
    }

    @Override public /**/double elementMin( DMatrixSparseCSC A ) {
        return CommonOps_DSCC.elementMin(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, (double)tol);
    }

    @Override public void print( PrintStream out, Matrix mat, String format ) {
        MatrixIO.print(out, (DMatrixSparseCSC)mat, format);
    }

    @Override public void elementOp( DMatrixSparseCSC A, ElementOpReal op, DMatrixSparseCSC output ) {
        // Ensure the output has the same non-zero elements as A
        output.copyStructure(A);

        for (int col = 0; col < A.numCols; col++) {
            int idx0 = A.col_idx[col];
            int idx1 = A.col_idx[col + 1];

            for (int i = idx0; i < idx1; i++) {
                int row = A.nz_rows[i];
                double value = A.nz_values[i];

                output.nz_values[i] = (double)op.op(row, col, value);
            }
        }
    }

    @Override public void elementOp( DMatrixSparseCSC A, ElementOpComplex op, DMatrixSparseCSC output ) {
        throw new ConvertToImaginaryException();
    }
}