FloatVectorAbs.java
package org.djunits.value.vfloat.vector.base;
import org.djunits.unit.AbsoluteLinearUnit;
import org.djunits.unit.Unit;
import org.djunits.value.Absolute;
import org.djunits.value.ValueRuntimeException;
import org.djunits.value.vfloat.function.FloatMathFunctions;
import org.djunits.value.vfloat.scalar.base.FloatScalarAbs;
import org.djunits.value.vfloat.scalar.base.FloatScalarRelWithAbs;
import org.djunits.value.vfloat.vector.data.FloatVectorData;
/**
* AbstractMutableFloatVectorRelWithAbs.java.
* <p>
* Copyright (c) 2019-2024 Delft University of Technology, PO Box 5, 2600 AA, Delft, the Netherlands. All rights reserved. <br>
* BSD-style license. See <a href="https://djunits.org/docs/license.html">DJUNITS License</a>.
* </p>
* @author <a href="https://www.tudelft.nl/averbraeck" target="_blank">Alexander Verbraeck</a>
* @param <AU> the absolute unit belonging to the relative unit
* @param <A> the absolute scalar type belonging to the absolute vector type
* @param <AV> the (immutable or mutable) absolute vector type
* @param <RU> the relative unit belonging to the absolute unit
* @param <R> the relative scalar type belonging to the relative vector type
* @param <RV> the relative (immutable or mutable) vector type with this unit
*/
// @formatter:off
public abstract class FloatVectorAbs<
AU extends AbsoluteLinearUnit<AU, RU>,
A extends FloatScalarAbs<AU, A, RU, R>,
AV extends FloatVectorAbs<AU, A, AV, RU, R, RV>,
RU extends Unit<RU>,
R extends FloatScalarRelWithAbs<AU, A, RU, R>,
RV extends FloatVectorRelWithAbs<AU, A, AV, RU, R, RV>>
extends FloatVector<AU, A, AV>
implements Absolute<AU, AV, RU, RV>
// @formatter:on
{
/** */
private static final long serialVersionUID = 20190908L;
/**
* Construct a new Relative Mutable FloatVector.
* @param data FloatVectorData; an internal data object
* @param unit AU; the unit
*/
protected FloatVectorAbs(final FloatVectorData data, final AU unit)
{
super(data.copy(), unit);
}
@Override
public AV plus(final RV increment) throws ValueRuntimeException
{
return instantiateVector(this.getData().plus(increment.getData()), getDisplayUnit());
}
@Override
public AV minus(final RV decrement) throws ValueRuntimeException
{
return instantiateVector(this.getData().minus(decrement.getData()), getDisplayUnit());
}
@Override
public RV minus(final AV decrement) throws ValueRuntimeException
{
return instantiateVectorRel(this.getData().minus(decrement.getData()), decrement.getDisplayUnit().getRelativeUnit());
}
/**
* Decrement all values of this vector by the decrement. This only works if the vector is mutable.
* @param decrement R; the scalar by which to decrement all values
* @return AV; this modified vector
* @throws ValueRuntimeException in case this vector is immutable
*/
@SuppressWarnings("unchecked")
public AV decrementBy(final R decrement)
{
checkCopyOnWrite();
assign(FloatMathFunctions.DEC(decrement.si));
return (AV) this;
}
/**
* Decrement all values of this vector by the decrement on a value by value basis. This only works if this vector is
* mutable.
* @param decrement RV; the vector that contains the values by which to decrement the corresponding values
* @return AV; this modified vector
* @throws ValueRuntimeException in case this vector is immutable or when the sizes of the vectors differ
*/
@SuppressWarnings("unchecked")
public AV decrementBy(final RV decrement)
{
checkCopyOnWrite();
getData().decrementBy(decrement.getData());
return (AV) this;
}
/**
* Instantiate a new relative vector of the class of this absolute vector. This can be used instead of the
* FloatVector.instiantiate() methods in case another vector of this absolute vector class is known. The method is faster
* than FloatVector.instantiate, and it will also work if the vector is user-defined.
* @param dvd FloatVectorData; the data used to instantiate the vector
* @param displayUnit RU; the display unit of the relative vector
* @return RV; a relative vector of the correct type, belonging to this absolute vector type
*/
public abstract RV instantiateVectorRel(FloatVectorData dvd, RU displayUnit);
/**
* Instantiate a new relative scalar for the class of this absolute vector. This can be used instead of the
* FloatScalar.instiantiate() methods in case a vector of this class is known. The method is faster than
* FloatScalar.instantiate, and it will also work if the vector and/or scalar are user-defined.
* @param valueSI float; the SI value of the relative scalar
* @param displayUnit RU; the unit in which the relative value will be displayed
* @return R; a relative scalar of the correct type, belonging to this absolute vector type
*/
public abstract R instantiateScalarRelSI(float valueSI, RU displayUnit);
}