umontreal.iro.lecuyer.hups
Interface PointSetIterator

All Superinterfaces:

RandomStream

All Known Implementing Classes:

CycleBasedPointSet.CycleBasedPointSetIterator, CycleBasedPointSetBase2.CycleBasedPointSetBase2Iterator

public interface PointSetIterator
extends RandomStream

Objects of classes that implement this interface are iterators that permit one to enumerate (or observe) the successive points of a point set and the successive coordinates of these points. Each PointSetIterator is associated with a given point set and maintains a current point index i and a current coordinate index j, which are both initialized to zero.

Successive coordinates can be accessed one or many at a time by the methods nextCoordinate and nextCoordinates, respectively. The current coordinate index j can be set explicitely by setCurCoordIndex and resetCurCoordIndex. Similar methods are available for resetting and accessing the current point. The method nextPoint permits one to enumerate the successive points in natural order.

This class also implements the RandomStream interface. This permits one to replace random numbers by the coordinates of (randomized) quasi-Monte Carlo points without changing the code that calls the generators in a simulation program. That is, the same simulation program can be used for both Monte Carlo and quasi-Monte Carlo simulations. The method nextDouble does exactly the same as nextCoordinate, it returns the current coordinate of the current point and advances the current coordinate by one. The substreams correspond to the points, so resetStartSubstream resets the current point coordinate to zero, resetNextSubstream resets the iterator to the next point, and resetStartStream resets the iterator to the first point of the point set.

There can be several iterators over the same point set. These iterators are independent from each other. Classes that implement this interface must maintain enough information so that each iterator is unaffected by other iterator's operations. However, the iterator does not need to be independent of the underlying point set. If the point set is modified (e.g., randomized), the iterator may continue to work as usual.

Point set iterators are implemented as inner classes because this gives a direct access to the private members (or variables) of the class. This is important for efficiency. They are quite similar to the iterators in Java collections.

Method Summary

int	getCurCoordIndex() Returns the index j of the current coordinate.
int	getCurPointIndex() Returns the index i of the current point.
boolean	hasNextCoordinate() Returns true if the current point has another coordinate.
boolean	hasNextPoint() Returns true if there is a next point.
double	nextCoordinate() Returns the current coordinate ui, j and advances to the next one.
void	nextCoordinates(double[] p, int d) Returns the next d coordinates of the current point in p and advances the current coordinate index by d.
int	nextPoint(double[] p, int d) Returns the first d coordinates of the current point in p, advances to the next point, and returns the index of the new current point.
void	resetCurCoordIndex() Equivalent to setCurCoordIndex (0).
void	resetCurPointIndex() Equivalent to setCurPointIndex (0).
int	resetToNextPoint() Increases the current point index by 1 and returns its new value.
void	setCurCoordIndex(int j) Sets the current coordinate index to j, so that the next calls to nextCoordinate or nextCoordinates will return the values ui, j, ui, j+1,..., where i is the index of the current point.
void	setCurPointIndex(int i) Resets the current point index to i and the current coordinate index to zero.

Methods inherited from interface umontreal.iro.lecuyer.rng.RandomStream

nextArrayOfDouble, nextArrayOfInt, nextDouble, nextInt, resetNextSubstream, resetStartStream, resetStartSubstream, toString

Method Detail

setCurCoordIndex

void setCurCoordIndex(int j)

Sets the current coordinate index to j, so that the next calls to nextCoordinate or nextCoordinates will return the values u_{i, j}, u_{i, j+1},..., where i is the index of the current point.

Parameters:

j - index of the new current coordinate

resetCurCoordIndex

void resetCurCoordIndex()

Equivalent to setCurCoordIndex (0).

getCurCoordIndex

int getCurCoordIndex()

Returns the index j of the current coordinate. This may be useful, e.g., for testing if enough coordinates are still available.

Returns:

index of the current coordinate

hasNextCoordinate

boolean hasNextCoordinate()

Returns true if the current point has another coordinate. This can be useful for testing if coordinates are still available.

Returns:

true if the current point has another coordinate

nextCoordinate

double nextCoordinate()

Returns the current coordinate u_{i, j} and advances to the next one. If no current coordinate is available (either because the current point index has reached the number of points or because the current coordinate index has reached the number of dimensions), it throws a NoSuchElementException.

Returns:

value of the current coordinate

Throws:

NoSuchElementException - if no such coordinate is available

nextCoordinates

void nextCoordinates(double[] p,
                     int d)

Returns the next d coordinates of the current point in p and advances the current coordinate index by d. If the remaining number of coordinates is too small, a NoSuchElementException is thrown, as in nextCoordinate.

Parameters:

p - array to be filled with the coordinates, starting at index 0

d - number of coordinates to get

Throws:

NoSuchElementException - if there are not enough remaining coordinates in the current point

setCurPointIndex

void setCurPointIndex(int i)

Resets the current point index to i and the current coordinate index to zero. If i is larger or equal to the number of points, an exception will not be raised here, but only later if we ask for a new coordinate or point.

Parameters:

i - new index of the current point

resetCurPointIndex

void resetCurPointIndex()

Equivalent to setCurPointIndex (0).

resetToNextPoint

int resetToNextPoint()

Increases the current point index by 1 and returns its new value. If there is no more point, an exception will be raised only if we ask for a new coordinate or point later on.

Returns:

index of the new current point

getCurPointIndex

int getCurPointIndex()

Returns the index i of the current point. This can be useful, e.g., for caching point sets.

Returns:

index of the current point

hasNextPoint

boolean hasNextPoint()

Returns true if there is a next point. This can be useful for testing if points are still available.

Returns:

true if a next point is available from the iterated point set

nextPoint

int nextPoint(double[] p,
              int d)

Returns the first d coordinates of the current point in p, advances to the next point, and returns the index of the new current point. Even if the current coordinate index is 0, the point returned starts from coordinate 0. After obtaining the last point via this method, the current point index (returned by the method) is equal to the number of points, so it is no longer a valid point index. An exception will then be raised if we attempt to generate additional points or coordinates.

Specialized implementations of this method often allow for increased efficiency, e.g., for cycle-based point sets where the cycles (but not the points) are stored explicitly or for digital nets by allowing non-incremental point enumerations via Gray-code counters.

Parameters:

p - array to be filled with the coordinates, starting from array index 0

d - number of coordinates to return

Returns:

index of the new current point

Throws:

NoSuchElementException - if there are not enough coordinates available in the current point for filling p