Package org.orekit.utils

Class TimeStampedFieldAngularCoordinatesHermiteInterpolator<KK extends CalculusFieldElement<KK>>

    • Constructor Detail

      • TimeStampedFieldAngularCoordinatesHermiteInterpolator

        public TimeStampedFieldAngularCoordinatesHermiteInterpolator()
        Constructor with :
        • Default number of interpolation points of DEFAULT_INTERPOLATION_POINTS
        • Default extrapolation threshold value (DEFAULT_EXTRAPOLATION_THRESHOLD_SEC s)
        • Use of angular and first time derivative for attitude interpolation
        As this implementation of interpolation is polynomial, it should be used only with small number of interpolation points (about 10-20 points) in order to avoid Runge's phenomenon and numerical problems (including NaN appearing).

      • TimeStampedFieldAngularCoordinatesHermiteInterpolator

        public TimeStampedFieldAngularCoordinatesHermiteInterpolator​(int interpolationPoints)
        /** Constructor with :
        • Default extrapolation threshold value (DEFAULT_EXTRAPOLATION_THRESHOLD_SEC s)
        • Use of angular and first time derivative for attitude interpolation
        As this implementation of interpolation is polynomial, it should be used only with small number of interpolation points (about 10-20 points) in order to avoid Runge's phenomenon and numerical problems (including NaN appearing).
        Parameters:
        interpolationPoints - number of interpolation points

      • TimeStampedFieldAngularCoordinatesHermiteInterpolator

        public TimeStampedFieldAngularCoordinatesHermiteInterpolator​(int interpolationPoints,
                                                             AngularDerivativesFilter filter)
        Constructor with :
        • Default extrapolation threshold value (DEFAULT_EXTRAPOLATION_THRESHOLD_SEC s)
        As this implementation of interpolation is polynomial, it should be used only with small number of interpolation points (about 10-20 points) in order to avoid Runge's phenomenon and numerical problems (including NaN appearing).
        Parameters:
        interpolationPoints - number of interpolation points
        filter - filter for derivatives from the sample to use in interpolation

      • TimeStampedFieldAngularCoordinatesHermiteInterpolator

        public TimeStampedFieldAngularCoordinatesHermiteInterpolator​(int interpolationPoints,
                                                             double extrapolationThreshold,
                                                             AngularDerivativesFilter filter)
        Constructor.

        As this implementation of interpolation is polynomial, it should be used only with small number of interpolation points (about 10-20 points) in order to avoid Runge's phenomenon and numerical problems (including NaN appearing).

        Parameters:
        interpolationPoints - number of interpolation points
        extrapolationThreshold - extrapolation threshold beyond which the propagation will fail
        filter - filter for derivatives from the sample to use in interpolation

    • Method Detail

      • getFilter

        public AngularDerivativesFilter getFilter()
        Get filter for derivatives from the sample to use in interpolation.
        Returns:
        filter for derivatives from the sample to use in interpolation

      • interpolate

        protected TimeStampedFieldAngularCoordinates<KK> interpolate​(AbstractFieldTimeInterpolator.InterpolationData interpolationData)
        Interpolate instance from given interpolation data.

        The interpolated instance is created by polynomial Hermite interpolation on Rodrigues vector ensuring rotation rate remains the exact derivative of rotation.

        This method is based on Sergei Tanygin's paper Attitude Interpolation, changing the norm of the vector to match the modified Rodrigues vector as described in Malcolm D. Shuster's paper A Survey of Attitude Representations. This change avoids the singularity at π. There is still a singularity at 2π, which is handled by slightly offsetting all rotations when this singularity is detected. Another change is that the mean linear motion is first removed before interpolation and added back after interpolation. This allows to use interpolation even when the sample covers much more than one turn and even when sample points are separated by more than one turn.

        Note that even if first and second time derivatives (rotation rates and acceleration) from sample can be ignored, the interpolated instance always includes interpolated derivatives. This feature can be used explicitly to compute these derivatives when it would be too complex to compute them from an analytical formula: just compute a few sample points from the explicit formula and set the derivatives to zero in these sample points, then use interpolation to add derivatives consistent with the rotations.

        Specified by:
        interpolate in class AbstractFieldTimeInterpolator<TimeStampedFieldAngularCoordinates<KK extends CalculusFieldElement<KK>>,​KK extends CalculusFieldElement<KK>>
        Parameters:
        interpolationData - interpolation data
        Returns:
        interpolated instance from given interpolation data.