Package org.apache.commons.math3.ode.nonstiff

Class GraggBulirschStoerIntegrator

java.lang.Object
org.apache.commons.math3.ode.AbstractIntegrator
org.apache.commons.math3.ode.nonstiff.AdaptiveStepsizeIntegrator
org.apache.commons.math3.ode.nonstiff.GraggBulirschStoerIntegrator
All Implemented Interfaces:
FirstOrderIntegrator, ODEIntegrator
public class GraggBulirschStoerIntegrator extends AdaptiveStepsizeIntegrator
This class implements a Gragg-Bulirsch-Stoer integrator for Ordinary Differential Equations.

The Gragg-Bulirsch-Stoer algorithm is one of the most efficient ones currently available for smooth problems. It uses Richardson extrapolation to estimate what would be the solution if the step size could be decreased down to zero.

This method changes both the step size and the order during integration, in order to minimize computation cost. It is particularly well suited when a very high precision is needed. The limit where this method becomes more efficient than high-order embedded Runge-Kutta methods like Dormand-Prince 8(5,3) depends on the problem. Results given in the Hairer, Norsett and Wanner book show for example that this limit occurs for accuracy around 1e-6 when integrating Saltzam-Lorenz equations (the authors note this problem is extremely sensitive to the errors in the first integration steps), and around 1e-11 for a two dimensional celestial mechanics problems with seven bodies (pleiades problem, involving quasi-collisions for which automatic step size control is essential).

This implementation is basically a reimplementation in Java of the odex fortran code by E. Hairer and G. Wanner. The redistribution policy for this code is available here, for convenience, it is reproduced below.

Copyright (c) 2004, Ernst Hairer
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
  • Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
  • Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Since:
1.2

  • Constructor Details

    • GraggBulirschStoerIntegrator

      public GraggBulirschStoerIntegrator(double minStep, double maxStep, double scalAbsoluteTolerance, double scalRelativeTolerance)
      Simple constructor. Build a Gragg-Bulirsch-Stoer integrator with the given step bounds. All tuning parameters are set to their default values. The default step handler does nothing.
      Parameters:
      minStep - minimal step (sign is irrelevant, regardless of integration direction, forward or backward), the last step can be smaller than this
      maxStep - maximal step (sign is irrelevant, regardless of integration direction, forward or backward), the last step can be smaller than this
      scalAbsoluteTolerance - allowed absolute error
      scalRelativeTolerance - allowed relative error

    • GraggBulirschStoerIntegrator

      public GraggBulirschStoerIntegrator(double minStep, double maxStep, double[] vecAbsoluteTolerance, double[] vecRelativeTolerance)
      Simple constructor. Build a Gragg-Bulirsch-Stoer integrator with the given step bounds. All tuning parameters are set to their default values. The default step handler does nothing.
      Parameters:
      minStep - minimal step (must be positive even for backward integration), the last step can be smaller than this
      maxStep - maximal step (must be positive even for backward integration)
      vecAbsoluteTolerance - allowed absolute error
      vecRelativeTolerance - allowed relative error

  • Method Details

    • setStabilityCheck

      public void setStabilityCheck(boolean performStabilityCheck, int maxNumIter, int maxNumChecks, double stepsizeReductionFactor)
      Set the stability check controls.

      The stability check is performed on the first few iterations of the extrapolation scheme. If this test fails, the step is rejected and the stepsize is reduced.

      By default, the test is performed, at most during two iterations at each step, and at most once for each of these iterations. The default stepsize reduction factor is 0.5.

      Parameters:
      performStabilityCheck - if true, stability check will be performed, if false, the check will be skipped
      maxNumIter - maximal number of iterations for which checks are performed (the number of iterations is reset to default if negative or null)
      maxNumChecks - maximal number of checks for each iteration (the number of checks is reset to default if negative or null)
      stepsizeReductionFactor - stepsize reduction factor in case of failure (the factor is reset to default if lower than 0.0001 or greater than 0.9999)

    • setControlFactors

      public void setControlFactors(double control1, double control2, double control3, double control4)
      Set the step size control factors.

      The new step size hNew is computed from the old one h by: 

       hNew = h * stepControl2 / (err/stepControl1)^(1/(2k+1))
      where err is the scaled error and k the iteration number of the extrapolation scheme (counting from 0). The default values are 0.65 for stepControl1 and 0.94 for stepControl2.

      The step size is subject to the restriction: 

       stepControl3^(1/(2k+1))/stepControl4 invalid input: '<'= hNew/h invalid input: '<'= 1/stepControl3^(1/(2k+1))
      The default values are 0.02 for stepControl3 and 4.0 for stepControl4.

      Parameters:
      control1 - first stepsize control factor (the factor is reset to default if lower than 0.0001 or greater than 0.9999)
      control2 - second stepsize control factor (the factor is reset to default if lower than 0.0001 or greater than 0.9999)
      control3 - third stepsize control factor (the factor is reset to default if lower than 0.0001 or greater than 0.9999)
      control4 - fourth stepsize control factor (the factor is reset to default if lower than 1.0001 or greater than 999.9)

    • setOrderControl

      public void setOrderControl(int maximalOrder, double control1, double control2)
      Set the order control parameters.

      The Gragg-Bulirsch-Stoer method changes both the step size and the order during integration, in order to minimize computation cost. Each extrapolation step increases the order by 2, so the maximal order that will be used is always even, it is twice the maximal number of columns in the extrapolation table.

       order is decreased if w(k-1) invalid input: '<'= w(k)   * orderControl1
 order is increased if w(k)   invalid input: '<'= w(k-1) * orderControl2

      where w is the table of work per unit step for each order (number of function calls divided by the step length), and k is the current order.

      The default maximal order after construction is 18 (i.e. the maximal number of columns is 9). The default values are 0.8 for orderControl1 and 0.9 for orderControl2.

      Parameters:
      maximalOrder - maximal order in the extrapolation table (the maximal order is reset to default if order invalid input: '<'= 6 or odd)
      control1 - first order control factor (the factor is reset to default if lower than 0.0001 or greater than 0.9999)
      control2 - second order control factor (the factor is reset to default if lower than 0.0001 or greater than 0.9999)

    • addStepHandler

      public void addStepHandler(StepHandler handler)
      Add a step handler to this integrator.

      The handler will be called by the integrator for each accepted step.

      Specified by:
      addStepHandler in interface ODEIntegrator
      Overrides:
      addStepHandler in class AbstractIntegrator
      Parameters:
      handler - handler for the accepted steps
      See Also:

    • addEventHandler

      public void addEventHandler(EventHandler function, double maxCheckInterval, double convergence, int maxIterationCount, UnivariateSolver solver)
      Add an event handler to the integrator.
      Specified by:
      addEventHandler in interface ODEIntegrator
      Overrides:
      addEventHandler in class AbstractIntegrator
      Parameters:
      function - event handler
      maxCheckInterval - maximal time interval between switching function checks (this interval prevents missing sign changes in case the integration steps becomes very large)
      convergence - convergence threshold in the event time search
      maxIterationCount - upper limit of the iteration count in the event time search
      solver - The root-finding algorithm to use to detect the state events.
      See Also:

    • setInterpolationControl

      public void setInterpolationControl(boolean useInterpolationErrorForControl, int mudifControlParameter)
      Set the interpolation order control parameter. The interpolation order for dense output is 2k - mudif + 1. The default value for mudif is 4 and the interpolation error is used in stepsize control by default.
      Parameters:
      useInterpolationErrorForControl - if true, interpolation error is used for stepsize control
      mudifControlParameter - interpolation order control parameter (the parameter is reset to default if invalid input: '<'= 0 or >= 7)

    • integrate

      public void integrate(ExpandableStatefulODE equations, double t) throws NumberIsTooSmallException, DimensionMismatchException, MaxCountExceededException, NoBracketingException
      Integrate a set of differential equations up to the given time.

      This method solves an Initial Value Problem (IVP).

      The set of differential equations is composed of a main set, which can be extended by some sets of secondary equations. The set of equations must be already set up with initial time and partial states. At integration completion, the final time and partial states will be available in the same object.

      Since this method stores some internal state variables made available in its public interface during integration (AbstractIntegrator.getCurrentSignedStepsize()), it is not thread-safe.

      Specified by:
      integrate in class AdaptiveStepsizeIntegrator
      Parameters:
      equations - complete set of differential equations to integrate
      t - target time for the integration (can be set to a value smaller than t0 for backward integration)
      Throws:
      NumberIsTooSmallException - if integration step is too small
      DimensionMismatchException - if the dimension of the complete state does not match the complete equations sets dimension
      MaxCountExceededException - if the number of functions evaluations is exceeded
      NoBracketingException - if the location of an event cannot be bracketed