View Javadoc
1   /*
2    * Licensed to the Apache Software Foundation (ASF) under one or more
3    * contributor license agreements.  See the NOTICE file distributed with
4    * this work for additional information regarding copyright ownership.
5    * The ASF licenses this file to You under the Apache License, Version 2.0
6    * (the "License"); you may not use this file except in compliance with
7    * the License.  You may obtain a copy of the License at
8    *
9    *      https://www.apache.org/licenses/LICENSE-2.0
10   *
11   * Unless required by applicable law or agreed to in writing, software
12   * distributed under the License is distributed on an "AS IS" BASIS,
13   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14   * See the License for the specific language governing permissions and
15   * limitations under the License.
16   */
17  
18  /*
19   * This is not the original file distributed by the Apache Software Foundation
20   * It has been modified by the Hipparchus project
21   */
22  
23  package org.hipparchus.ode;
24  
25  import org.hipparchus.exception.MathIllegalArgumentException;
26  import org.hipparchus.exception.MathIllegalStateException;
27  import org.hipparchus.linear.Array2DRowRealMatrix;
28  import org.hipparchus.ode.nonstiff.AdaptiveStepsizeIntegrator;
29  import org.hipparchus.ode.nonstiff.DormandPrince853Integrator;
30  import org.hipparchus.ode.sampling.ODEStateInterpolator;
31  import org.hipparchus.ode.sampling.ODEStepHandler;
32  import org.hipparchus.util.FastMath;
33  
34  /**
35   * This class is the base class for multistep integrators for Ordinary
36   * Differential Equations.
37   * <p>We define scaled derivatives s<sub>i</sub>(n) at step n as:
38   * \[
39   *   \left\{\begin{align}
40   *   s_1(n) &amp;= h y'_n \text{ for first derivative}\\
41   *   s_2(n) &amp;= \frac{h^2}{2} y_n'' \text{ for second derivative}\\
42   *   s_3(n) &amp;= \frac{h^3}{6} y_n''' \text{ for third derivative}\\
43   *   &amp;\cdots\\
44   *   s_k(n) &amp;= \frac{h^k}{k!} y_n^{(k)} \text{ for } k^\mathrm{th} \text{ derivative}
45   *   \end{align}\right.
46   * \]</p>
47   * <p>Rather than storing several previous steps separately, this implementation uses
48   * the Nordsieck vector with higher degrees scaled derivatives all taken at the same
49   * step (y<sub>n</sub>, s<sub>1</sub>(n) and r<sub>n</sub>) where r<sub>n</sub> is defined as:
50   * \[
51   * r_n = [ s_2(n), s_3(n) \ldots s_k(n) ]^T
52   * \]
53   * (we omit the k index in the notation for clarity)</p>
54   * <p>
55   * Multistep integrators with Nordsieck representation are highly sensitive to
56   * large step changes because when the step is multiplied by factor a, the
57   * k<sup>th</sup> component of the Nordsieck vector is multiplied by a<sup>k</sup>
58   * and the last components are the least accurate ones. The default max growth
59   * factor is therefore set to a quite low value: 2<sup>1/order</sup>.
60   * </p>
61   *
62   * @see org.hipparchus.ode.nonstiff.AdamsBashforthIntegrator
63   * @see org.hipparchus.ode.nonstiff.AdamsMoultonIntegrator
64   */
65  public abstract class MultistepIntegrator extends AdaptiveStepsizeIntegrator {
66  
67      /** First scaled derivative (h y'). */
68      protected double[] scaled;
69  
70      /** Nordsieck matrix of the higher scaled derivatives.
71       * <p>(h<sup>2</sup>/2 y'', h<sup>3</sup>/6 y''' ..., h<sup>k</sup>/k! y<sup>(k)</sup>)</p>
72       */
73      protected Array2DRowRealMatrix nordsieck;
74  
75      /** Starter integrator. */
76      private ODEIntegrator starter;
77  
78      /** Number of steps of the multistep method (excluding the one being computed). */
79      private final int nSteps;
80  
81      /** Stepsize control exponent. */
82      private double exp;
83  
84      /** Safety factor for stepsize control. */
85      private double safety;
86  
87      /** Minimal reduction factor for stepsize control. */
88      private double minReduction;
89  
90      /** Maximal growth factor for stepsize control. */
91      private double maxGrowth;
92  
93      /**
94       * Build a multistep integrator with the given stepsize bounds.
95       * <p>The default starter integrator is set to the {@link
96       * DormandPrince853Integrator Dormand-Prince 8(5,3)} integrator with
97       * some defaults settings.</p>
98       * <p>
99       * The default max growth factor is set to a quite low value: 2<sup>1/order</sup>.
100      * </p>
101      * @param name name of the method
102      * @param nSteps number of steps of the multistep method
103      * (excluding the one being computed)
104      * @param order order of the method
105      * @param minStep minimal step (must be positive even for backward
106      * integration), the last step can be smaller than this
107      * @param maxStep maximal step (must be positive even for backward
108      * integration)
109      * @param scalAbsoluteTolerance allowed absolute error
110      * @param scalRelativeTolerance allowed relative error
111      * @exception MathIllegalArgumentException if number of steps is smaller than 2
112      */
113     protected MultistepIntegrator(final String name, final int nSteps,
114                                   final int order,
115                                   final double minStep, final double maxStep,
116                                   final double scalAbsoluteTolerance,
117                                   final double scalRelativeTolerance)
118         throws MathIllegalArgumentException {
119 
120         super(name, minStep, maxStep, scalAbsoluteTolerance, scalRelativeTolerance);
121 
122         if (nSteps < 2) {
123             throw new MathIllegalArgumentException(LocalizedODEFormats.INTEGRATION_METHOD_NEEDS_AT_LEAST_TWO_PREVIOUS_POINTS,
124                                                    nSteps, 2, true);
125         }
126 
127         starter = new DormandPrince853Integrator(minStep, maxStep,
128                                                  scalAbsoluteTolerance,
129                                                  scalRelativeTolerance);
130         this.nSteps = nSteps;
131 
132         exp = -1.0 / order;
133 
134         // set the default values of the algorithm control parameters
135         setSafety(0.9);
136         setMinReduction(0.2);
137         setMaxGrowth(FastMath.pow(2.0, -exp));
138 
139     }
140 
141     /**
142      * Build a multistep integrator with the given stepsize bounds.
143      * <p>The default starter integrator is set to the {@link
144      * DormandPrince853Integrator Dormand-Prince 8(5,3)} integrator with
145      * some defaults settings.</p>
146      * <p>
147      * The default max growth factor is set to a quite low value: 2<sup>1/order</sup>.
148      * </p>
149      * @param name name of the method
150      * @param nSteps number of steps of the multistep method
151      * (excluding the one being computed)
152      * @param order order of the method
153      * @param minStep minimal step (must be positive even for backward
154      * integration), the last step can be smaller than this
155      * @param maxStep maximal step (must be positive even for backward
156      * integration)
157      * @param vecAbsoluteTolerance allowed absolute error
158      * @param vecRelativeTolerance allowed relative error
159      */
160     protected MultistepIntegrator(final String name, final int nSteps,
161                                   final int order,
162                                   final double minStep, final double maxStep,
163                                   final double[] vecAbsoluteTolerance,
164                                   final double[] vecRelativeTolerance) {
165         super(name, minStep, maxStep, vecAbsoluteTolerance, vecRelativeTolerance);
166 
167         if (nSteps < 2) {
168             throw new MathIllegalArgumentException(LocalizedODEFormats.INTEGRATION_METHOD_NEEDS_AT_LEAST_TWO_PREVIOUS_POINTS,
169                                                    nSteps, 2, true);
170         }
171 
172         starter = new DormandPrince853Integrator(minStep, maxStep,
173                                                  vecAbsoluteTolerance,
174                                                  vecRelativeTolerance);
175         this.nSteps = nSteps;
176 
177         exp = -1.0 / order;
178 
179         // set the default values of the algorithm control parameters
180         setSafety(0.9);
181         setMinReduction(0.2);
182         setMaxGrowth(FastMath.pow(2.0, -exp));
183 
184     }
185 
186     /**
187      * Get the starter integrator.
188      * @return starter integrator
189      */
190     public ODEIntegrator getStarterIntegrator() {
191         return starter;
192     }
193 
194     /**
195      * Set the starter integrator.
196      * <p>The various step and event handlers for this starter integrator
197      * will be managed automatically by the multi-step integrator. Any
198      * user configuration for these elements will be cleared before use.</p>
199      * @param starterIntegrator starter integrator
200      */
201     public void setStarterIntegrator(ODEIntegrator starterIntegrator) {
202         this.starter = starterIntegrator;
203     }
204 
205     /** Start the integration.
206      * <p>This method computes one step using the underlying starter integrator,
207      * and initializes the Nordsieck vector at step start. The starter integrator
208      * purpose is only to establish initial conditions, it does not really change
209      * time by itself. The top level multistep integrator remains in charge of
210      * handling time propagation and events handling as it will starts its own
211      * computation right from the beginning. In a sense, the starter integrator
212      * can be seen as a dummy one and so it will never trigger any user event nor
213      * call any user step handler.</p>
214      * @param equations complete set of differential equations to integrate
215      * @param initialState initial state (time, primary and secondary state vectors)
216      * @param finalTime target time for the integration
217      * (can be set to a value smaller than {@code initialState.getTime()} for backward integration)
218      * @exception MathIllegalArgumentException if arrays dimension do not match equations settings
219      * @exception MathIllegalArgumentException if integration step is too small
220      * @exception MathIllegalStateException if the number of functions evaluations is exceeded
221      * @exception MathIllegalArgumentException if the location of an event cannot be bracketed
222      */
223     protected void start(final ExpandableODE equations, final ODEState initialState, final double finalTime)
224         throws MathIllegalArgumentException, MathIllegalStateException {
225 
226         // make sure NO user events nor user step handlers are triggered,
227         // this is the task of the top level integrator, not the task of the starter integrator
228         starter.clearEventDetectors();
229         starter.clearStepHandlers();
230 
231         // set up one specific step handler to extract initial Nordsieck vector
232         starter.addStepHandler(new NordsieckInitializer((nSteps + 3) / 2));
233 
234         // start integration, expecting a InitializationCompletedMarkerException
235         try {
236 
237             starter.integrate(getEquations(), initialState, finalTime);
238 
239             // we should not reach this step
240             throw new MathIllegalStateException(LocalizedODEFormats.MULTISTEP_STARTER_STOPPED_EARLY);
241 
242         } catch (InitializationCompletedMarkerException icme) { // NOPMD
243             // this is the expected nominal interruption of the start integrator
244 
245             // count the evaluations used by the starter
246             getEvaluationsCounter().increment(starter.getEvaluations());
247 
248         }
249 
250         // remove the specific step handler
251         starter.clearStepHandlers();
252 
253     }
254 
255     /** Initialize the high order scaled derivatives at step start.
256      * @param h step size to use for scaling
257      * @param t first steps times
258      * @param y first steps states
259      * @param yDot first steps derivatives
260      * @return Nordieck vector at first step (h<sup>2</sup>/2 y''<sub>n</sub>,
261      * h<sup>3</sup>/6 y'''<sub>n</sub> ... h<sup>k</sup>/k! y<sup>(k)</sup><sub>n</sub>)
262      */
263     protected abstract Array2DRowRealMatrix initializeHighOrderDerivatives(double h, double[] t, double[][] y, double[][] yDot);
264 
265     /** Get the minimal reduction factor for stepsize control.
266      * @return minimal reduction factor
267      */
268     public double getMinReduction() {
269         return minReduction;
270     }
271 
272     /** Set the minimal reduction factor for stepsize control.
273      * @param minReduction minimal reduction factor
274      */
275     public void setMinReduction(final double minReduction) {
276         this.minReduction = minReduction;
277     }
278 
279     /** Get the maximal growth factor for stepsize control.
280      * @return maximal growth factor
281      */
282     public double getMaxGrowth() {
283         return maxGrowth;
284     }
285 
286     /** Set the maximal growth factor for stepsize control.
287      * @param maxGrowth maximal growth factor
288      */
289     public void setMaxGrowth(final double maxGrowth) {
290         this.maxGrowth = maxGrowth;
291     }
292 
293     /** Get the safety factor for stepsize control.
294      * @return safety factor
295      */
296     public double getSafety() {
297       return safety;
298     }
299 
300     /** Set the safety factor for stepsize control.
301      * @param safety safety factor
302      */
303     public void setSafety(final double safety) {
304       this.safety = safety;
305     }
306 
307     /** Get the number of steps of the multistep method (excluding the one being computed).
308      * @return number of steps of the multistep method (excluding the one being computed)
309      */
310     public int getNSteps() {
311       return nSteps;
312     }
313 
314     /** Rescale the instance.
315      * <p>Since the scaled and Nordsieck arrays are shared with the caller,
316      * this method has the side effect of rescaling this arrays in the caller too.</p>
317      * @param newStepSize new step size to use in the scaled and Nordsieck arrays
318      */
319     protected void rescale(final double newStepSize) {
320 
321         final double ratio = newStepSize / getStepSize();
322         for (int i = 0; i < scaled.length; ++i) {
323             scaled[i] = scaled[i] * ratio;
324         }
325 
326         final double[][] nData = nordsieck.getDataRef();
327         double power = ratio;
328         for (int i = 0; i < nData.length; ++i) {
329             power = power * ratio;
330             final double[] nDataI = nData[i];
331             for (int j = 0; j < nDataI.length; ++j) {
332                 nDataI[j] = nDataI[j] * power;
333             }
334         }
335 
336         setStepSize(newStepSize);
337 
338     }
339 
340     /** Compute step grow/shrink factor according to normalized error.
341      * @param error normalized error of the current step
342      * @return grow/shrink factor for next step
343      */
344     protected double computeStepGrowShrinkFactor(final double error) {
345         return FastMath.min(maxGrowth, FastMath.max(minReduction, safety * FastMath.pow(error, exp)));
346     }
347 
348     /** Specialized step handler storing the first step. */
349     private class NordsieckInitializer implements ODEStepHandler {
350 
351         /** Steps counter. */
352         private int count;
353 
354         /** Start of the integration. */
355         private ODEStateAndDerivative savedStart;
356 
357         /** First steps times. */
358         private final double[] t;
359 
360         /** First steps states. */
361         private final double[][] y;
362 
363         /** First steps derivatives. */
364         private final double[][] yDot;
365 
366         /** Simple constructor.
367          * @param nbStartPoints number of start points (including the initial point)
368          */
369         NordsieckInitializer(final int nbStartPoints) {
370             this.count  = 0;
371             this.t      = new double[nbStartPoints];
372             this.y      = new double[nbStartPoints][];
373             this.yDot   = new double[nbStartPoints][];
374         }
375 
376         /** {@inheritDoc} */
377         @Override
378         public void handleStep(ODEStateInterpolator interpolator) {
379 
380             if (count == 0) {
381                 // first step, we need to store also the point at the beginning of the step
382                 savedStart   = interpolator.getPreviousState();
383                 t[0]    = savedStart.getTime();
384                 y[0]    = savedStart.getCompleteState();
385                 yDot[0] = savedStart.getCompleteDerivative();
386             }
387 
388             // store the point at the end of the step
389             ++count;
390             final ODEStateAndDerivative curr = interpolator.getCurrentState();
391             t[count]    = curr.getTime();
392             y[count]    = curr.getCompleteState();
393             yDot[count] = curr.getCompleteDerivative();
394 
395             if (count == t.length - 1) {
396 
397                 // this was the last point we needed, we can compute the derivatives
398                 setStepStart(savedStart);
399                 final double rawStep = (t[t.length - 1] - t[0]) / (t.length - 1);
400                 setStepSize(getStepSizeHelper().filterStep(rawStep, rawStep >= 0, true));
401 
402                 // first scaled derivative
403                 scaled = yDot[0].clone();
404                 for (int j = 0; j < scaled.length; ++j) {
405                     scaled[j] *= getStepSize();
406                 }
407 
408                 // higher order derivatives
409                 nordsieck = initializeHighOrderDerivatives(getStepSize(), t, y, yDot);
410 
411                 // stop the integrator now that all needed steps have been handled
412                 throw new InitializationCompletedMarkerException();
413 
414             }
415 
416         }
417 
418     }
419 
420     /** Marker exception used ONLY to stop the starter integrator after first step. */
421     private static class InitializationCompletedMarkerException
422         extends RuntimeException {
423 
424         /** Serializable version identifier. */
425         private static final long serialVersionUID = -1914085471038046418L;
426 
427         /** Simple constructor. */
428         InitializationCompletedMarkerException() {
429             super((Throwable) null);
430         }
431 
432     }
433 
434 }
435