1 | <!-- ******************************************************** --> |
---|
2 | <!-- --> |
---|
3 | <!-- [History] --> |
---|
4 | <!-- New section on Reverse MC: L. Desorgher, Dec-2009 --> |
---|
5 | <!-- Converted to DocBook: Katsuya Amako, Aug-2006 --> |
---|
6 | <!-- Changed by: Dennis Wright, 25-Jun-2002 --> |
---|
7 | <!-- --> |
---|
8 | <!-- ******************************************************** --> |
---|
9 | |
---|
10 | |
---|
11 | <!-- ******************* Section (Level#1) ****************** --> |
---|
12 | <sect1 id="sect.EvtBias"> |
---|
13 | <title> |
---|
14 | Event Biasing Techniques |
---|
15 | </title> |
---|
16 | |
---|
17 | <!-- ******************* Section (Level#2) ****************** --> |
---|
18 | <sect2 id="sect.EvtBias.ScorImpRoul"> |
---|
19 | <title> |
---|
20 | Scoring, Geometrical Importance Sampling and Weight Roulette |
---|
21 | </title> |
---|
22 | |
---|
23 | <para> |
---|
24 | Geant4 provides event biasing techniques which may be used to save |
---|
25 | computing time in such applications as the simulation of radiation |
---|
26 | shielding. These are <emphasis>geometrical splitting |
---|
27 | </emphasis> and <emphasis>Russian roulette</emphasis> |
---|
28 | (also called geometrical importance sampling), and |
---|
29 | <emphasis>weight roulette</emphasis>. Scoring is carried out by <emphasis>G4MultiFunctionalDetector</emphasis> (see <xref |
---|
30 | linkend="sect.Hits.G4Multi" /> and <xref linkend="sect.Hits.G4VPrim" />) using the standard Geant4 scoring technique. |
---|
31 | Biasing specific scorers have been implemented and are described within |
---|
32 | <emphasis>G4MultiFunctionDetector</emphasis> documentation. In this chapter, it is assumed that |
---|
33 | the reader is familiar with both the usage of Geant4 and the |
---|
34 | concepts of importance sampling. More detailed documentation may be |
---|
35 | found in the documents |
---|
36 | <ulink url="http://geant4.cern.ch/collaboration/working_groups/geometry/biasing/Sampling.html"> |
---|
37 | 'Scoring, geometrical importance sampling and weight roulette' |
---|
38 | </ulink>. |
---|
39 | A detailed description of different use-cases which employ the sampling |
---|
40 | and scoring techniques can be found in the document |
---|
41 | <ulink url="http://geant4.cern.ch/collaboration/working_groups/geometry/biasing/BiasScoreUseCases.html"> |
---|
42 | 'Use cases of importance sampling and scoring in Geant4' |
---|
43 | </ulink>. |
---|
44 | </para> |
---|
45 | |
---|
46 | <para> |
---|
47 | The purpose of importance sampling is to save computing time by |
---|
48 | sampling less often the particle histories entering "less |
---|
49 | important" geometry regions, and more often in more "important" |
---|
50 | regions. Given the same amount of computing time, an |
---|
51 | importance-sampled and an analogue-sampled simulation must show |
---|
52 | equal mean values, while the importance-sampled simulation will |
---|
53 | have a decreased variance. |
---|
54 | </para> |
---|
55 | |
---|
56 | <para> |
---|
57 | The implementation of scoring is independent of the implementation |
---|
58 | of importance sampling. However both share common concepts. |
---|
59 | <emphasis>Scoring and importance sampling apply to particle types chosen |
---|
60 | by the user</emphasis>, which should be borne in mind when interpreting the |
---|
61 | output of any biased simulation. |
---|
62 | </para> |
---|
63 | |
---|
64 | <para> |
---|
65 | Examples on how to use scoring and importance sampling may be found |
---|
66 | in <literal>examples/extended/biasing</literal>. |
---|
67 | </para> |
---|
68 | |
---|
69 | |
---|
70 | <!-- ******************* Section (Level#3) ****************** --> |
---|
71 | <sect3 id="sect.EvtBias.ScorImpRoul.Geom"> |
---|
72 | <title> |
---|
73 | Geometries |
---|
74 | </title> |
---|
75 | |
---|
76 | <para> |
---|
77 | The kind of scoring referred to in this note and the importance |
---|
78 | sampling apply to spatial cells provided by the user. |
---|
79 | </para> |
---|
80 | |
---|
81 | <para> |
---|
82 | A <emphasis role="bold">cell</emphasis> is a physical volume (further specified |
---|
83 | by it's replica number, if the volume is a replica). Cells may be defined |
---|
84 | in two kinds of geometries: |
---|
85 | |
---|
86 | <orderedlist spacing="compact"> |
---|
87 | <listitem><para> |
---|
88 | <emphasis role="bold">mass geometry</emphasis>: the geometry setup of the |
---|
89 | experiment to be simulated. Physics processes apply to this geometry. |
---|
90 | </para></listitem> |
---|
91 | <listitem><para> |
---|
92 | <emphasis role="bold">parallel-geometry</emphasis>: a geometry constructed |
---|
93 | to define the physical volumes according to which scoring and/or importance |
---|
94 | sampling is applied. |
---|
95 | </para></listitem> |
---|
96 | </orderedlist> |
---|
97 | </para> |
---|
98 | |
---|
99 | <para> |
---|
100 | The user has the choice to score and/or sample by importance the |
---|
101 | particles of the chosen type, according to mass geometry or to |
---|
102 | parallel geometry. It is possible to utilize several parallel |
---|
103 | geometries in addition to the mass geometry. This provides the user |
---|
104 | with a lot of flexibility to define separate geometries for |
---|
105 | different particle types in order to apply scoring or/and |
---|
106 | importance sampling. |
---|
107 | </para> |
---|
108 | |
---|
109 | <para> |
---|
110 | <note> |
---|
111 | Parallel geometries should be constructed using the implementation as |
---|
112 | described in <xref linkend="sect.ParaGeom"/>. |
---|
113 | |
---|
114 | There are a few conditions for parallel geometries: |
---|
115 | |
---|
116 | <itemizedlist spacing="compact"> |
---|
117 | <listitem><para> |
---|
118 | The world volume for parallel and mass geometries must be identical copies. |
---|
119 | </para></listitem> |
---|
120 | <listitem><para> |
---|
121 | Scoring and importance cells must not share boundaries with the world volume. |
---|
122 | </para></listitem> |
---|
123 | </itemizedlist> |
---|
124 | </note> |
---|
125 | </para> |
---|
126 | |
---|
127 | </sect3> |
---|
128 | |
---|
129 | <!-- ******************* Section (Level#3) ****************** --> |
---|
130 | <sect3 id="sect.EvtBias.ScorImpRoul.ChgSamp"> |
---|
131 | <title> |
---|
132 | Changing the Sampling |
---|
133 | </title> |
---|
134 | |
---|
135 | <para> |
---|
136 | Samplers are higher level tools which perform the necessary |
---|
137 | changes of the Geant4 sampling in order to apply importance |
---|
138 | sampling and weight roulette. |
---|
139 | </para> |
---|
140 | |
---|
141 | <para> |
---|
142 | Variance reduction (and scoring through the |
---|
143 | <emphasis>G4MultiFunctionalDetector</emphasis>) |
---|
144 | may be combined arbitrarily for chosen particle types and may be applied to the |
---|
145 | mass or to parallel geometries. |
---|
146 | </para> |
---|
147 | |
---|
148 | <para> |
---|
149 | The <literal>G4GeometrySampler</literal> can be applied equally to mass or |
---|
150 | parallel geometries with an abstract interface supplied by <literal>G4VSampler</literal>. |
---|
151 | <literal>G4VSampler</literal> provides |
---|
152 | <literal>Prepare...</literal> methods and a <literal>Configure</literal> |
---|
153 | method: |
---|
154 | |
---|
155 | <anchor id="anchor_EvtBias_G4VSampler" /> |
---|
156 | <informalexample> |
---|
157 | <programlisting> |
---|
158 | class G4VSampler |
---|
159 | { |
---|
160 | public: |
---|
161 | G4VSampler(); |
---|
162 | virtual ~G4VSampler(); |
---|
163 | virtual void PrepareImportanceSampling(G4VIStore *istore, |
---|
164 | const G4VImportanceAlgorithm |
---|
165 | *ialg = 0) = 0; |
---|
166 | virtual void PrepareWeightRoulett(G4double wsurvive = 0.5, |
---|
167 | G4double wlimit = 0.25, |
---|
168 | G4double isource = 1) = 0; |
---|
169 | virtual void PrepareWeightWindow(G4VWeightWindowStore *wwstore, |
---|
170 | G4VWeightWindowAlgorithm *wwAlg = 0, |
---|
171 | G4PlaceOfAction placeOfAction = |
---|
172 | onBoundary) = 0; |
---|
173 | virtual void Configure() = 0; |
---|
174 | virtual void ClearSampling() = 0; |
---|
175 | virtual G4bool IsConfigured() const = 0; |
---|
176 | }; |
---|
177 | </programlisting> |
---|
178 | </informalexample> |
---|
179 | </para> |
---|
180 | |
---|
181 | <para> |
---|
182 | The methods for setting up the desired combination need specific |
---|
183 | information: |
---|
184 | |
---|
185 | <itemizedlist spacing="compact"> |
---|
186 | <listitem><para> |
---|
187 | Importance sampling: message <literal>PrepareImportanceSampling</literal> |
---|
188 | with a |
---|
189 | <link linkend="anchor_EvtBias_G4VIStore"> |
---|
190 | <literal>G4VIStore</literal></link> |
---|
191 | and optionally a <literal>G4VImportanceAlgorithm</literal> |
---|
192 | </para></listitem> |
---|
193 | <listitem><para> |
---|
194 | Weight window: message <literal>PrepareWeightWindow</literal> with the |
---|
195 | arguments: |
---|
196 | <itemizedlist spacing="compact"> |
---|
197 | <listitem><para> |
---|
198 | <emphasis>*wwstore</emphasis>: a |
---|
199 | <literal>G4VWeightWindowStore</literal> for retrieving the lower |
---|
200 | weight bounds for the energy-space cells |
---|
201 | </para></listitem> |
---|
202 | <listitem><para> |
---|
203 | <emphasis>*wwAlg</emphasis>: a |
---|
204 | <literal>G4VWeightWindowAlgorithm</literal> if a customized algorithm |
---|
205 | should be used |
---|
206 | </para></listitem> |
---|
207 | <listitem><para> |
---|
208 | <emphasis>placeOfAction</emphasis>: a |
---|
209 | <literal>G4PlaceOfAction</literal> specifying where to perform the |
---|
210 | biasing |
---|
211 | </para></listitem> |
---|
212 | </itemizedlist> |
---|
213 | </para></listitem> |
---|
214 | <listitem><para> |
---|
215 | Weight roulette: message <literal>PrepareWeightRoulett</literal> with the |
---|
216 | optional parameters: |
---|
217 | <itemizedlist spacing="compact"> |
---|
218 | <listitem><para> |
---|
219 | <emphasis>wsurvive</emphasis>: survival weight |
---|
220 | </para></listitem> |
---|
221 | <listitem><para> |
---|
222 | <emphasis>wlimit</emphasis>: minimal allowed |
---|
223 | value of weight * source importance / cell importance |
---|
224 | </para></listitem> |
---|
225 | <listitem><para> |
---|
226 | <emphasis>isource</emphasis>: importance of the source cell |
---|
227 | </para></listitem> |
---|
228 | </itemizedlist> |
---|
229 | </para></listitem> |
---|
230 | </itemizedlist> |
---|
231 | </para> |
---|
232 | |
---|
233 | <para> |
---|
234 | Each object of a sampler class is responsible for one particle |
---|
235 | type. The particle type is given to the constructor of the sampler |
---|
236 | classes via the particle type name, e.g. "neutron". Depending on |
---|
237 | the specific purpose, the <literal>Configure()</literal> of a sampler will |
---|
238 | set up specialized processes (derived from <literal>G4VProcess</literal>) for |
---|
239 | transportation in the parallel geometry, importance |
---|
240 | sampling and weight roulette for the given particle type. When |
---|
241 | <literal>Configure()</literal> is invoked the sampler places the processes in |
---|
242 | the correct order independent of the order in which user invoked |
---|
243 | the <literal>Prepare...</literal> methods. |
---|
244 | </para> |
---|
245 | |
---|
246 | <para> |
---|
247 | <note> |
---|
248 | <para> |
---|
249 | <itemizedlist spacing="compact"> |
---|
250 | <listitem><para> |
---|
251 | The <literal>Prepare...()</literal> functions may each only be invoked |
---|
252 | once. |
---|
253 | </para></listitem> |
---|
254 | <listitem><para> |
---|
255 | To configure the sampling the function <literal>Configure()</literal> |
---|
256 | must be called <emphasis>after</emphasis> the |
---|
257 | <literal>G4RunManager</literal> has been initialized and the PhysicsList has |
---|
258 | been instantiated. |
---|
259 | </para></listitem> |
---|
260 | </itemizedlist> |
---|
261 | </para> |
---|
262 | </note> |
---|
263 | </para> |
---|
264 | |
---|
265 | |
---|
266 | |
---|
267 | |
---|
268 | <para> |
---|
269 | The interface and framework are demonstrated in the |
---|
270 | <literal>examples/extended/biasing</literal> directory, with the main changes being to the |
---|
271 | G4GeometrySampler class and the fact that in the parallel case the WorldVolume |
---|
272 | is a copy of the Mass World. |
---|
273 | |
---|
274 | The parallel geometry now has to inherit from |
---|
275 | <emphasis>G4VUserParallelWorld</emphasis> which also has the <emphasis>GetWorld()</emphasis> method |
---|
276 | in order to retrieve a copy of the mass geometry WorldVolume. |
---|
277 | |
---|
278 | <informalexample> |
---|
279 | <programlisting> |
---|
280 | class B02ImportanceDetectorConstruction : public G4VUserParallelWorld |
---|
281 | ghostWorld = GetWorld(); |
---|
282 | </programlisting> |
---|
283 | </informalexample> |
---|
284 | |
---|
285 | </para> |
---|
286 | |
---|
287 | <para> |
---|
288 | The constructor for <emphasis>G4GeometrySampler</emphasis> takes a pointer to |
---|
289 | the physical world volume and the particle type name (e.g. "neutron") as arguments. |
---|
290 | In a single mass geometry the sampler is created as follows: |
---|
291 | |
---|
292 | <informalexample> |
---|
293 | <programlisting> |
---|
294 | G4GeometrySampler mgs(detector->GetWorldVolume(),"neutron"); |
---|
295 | mgs.SetParallel(false); |
---|
296 | </programlisting> |
---|
297 | </informalexample> |
---|
298 | |
---|
299 | |
---|
300 | Whilst the following lines of code are required in order to set up the sampler for the |
---|
301 | parallel geometry case: |
---|
302 | |
---|
303 | <informalexample> |
---|
304 | <programlisting> |
---|
305 | G4VPhysicalVolume* ghostWorld = pdet->GetWorldVolume(); |
---|
306 | |
---|
307 | G4GeometrySampler pgs(ghostWorld,"neutron"); |
---|
308 | |
---|
309 | pgs.SetParallel(true); |
---|
310 | </programlisting> |
---|
311 | </informalexample> |
---|
312 | |
---|
313 | |
---|
314 | Also note that the preparation and configuration of the samplers has to be |
---|
315 | carried out <emphasis>after</emphasis> the instantiation of the UserPhysicsList |
---|
316 | and after the initialisation of the <emphasis>G4RunManager</emphasis>: |
---|
317 | |
---|
318 | <informalexample> |
---|
319 | <programlisting> |
---|
320 | pgs.PrepareImportanceSampling(&aIstore, 0); |
---|
321 | pgs.Configure(); |
---|
322 | </programlisting> |
---|
323 | </informalexample> |
---|
324 | |
---|
325 | Due to the fact that biasing is a process and has to be inserted after all the |
---|
326 | other processes have been created. |
---|
327 | </para> |
---|
328 | |
---|
329 | |
---|
330 | </sect3> |
---|
331 | |
---|
332 | <!-- ******************* Section (Level#3) ****************** --> |
---|
333 | <sect3 id="sect.EvtBias.ScorImpRoul.ImpSamp"> |
---|
334 | <title> |
---|
335 | Importance Sampling |
---|
336 | </title> |
---|
337 | |
---|
338 | <para> |
---|
339 | Importance sampling acts on particles crossing boundaries |
---|
340 | between "importance cells". The action taken depends on the |
---|
341 | importance values assigned to the cells. In general a particle |
---|
342 | history is either split or Russian roulette is played if the |
---|
343 | importance increases or decreases, respectively. A weight assigned |
---|
344 | to the history is changed according to the action taken. |
---|
345 | </para> |
---|
346 | |
---|
347 | <para> |
---|
348 | The tools provided for importance sampling require the user to |
---|
349 | have a good understanding of the physics in the problem. This is |
---|
350 | because the user has to decide which particle types require |
---|
351 | importance sampled, define the cells, and assign importance values |
---|
352 | to the cells. If this is not done properly the results cannot be |
---|
353 | expected to describe a real experiment. |
---|
354 | </para> |
---|
355 | |
---|
356 | <para> |
---|
357 | The assignment of importance values to a cell is done using an |
---|
358 | importance store described below. |
---|
359 | </para> |
---|
360 | |
---|
361 | <para> |
---|
362 | An "importance store" with the interface <literal>G4VIStore</literal> is |
---|
363 | used to store importance values related to cells. In order to do |
---|
364 | importance sampling the user has to create an object (e.g. of class |
---|
365 | <literal>G4IStore</literal>) of type <literal>G4VIStore</literal>. The samplers may be |
---|
366 | given a <literal>G4VIStore</literal>. The user fills the store with cells and |
---|
367 | their importance values. |
---|
368 | </para> |
---|
369 | |
---|
370 | <para> |
---|
371 | An importance store has to be constructed with a reference to |
---|
372 | the world volume of the geometry used for importance sampling. This |
---|
373 | may be the world volume of the mass or of a parallel geometry. |
---|
374 | Importance stores derive from the interface <literal>G4VIStore</literal>: |
---|
375 | |
---|
376 | <anchor id="anchor_EvtBias_G4VIStore" /> |
---|
377 | <informalexample> |
---|
378 | <programlisting> |
---|
379 | class G4VIStore |
---|
380 | { |
---|
381 | public: |
---|
382 | G4VIStore(); |
---|
383 | virtual ~G4VIStore(); |
---|
384 | virtual G4double GetImportance(const G4GeometryCell &gCell) const = 0; |
---|
385 | virtual G4bool IsKnown(const G4GeometryCell &gCell) const = 0; |
---|
386 | virtual const G4VPhysicalVolume &GetWorldVolume() const = 0; |
---|
387 | }; |
---|
388 | </programlisting> |
---|
389 | </informalexample> |
---|
390 | </para> |
---|
391 | |
---|
392 | <para> |
---|
393 | A concrete implementation of an importance store is provided by |
---|
394 | the class <literal>G4VStore</literal>. The <emphasis>public</emphasis> |
---|
395 | part of the class is: |
---|
396 | |
---|
397 | <informalexample> |
---|
398 | <programlisting> |
---|
399 | class G4IStore : public G4VIStore |
---|
400 | { |
---|
401 | public: |
---|
402 | explicit G4IStore(const G4VPhysicalVolume &worldvolume); |
---|
403 | virtual ~G4IStore(); |
---|
404 | virtual G4double GetImportance(const G4GeometryCell &gCell) const; |
---|
405 | virtual G4bool IsKnown(const G4GeometryCell &gCell) const; |
---|
406 | virtual const G4VPhysicalVolume &GetWorldVolume() const; |
---|
407 | void AddImportanceGeometryCell(G4double importance, |
---|
408 | const G4GeometryCell &gCell); |
---|
409 | void AddImportanceGeometryCell(G4double importance, |
---|
410 | const G4VPhysicalVolume &, |
---|
411 | G4int aRepNum = 0); |
---|
412 | void ChangeImportance(G4double importance, |
---|
413 | const G4GeometryCell &gCell); |
---|
414 | void ChangeImportance(G4double importance, |
---|
415 | const G4VPhysicalVolume &, |
---|
416 | G4int aRepNum = 0); |
---|
417 | G4double GetImportance(const G4VPhysicalVolume &, |
---|
418 | G4int aRepNum = 0) const ; |
---|
419 | private: ..... |
---|
420 | }; |
---|
421 | </programlisting> |
---|
422 | </informalexample> |
---|
423 | </para> |
---|
424 | |
---|
425 | <para> |
---|
426 | The member function <literal>AddImportanceGeometryCell()</literal> enters |
---|
427 | a cell and an importance value into the importance store. The |
---|
428 | importance values may be returned either according to a physical |
---|
429 | volume and a replica number or according to a |
---|
430 | <literal>G4GeometryCell</literal>. The user must be aware of the |
---|
431 | interpretation of assigning importance values to a cell. |
---|
432 | If scoring is also implemented then this is attached to logical volumes, in |
---|
433 | which case the physical volume and replica number method should be used for |
---|
434 | assigning importance values. See <literal>examples/extended/biasing |
---|
435 | B01</literal> and <literal>B02</literal> for examples of this. |
---|
436 | </para> |
---|
437 | |
---|
438 | <para> |
---|
439 | <note> |
---|
440 | <para> |
---|
441 | <itemizedlist spacing="compact"> |
---|
442 | <listitem><para> |
---|
443 | An importance value must be assigned to every cell. |
---|
444 | </para></listitem> |
---|
445 | </itemizedlist> |
---|
446 | </para> |
---|
447 | </note> |
---|
448 | </para> |
---|
449 | |
---|
450 | <para> |
---|
451 | The different cases: |
---|
452 | |
---|
453 | <itemizedlist spacing="compact"> |
---|
454 | <listitem><para> |
---|
455 | <emphasis>Cell is not in store</emphasis> |
---|
456 | <para> |
---|
457 | Not filling a certain cell in the store will cause an |
---|
458 | exception. |
---|
459 | </para> |
---|
460 | </para></listitem> |
---|
461 | <listitem><para> |
---|
462 | <emphasis>Importance value = zero</emphasis> |
---|
463 | <para> |
---|
464 | Tracks of the chosen particle type will be killed. |
---|
465 | </para> |
---|
466 | </para></listitem> |
---|
467 | <listitem><para> |
---|
468 | <emphasis>importance values > 0</emphasis> |
---|
469 | <para> |
---|
470 | Normal allowed values |
---|
471 | </para> |
---|
472 | </para></listitem> |
---|
473 | <listitem><para> |
---|
474 | <emphasis>Importance value smaller zero</emphasis> |
---|
475 | <para> |
---|
476 | Not allowed! |
---|
477 | </para> |
---|
478 | </para></listitem> |
---|
479 | </itemizedlist> |
---|
480 | </para> |
---|
481 | |
---|
482 | </sect3> |
---|
483 | |
---|
484 | <!-- ******************* Section (Level#3) ****************** --> |
---|
485 | <sect3 id="sect.EvtBias.ScorImpRoul.SampAlgor"> |
---|
486 | <title> |
---|
487 | The Importance Sampling Algorithm |
---|
488 | </title> |
---|
489 | |
---|
490 | <para> |
---|
491 | Importance sampling supports using a customized importance |
---|
492 | sampling algorithm. To this end, the sampler interface |
---|
493 | <link linkend="anchor_EvtBias_G4VSampler"> |
---|
494 | <literal>G4VSampler</literal></link> |
---|
495 | may be given a pointer to the interface |
---|
496 | <literal>G4VImportanceAlgorithm</literal>: |
---|
497 | |
---|
498 | <informalexample> |
---|
499 | <programlisting> |
---|
500 | class G4VImportanceAlgorithm |
---|
501 | { |
---|
502 | public: |
---|
503 | G4VImportanceAlgorithm(); |
---|
504 | virtual ~G4VImportanceAlgorithm(); |
---|
505 | virtual G4Nsplit_Weight Calculate(G4double ipre, |
---|
506 | G4double ipost, |
---|
507 | G4double init_w) const = 0; |
---|
508 | }; |
---|
509 | </programlisting> |
---|
510 | </informalexample> |
---|
511 | </para> |
---|
512 | |
---|
513 | <para> |
---|
514 | The method <literal>Calculate()</literal> takes the arguments: |
---|
515 | |
---|
516 | <itemizedlist spacing="compact"> |
---|
517 | <listitem><para> |
---|
518 | <emphasis>ipre, ipost</emphasis>: importance |
---|
519 | of the previous cell and the importance of the current cell, |
---|
520 | respectively. |
---|
521 | </para></listitem> |
---|
522 | <listitem><para> |
---|
523 | <emphasis>init_w</emphasis>: the particles weight |
---|
524 | </para></listitem> |
---|
525 | </itemizedlist> |
---|
526 | </para> |
---|
527 | |
---|
528 | <para> |
---|
529 | It returns the struct: |
---|
530 | |
---|
531 | <informalexample> |
---|
532 | <programlisting> |
---|
533 | class G4Nsplit_Weight |
---|
534 | { |
---|
535 | public: |
---|
536 | |
---|
537 | G4int fN; |
---|
538 | G4double fW; |
---|
539 | }; |
---|
540 | </programlisting> |
---|
541 | </informalexample> |
---|
542 | |
---|
543 | <itemizedlist spacing="compact"> |
---|
544 | <listitem><para> |
---|
545 | <emphasis>fN</emphasis>: the calculated |
---|
546 | number of particles to exit the importance sampling |
---|
547 | </para></listitem> |
---|
548 | <listitem><para> |
---|
549 | <emphasis>fW</emphasis>: the weight of the particles |
---|
550 | </para></listitem> |
---|
551 | </itemizedlist> |
---|
552 | </para> |
---|
553 | |
---|
554 | <para> |
---|
555 | The user may have a customized algorithm used by providing a |
---|
556 | class inheriting from <literal>G4VImportanceAlgorithm</literal>. |
---|
557 | </para> |
---|
558 | |
---|
559 | <para> |
---|
560 | If no customized algorithm is given to the sampler the default |
---|
561 | importance sampling algorithm is used. This algorithm is |
---|
562 | implemented in <literal>G4ImportanceAlgorithm</literal>. |
---|
563 | </para> |
---|
564 | |
---|
565 | </sect3> |
---|
566 | |
---|
567 | <!-- ******************* Section (Level#3) ****************** --> |
---|
568 | <sect3 id="sect.EvtBias.ScorImpRoul.WeightWin"> |
---|
569 | <title> |
---|
570 | The Weight Window Technique |
---|
571 | </title> |
---|
572 | |
---|
573 | <para> |
---|
574 | The weight window technique is a weight-based alternative to |
---|
575 | importance sampling: |
---|
576 | |
---|
577 | <itemizedlist spacing="compact"> |
---|
578 | <listitem><para> |
---|
579 | applies splitting and Russian roulette depending on space |
---|
580 | (cells) and energy |
---|
581 | </para></listitem> |
---|
582 | <listitem><para> |
---|
583 | user defines weight windows in contrast to defining importance |
---|
584 | values as in importance sampling |
---|
585 | </para></listitem> |
---|
586 | </itemizedlist> |
---|
587 | </para> |
---|
588 | |
---|
589 | <para> |
---|
590 | In contrast to importance sampling this technique is not weight |
---|
591 | blind. Instead the technique is applied according to the particle |
---|
592 | weight with respect to the current energy-space cell. |
---|
593 | </para> |
---|
594 | |
---|
595 | <para> |
---|
596 | Therefore the technique is convenient to apply in combination |
---|
597 | with other variance reduction techniques such as cross-section |
---|
598 | biasing and implicit capture. |
---|
599 | </para> |
---|
600 | |
---|
601 | <para> |
---|
602 | A weight window may be specified for every cell and for several |
---|
603 | energy regions: <emphasis>space-energy cell</emphasis>. |
---|
604 | |
---|
605 | <figure id="fig.EvtBias.WeightWindow"> |
---|
606 | <title> |
---|
607 | Weight window concept |
---|
608 | </title> |
---|
609 | |
---|
610 | <mediaobject> |
---|
611 | <imageobject role="fo"> |
---|
612 | <imagedata fileref="./AllResources/Fundamentals/wwconcept.jpg" |
---|
613 | format="JPG" contentwidth="9.0cm" align="center" /> |
---|
614 | </imageobject> |
---|
615 | <imageobject role="html"> |
---|
616 | <imagedata fileref="./AllResources/Fundamentals/wwconcept.jpg" |
---|
617 | format="JPG" align="center" /> |
---|
618 | </imageobject> |
---|
619 | <textobject> |
---|
620 | <phrase>Weight window concept</phrase> |
---|
621 | </textobject> |
---|
622 | </mediaobject> |
---|
623 | </figure> |
---|
624 | </para> |
---|
625 | |
---|
626 | <!-- ******* Bridgehead ******* --> |
---|
627 | <bridgehead renderas='sect4'> |
---|
628 | Weight window concept |
---|
629 | </bridgehead> |
---|
630 | |
---|
631 | <para> |
---|
632 | The user specifies a <emphasis>lower weight bound W_L</emphasis> |
---|
633 | for every space-energy cell. |
---|
634 | |
---|
635 | <itemizedlist spacing="compact"> |
---|
636 | <listitem><para> |
---|
637 | The upper weight bound W_U and the survival weight W_S are |
---|
638 | calculated as: |
---|
639 | <para> |
---|
640 | W_U = C_U <emphasis>W_L</emphasis> and |
---|
641 | </para> |
---|
642 | <para> |
---|
643 | W_S = C_S <emphasis>W_L</emphasis>. |
---|
644 | </para> |
---|
645 | </para></listitem> |
---|
646 | <listitem><para> |
---|
647 | The user specifies C_S and C_U once for the whole problem. |
---|
648 | </para></listitem> |
---|
649 | <listitem><para> |
---|
650 | The user may give different sets of energy bounds for every cell |
---|
651 | or one set for all geometrical cells |
---|
652 | </para></listitem> |
---|
653 | <listitem><para> |
---|
654 | Special case: if C_S = C_U = 1 for all energies then weight |
---|
655 | window is equivalent to importance sampling |
---|
656 | </para></listitem> |
---|
657 | <listitem><para> |
---|
658 | The user can choose to apply the technique: at boundaries, on |
---|
659 | collisions or on boundaries and collisions |
---|
660 | </para></listitem> |
---|
661 | </itemizedlist> |
---|
662 | </para> |
---|
663 | |
---|
664 | <para> |
---|
665 | The energy-space cells are realized by <literal>G4GeometryCell</literal> |
---|
666 | as in importance sampling. The cells are stored in a weight window |
---|
667 | store defined by <literal>G4VWeightWindowStore</literal>: |
---|
668 | |
---|
669 | <informalexample> |
---|
670 | <programlisting> |
---|
671 | class G4VWeightWindowStore { |
---|
672 | public: |
---|
673 | G4VWeightWindowStore(); |
---|
674 | virtual ~G4VWeightWindowStore(); |
---|
675 | virtual G4double GetLowerWeitgh(const G4GeometryCell &gCell, |
---|
676 | G4double partEnergy) const = 0; |
---|
677 | virtual G4bool IsKnown(const G4GeometryCell &gCell) const = 0; |
---|
678 | virtual const G4VPhysicalVolume &GetWorldVolume() const = 0; |
---|
679 | }; |
---|
680 | </programlisting> |
---|
681 | </informalexample> |
---|
682 | </para> |
---|
683 | |
---|
684 | <para> |
---|
685 | A concrete implementation is provided: |
---|
686 | |
---|
687 | <informalexample> |
---|
688 | <programlisting> |
---|
689 | class G4WeightWindowStore: public G4VWeightWindowStore { |
---|
690 | public: |
---|
691 | explicit G4WeightWindowStore(const G4VPhysicalVolume &worldvolume); |
---|
692 | virtual ~G4WeightWindowStore(); |
---|
693 | virtual G4double GetLowerWeitgh(const G4GeometryCell &gCell, |
---|
694 | G4double partEnergy) const; |
---|
695 | virtual G4bool IsKnown(const G4GeometryCell &gCell) const; |
---|
696 | virtual const G4VPhysicalVolume &GetWorldVolume() const; |
---|
697 | void AddLowerWeights(const G4GeometryCell &gCell, |
---|
698 | const std::vector<G4double> &lowerWeights); |
---|
699 | void AddUpperEboundLowerWeightPairs(const G4GeometryCell &gCell, |
---|
700 | const G4UpperEnergyToLowerWeightMap& |
---|
701 | enWeMap); |
---|
702 | void SetGeneralUpperEnergyBounds(const |
---|
703 | std::set<G4double, std::less<G4double> > & enBounds); |
---|
704 | |
---|
705 | private:: |
---|
706 | ... |
---|
707 | }; |
---|
708 | </programlisting> |
---|
709 | </informalexample> |
---|
710 | </para> |
---|
711 | |
---|
712 | <para> |
---|
713 | The user may choose equal energy bounds for all cells. In this |
---|
714 | case a set of upper energy bounds must be given to the store using |
---|
715 | the method <literal>SetGeneralUpperEnergyBounds</literal>. If a general set |
---|
716 | of energy bounds have been set <literal>AddLowerWeights</literal> can be used |
---|
717 | to add the cells. |
---|
718 | </para> |
---|
719 | |
---|
720 | <para> |
---|
721 | Alternatively, the user may chose different energy regions for |
---|
722 | different cells. In this case the user must provide a mapping of |
---|
723 | upper energy bounds to lower weight bounds for every cell using the |
---|
724 | method <literal>AddUpperEboundLowerWeightPairs</literal>. |
---|
725 | </para> |
---|
726 | |
---|
727 | <para> |
---|
728 | Weight window algorithms implementing the interface class |
---|
729 | <literal>G4VWeightWindowAlgorithm</literal> can be used to define a |
---|
730 | customized algorithm: |
---|
731 | |
---|
732 | <informalexample> |
---|
733 | <programlisting> |
---|
734 | class G4VWeightWindowAlgorithm { |
---|
735 | public: |
---|
736 | G4VWeightWindowAlgorithm(); |
---|
737 | virtual ~G4VWeightWindowAlgorithm(); |
---|
738 | virtual G4Nsplit_Weight Calculate(G4double init_w, |
---|
739 | G4double lowerWeightBound) const = 0; |
---|
740 | }; |
---|
741 | </programlisting> |
---|
742 | </informalexample> |
---|
743 | </para> |
---|
744 | |
---|
745 | <para> |
---|
746 | A concrete implementation is provided and used as a default: |
---|
747 | |
---|
748 | <informalexample> |
---|
749 | <programlisting> |
---|
750 | class G4WeightWindowAlgorithm : public G4VWeightWindowAlgorithm { |
---|
751 | public: |
---|
752 | G4WeightWindowAlgorithm(G4double upperLimitFaktor = 5, |
---|
753 | G4double survivalFaktor = 3, |
---|
754 | G4int maxNumberOfSplits = 5); |
---|
755 | virtual ~G4WeightWindowAlgorithm(); |
---|
756 | virtual G4Nsplit_Weight Calculate(G4double init_w, |
---|
757 | G4double lowerWeightBound) const; |
---|
758 | private: |
---|
759 | ... |
---|
760 | }; |
---|
761 | </programlisting> |
---|
762 | </informalexample> |
---|
763 | </para> |
---|
764 | |
---|
765 | <para> |
---|
766 | The constructor takes three parameters which are used to: |
---|
767 | calculate the upper weight bound (upperLimitFaktor), calculate the |
---|
768 | survival weight (survivalFaktor), and introduce a maximal number |
---|
769 | (maxNumberOfSplits) of copies to be created in one go. |
---|
770 | </para> |
---|
771 | |
---|
772 | <para> |
---|
773 | In addition, the inverse of the maxNumberOfSplits is used to |
---|
774 | specify the minimum survival probability in case of Russian |
---|
775 | roulette. |
---|
776 | </para> |
---|
777 | |
---|
778 | </sect3> |
---|
779 | |
---|
780 | <!-- ******************* Section (Level#3) ****************** --> |
---|
781 | <sect3 id="sect.EvtBias.ScorImpRoul.WeigRoul"> |
---|
782 | <title> |
---|
783 | The Weight Roulette Technique |
---|
784 | </title> |
---|
785 | |
---|
786 | <para> |
---|
787 | Weight roulette (also called weight cutoff) is usually applied |
---|
788 | if importance sampling and implicit capture are used together. |
---|
789 | Implicit capture is not described here but it is useful to note |
---|
790 | that this procedure reduces a particle weight in every collision |
---|
791 | instead of killing the particle with some probability. |
---|
792 | </para> |
---|
793 | |
---|
794 | <para> |
---|
795 | Together with importance sampling the weight of a particle may |
---|
796 | become so low that it does not change any result significantly. |
---|
797 | Hence tracking a very low weight particle is a waste of computing |
---|
798 | time. Weight roulette is applied in order to solve this |
---|
799 | problem. |
---|
800 | </para> |
---|
801 | |
---|
802 | <!-- ******* Bridgehead ******* --> |
---|
803 | <bridgehead renderas='sect4'> |
---|
804 | The weight roulette concept |
---|
805 | </bridgehead> |
---|
806 | |
---|
807 | <para> |
---|
808 | Weight roulette takes into account the importance "Ic" of the |
---|
809 | current cell and the importance "Is" of the cell in which the |
---|
810 | source is located, by using the ratio "R=Is/Ic". |
---|
811 | </para> |
---|
812 | |
---|
813 | <para> |
---|
814 | Weight roulette uses a relative minimal weight limit and a |
---|
815 | relative survival weight. When a particle falls below the weight |
---|
816 | limit Russian roulette is applied. If the particle survives, |
---|
817 | tracking will be continued and the particle weight will be set to |
---|
818 | the survival weight. |
---|
819 | </para> |
---|
820 | |
---|
821 | <para> |
---|
822 | The weight roulette uses the following parameters with their |
---|
823 | default values: |
---|
824 | |
---|
825 | <itemizedlist spacing="compact"> |
---|
826 | <listitem><para> |
---|
827 | <emphasis>wsurvival</emphasis>: 0.5 |
---|
828 | </para></listitem> |
---|
829 | <listitem><para> |
---|
830 | <emphasis>wlimit</emphasis>: 0.25 |
---|
831 | </para></listitem> |
---|
832 | <listitem><para> |
---|
833 | <emphasis>isource</emphasis>: 1 |
---|
834 | </para></listitem> |
---|
835 | </itemizedlist> |
---|
836 | </para> |
---|
837 | |
---|
838 | <para> |
---|
839 | The following algorithm is applied: |
---|
840 | </para> |
---|
841 | |
---|
842 | <para> |
---|
843 | If a particle weight "w" is lower than R*wlimit: |
---|
844 | |
---|
845 | <itemizedlist spacing="compact"> |
---|
846 | <listitem><para> |
---|
847 | the weight of the particle will be changed to "ws = wsurvival*R" |
---|
848 | </para></listitem> |
---|
849 | <listitem><para> |
---|
850 | the probability for the particle to survive is "p = w/ws" |
---|
851 | </para></listitem> |
---|
852 | </itemizedlist> |
---|
853 | </para> |
---|
854 | |
---|
855 | </sect3> |
---|
856 | </sect2> |
---|
857 | |
---|
858 | |
---|
859 | <!-- ******************* Section (Level#2) ****************** --> |
---|
860 | <sect2 id="sect.EvtBias.PhysBias"> |
---|
861 | <title> |
---|
862 | Physics Based Biasing |
---|
863 | </title> |
---|
864 | |
---|
865 | <para> |
---|
866 | Geant4 supports physics based biasing through a number of general |
---|
867 | use, built in biasing techniques. A utility class, |
---|
868 | G4WrapperProcess, is also available to support user defined |
---|
869 | biasing. |
---|
870 | </para> |
---|
871 | |
---|
872 | <!-- ******************* Section (Level#3) ****************** --> |
---|
873 | <sect3 id="sect.EvtBias.PhysBias.BuiltInOpt"> |
---|
874 | <title> |
---|
875 | Built in Biasing Options |
---|
876 | </title> |
---|
877 | |
---|
878 | <!-- ******************* Section (Level#4) ****************** --> |
---|
879 | <sect4 id="sect.EvtBias.PhysBias.BuiltInOpt.PrimPart"> |
---|
880 | <title> |
---|
881 | Primary Particle Biasing |
---|
882 | </title> |
---|
883 | |
---|
884 | <para> |
---|
885 | Primary particle biasing can be used to increase the number of |
---|
886 | primary particles generated in a particular phase space region of |
---|
887 | interest. The weight of the primary particle is modified as |
---|
888 | appropriate. A general implementation is provided through the |
---|
889 | G4GeneralParticleSource class. It is possible to bias position, |
---|
890 | angular and energy distributions. |
---|
891 | </para> |
---|
892 | |
---|
893 | <para> |
---|
894 | G4GeneralParticleSource is a concrete implementation of |
---|
895 | G4VPrimaryGenerator. To use, instantiate G4GeneralParticleSource in |
---|
896 | the G4VUserPrimaryGeneratorAction class, as demonstrated below. |
---|
897 | |
---|
898 | <informalexample> |
---|
899 | <programlisting> |
---|
900 | MyPrimaryGeneratorAction::MyPrimaryGeneratorAction() { |
---|
901 | generator = new G4GeneralParticleSource; |
---|
902 | } |
---|
903 | |
---|
904 | void |
---|
905 | MyPrimaryGeneratorAction::GeneratePrimaries(G4Event*anEvent){ |
---|
906 | generator->GeneratePrimaryVertex(anEvent); |
---|
907 | } |
---|
908 | </programlisting> |
---|
909 | </informalexample> |
---|
910 | </para> |
---|
911 | |
---|
912 | <para> |
---|
913 | The biasing can be configured through interactive commands. |
---|
914 | Extensive documentation can be found in |
---|
915 | <ulink url="http://reat.space.qinetiq.com/gps/"> |
---|
916 | Primary particle biasing</ulink>. Examples are also distributed |
---|
917 | with the Geant4 distribution in |
---|
918 | <emphasis role="bold">examples/extended/eventgenerator/exgps</emphasis>. |
---|
919 | </para> |
---|
920 | |
---|
921 | </sect4> |
---|
922 | |
---|
923 | <!-- ******************* Section (Level#4) ****************** --> |
---|
924 | <sect4 id="sect.EvtBias.PhysBias.BuiltInOpt.RadDcy"> |
---|
925 | <title> |
---|
926 | Radioactive Decay Biasing |
---|
927 | </title> |
---|
928 | |
---|
929 | <para> |
---|
930 | The G4RadioactiveDecay class simulates the decay of radioactive |
---|
931 | nuclei and implements the following biasing options: |
---|
932 | |
---|
933 | <itemizedlist spacing="compact"> |
---|
934 | <listitem><para> |
---|
935 | Increase the sampling rate of radionuclides within observation |
---|
936 | times through a user defined probability distribution function |
---|
937 | </para></listitem> |
---|
938 | <listitem><para> |
---|
939 | Nuclear splitting, where the parent nuclide is split into a |
---|
940 | user defined number of nuclides |
---|
941 | </para></listitem> |
---|
942 | <listitem><para> |
---|
943 | Branching ratio biasing where branching ratios are sampled with |
---|
944 | equal probability |
---|
945 | </para></listitem> |
---|
946 | </itemizedlist> |
---|
947 | </para> |
---|
948 | |
---|
949 | <para> |
---|
950 | G4RadioactiveDecay is a process which must be registered with a |
---|
951 | process manager, as demonstrated below. |
---|
952 | |
---|
953 | <informalexample> |
---|
954 | <programlisting> |
---|
955 | void MyPhysicsList::ConstructProcess() |
---|
956 | { |
---|
957 | ... |
---|
958 | G4RadioactiveDecay* theRadioactiveDecay = |
---|
959 | new G4RadioactiveDecay(); |
---|
960 | |
---|
961 | G4ProcessManager* pmanager = ... |
---|
962 | pmanager ->AddProcess(theRadioactiveDecay); |
---|
963 | ... |
---|
964 | } |
---|
965 | </programlisting> |
---|
966 | </informalexample> |
---|
967 | </para> |
---|
968 | |
---|
969 | <para> |
---|
970 | The biasing can be controlled either in compiled code or through |
---|
971 | interactive commands. Extensive documentation can be found in |
---|
972 | |
---|
973 | <ulink url="http://reat.space.qinetiq.com/septimess/exrdm/"> |
---|
974 | Radioactive decay biasing example |
---|
975 | </ulink> |
---|
976 | and |
---|
977 | <ulink url="http://www.space.qinetiq.com/geant4/rdm.html"> |
---|
978 | Radioactive decay biasing |
---|
979 | </ulink>. |
---|
980 | </para> |
---|
981 | |
---|
982 | <para> |
---|
983 | Radioactive decay biasing examples are also distributed with the Geant4 |
---|
984 | distribution in |
---|
985 | <emphasis role="bold">examples/extended/radioactivedecay/exrdm</emphasis>. |
---|
986 | </para> |
---|
987 | |
---|
988 | </sect4> |
---|
989 | |
---|
990 | <!-- ******************* Section (Level#4) ****************** --> |
---|
991 | <sect4 id="sect.EvtBias.PhysBias.BuiltInOpt.HadLeadPar"> |
---|
992 | <title> |
---|
993 | Hadronic Leading Particle Biasing |
---|
994 | </title> |
---|
995 | |
---|
996 | <para> |
---|
997 | One hadronic leading particle biasing technique is |
---|
998 | implemented in the G4HadLeadBias utility. This method keeps only |
---|
999 | the most important part of the event, as well as representative |
---|
1000 | tracks of each given particle type. So the track with the highest |
---|
1001 | energy as well as one of each of Baryon, pi0, mesons and leptons. |
---|
1002 | As usual, appropriate weights are assigned to the particles. |
---|
1003 | Setting the <emphasis role="bold">SwitchLeadBiasOn</emphasis> |
---|
1004 | environmental variable will activate this utility. |
---|
1005 | </para> |
---|
1006 | |
---|
1007 | </sect4> |
---|
1008 | |
---|
1009 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1010 | <sect4 id="sect.EvtBias.PhysBias.BuiltInOpt.HadCrsSect"> |
---|
1011 | <title> |
---|
1012 | Hadronic Cross Section Biasing |
---|
1013 | </title> |
---|
1014 | |
---|
1015 | <para> |
---|
1016 | Cross section biasing artificially enhances/reduces the cross |
---|
1017 | section of a process. This may be useful for studying thin layer |
---|
1018 | interactions or thick layer shielding. The built in hadronic cross |
---|
1019 | section biasing applies to photon inelastic, electron nuclear and |
---|
1020 | positron nuclear processes. |
---|
1021 | </para> |
---|
1022 | |
---|
1023 | <para> |
---|
1024 | The biasing is controlled through the |
---|
1025 | <emphasis role="bold">BiasCrossSectionByFactor</emphasis> method |
---|
1026 | in G4HadronicProcess, as demonstrated below. |
---|
1027 | |
---|
1028 | <informalexample> |
---|
1029 | <programlisting> |
---|
1030 | void MyPhysicsList::ConstructProcess() |
---|
1031 | { |
---|
1032 | ... |
---|
1033 | G4ElectroNuclearReaction * theElectroReaction = |
---|
1034 | new G4ElectroNuclearReaction; |
---|
1035 | |
---|
1036 | G4ElectronNuclearProcess theElectronNuclearProcess; |
---|
1037 | theElectronNuclearProcess.RegisterMe(theElectroReaction); |
---|
1038 | theElectronNuclearProcess.BiasCrossSectionByFactor(100); |
---|
1039 | |
---|
1040 | pManager->AddDiscreteProcess(&theElectronNuclearProcess); |
---|
1041 | ... |
---|
1042 | } |
---|
1043 | </programlisting> |
---|
1044 | </informalexample> |
---|
1045 | </para> |
---|
1046 | |
---|
1047 | </sect4> |
---|
1048 | </sect3> |
---|
1049 | |
---|
1050 | |
---|
1051 | <!-- ******************* Section (Level#3) ****************** --> |
---|
1052 | <sect3 id="sect.EvtBias.PhysBias."> |
---|
1053 | <title> |
---|
1054 | G4WrapperProcess |
---|
1055 | </title> |
---|
1056 | |
---|
1057 | <para> |
---|
1058 | G4WrapperProcess can be used to implement user defined event |
---|
1059 | biasing. G4WrapperProcess, which is a process itself, wraps an |
---|
1060 | existing process. By default, all function calls are forwared to |
---|
1061 | the wrapped process. It is a non-invasive way to modify the |
---|
1062 | behaviour of an existing process. |
---|
1063 | </para> |
---|
1064 | |
---|
1065 | <para> |
---|
1066 | To use this utility, first create a derived class inheriting |
---|
1067 | from G4WrapperProcess. Override the methods whose behaviour you |
---|
1068 | would like to modify, for example, PostStepDoIt, and register the |
---|
1069 | derived class in place of the process to be wrapped. Finally, |
---|
1070 | register the wrapped process with G4WrapperProcess. The code |
---|
1071 | snippets below demonstrate its use. |
---|
1072 | |
---|
1073 | <informalexample> |
---|
1074 | <programlisting> |
---|
1075 | class MyWrapperProcess : public G4WrapperProcess { |
---|
1076 | ... |
---|
1077 | G4VParticleChange* PostStepDoIt(const G4Track& track, |
---|
1078 | const G4Step& step) { |
---|
1079 | // Do something interesting |
---|
1080 | } |
---|
1081 | }; |
---|
1082 | |
---|
1083 | |
---|
1084 | void MyPhysicsList::ConstructProcess() |
---|
1085 | { |
---|
1086 | ... |
---|
1087 | G4LowEnergyBremsstrahlung* bremProcess = |
---|
1088 | new G4LowEnergyBremsstrahlung(); |
---|
1089 | |
---|
1090 | MyWrapperProcess* wrapper = new MyWrapperProcess(); |
---|
1091 | wrapper->RegisterProcess(bremProcess); |
---|
1092 | |
---|
1093 | processManager->AddProcess(wrapper, -1, -1, 3); |
---|
1094 | } |
---|
1095 | </programlisting> |
---|
1096 | </informalexample> |
---|
1097 | </para> |
---|
1098 | |
---|
1099 | </sect3> |
---|
1100 | |
---|
1101 | </sect2> |
---|
1102 | |
---|
1103 | <!-- ******************* Section (Level#2) ****************** --> |
---|
1104 | <sect2 id="sect.EvtBias.ReverseMC"> |
---|
1105 | <title> |
---|
1106 | Adjoint/Reverse Monte Carlo |
---|
1107 | </title> |
---|
1108 | |
---|
1109 | <para> |
---|
1110 | Another powerful biasing technique available in Geant4 is the Reverse Monte Carlo (RMC) method, |
---|
1111 | also known as the Adjoint Monte Carlo method. In this method particles are generated on the external |
---|
1112 | boundary of the sensitive part of the geometry and then are tracked backward in the geometry till they |
---|
1113 | reach the external source surface, |
---|
1114 | or exceed an energy threshold. By this way the computing time is focused only on particle tracks |
---|
1115 | that are contributing to the tallies. |
---|
1116 | The RMC method is much rapid than the Forward MC method when the sensitive part of the geometry is |
---|
1117 | small compared to the rest of the geometry and to the external source, that has to be extensive and |
---|
1118 | not beam like. At the moment the RMC method is implemented in Geant4 only for some electromagnetic |
---|
1119 | processes (see <xref linkend="sect.EvtBias.ReverseMC.InG4.Process" />). An example illustrating the use |
---|
1120 | of the Reverse MC method in Geant4 is distributed within the Geant4 |
---|
1121 | toolkit in <emphasis role="bold">examples/extended/biasing/ReverseMC01</emphasis>. |
---|
1122 | </para> |
---|
1123 | |
---|
1124 | <!-- ******************* Section (Level#3) ****************** --> |
---|
1125 | <sect3 id="sect.EvtBias.ReverseMC.InG4"> |
---|
1126 | <title> |
---|
1127 | Treatment of the Reverse MC method in Geant4 |
---|
1128 | </title> |
---|
1129 | |
---|
1130 | <para> |
---|
1131 | Different G4Adjoint classes have been implemented into the Geant4 |
---|
1132 | toolkit in order to run an adjoint/reverse simulation in a Geant4 application. |
---|
1133 | This implementation is illustrated in <xref linkend="fig.EvtBias.ReverseMC.InG4_1" />. |
---|
1134 | An adjoint run is divided in a serie of alternative adjoint and forward tracking of adjoint and normal particles. |
---|
1135 | One Geant4 event treats one of this tracking phase. |
---|
1136 | </para> |
---|
1137 | |
---|
1138 | <figure id="fig.EvtBias.ReverseMC.InG4_1"> |
---|
1139 | <title> |
---|
1140 | Schematic view of an adjoint/reverse simulation in Geant4 |
---|
1141 | </title> |
---|
1142 | <mediaobject> |
---|
1143 | <imageobject role="fo"> |
---|
1144 | <imagedata fileref="./AllResources/Fundamentals/ReverseMC_tracking.png" |
---|
1145 | format="png" align="center" /> |
---|
1146 | </imageobject> |
---|
1147 | <imageobject role="html"> |
---|
1148 | <imagedata fileref="./AllResources/Fundamentals/ReverseMC_tracking.png" |
---|
1149 | format="png" align="center" /> |
---|
1150 | </imageobject> |
---|
1151 | </mediaobject> |
---|
1152 | </figure> |
---|
1153 | |
---|
1154 | |
---|
1155 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1156 | <sect4 id="sect.EvtBias.ReverseMC.InG4.AdjointTracking"> |
---|
1157 | <title> |
---|
1158 | Adjoint tracking phase |
---|
1159 | </title> |
---|
1160 | |
---|
1161 | <para> |
---|
1162 | Adjoint particles (adjoint_e-, adjoint_gamma,...) are generated one by one on the so called adjoint source |
---|
1163 | with random position, energy (1/E distribution) and direction. The adjoint source is the |
---|
1164 | external surface of a user defined volume or of a user defined sphere. The adjoint |
---|
1165 | source should contain one or several sensitive volumes and should be small compared to the entire geometry. |
---|
1166 | The user can set the minimum and maximum energy of the adjoint source. After its |
---|
1167 | generation the adjoint primary particle is tracked backward in the geometry till a user defined |
---|
1168 | external surface |
---|
1169 | (spherical or boundary of a volume) or is killed before if it reaches a user defined upper energy limit that |
---|
1170 | represents the maximum energy of the external source. During the reverse tracking, reverse |
---|
1171 | processes take place where the adjoint particle being tracked can be either scattered |
---|
1172 | or transformed in another type of adjoint particle. During the reverse tracking the |
---|
1173 | G4AdjointSimulationManager replaces the user defined primary, run, stepping, ... actions, by its own actions. |
---|
1174 | A reverse tracking phase corresponds to one Geant4 event. |
---|
1175 | </para> |
---|
1176 | |
---|
1177 | </sect4> |
---|
1178 | |
---|
1179 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1180 | <sect4 id="sect.EvtBias.ReverseMC.InG4.ForwardTracking"> |
---|
1181 | <title> |
---|
1182 | Forward tracking phase |
---|
1183 | </title> |
---|
1184 | |
---|
1185 | <para> |
---|
1186 | When an adjoint particle reaches the external surface its weight, type, position, |
---|
1187 | and direction are registered and a normal primary particle, with a type equivalent to the last generated primary |
---|
1188 | adjoint, is generated with the same energy, position but opposite direction and is tracked in the forward direction |
---|
1189 | in the sensitive region as in a forward MC simulation. |
---|
1190 | During this forward tracking phase the |
---|
1191 | event, stacking, stepping, tracking actions defined by the user for his forward simulation are used. |
---|
1192 | By this clear separation between |
---|
1193 | adjoint and forward tracking phases, the code of the user developed for a forward simulation |
---|
1194 | should be only slightly modified to adapt it for an adjoint simulation (see <xref linkend="sect.EvtBias.ReverseMC.UserCoding" />). |
---|
1195 | Indeed the computation of the signals is done by the same actions |
---|
1196 | or classes that the one used in the forward simulation mode. |
---|
1197 | A forward tracking phase corresponds to one G4 event. |
---|
1198 | </para> |
---|
1199 | |
---|
1200 | </sect4> |
---|
1201 | |
---|
1202 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1203 | <sect4 id="sect.EvtBias.ReverseMC.InG4.Process"> |
---|
1204 | <title> |
---|
1205 | Reverse processes |
---|
1206 | </title> |
---|
1207 | |
---|
1208 | <para> |
---|
1209 | During the reverse tracking, reverse processes act on the adjoint particles. |
---|
1210 | The reverse processes that are at the moment available in Geant4 are the: |
---|
1211 | <itemizedlist spacing="compact"> |
---|
1212 | <listitem><para> |
---|
1213 | Reverse discrete ionization for e-, proton and ions |
---|
1214 | </para></listitem> |
---|
1215 | <listitem><para> |
---|
1216 | Continuous gain of energy by ionization and bremsstrahlung for e- and by ionization for protons and ions |
---|
1217 | </para></listitem> |
---|
1218 | <listitem><para> |
---|
1219 | Reverse discrete e- bremsstrahlung |
---|
1220 | </para></listitem> |
---|
1221 | <listitem><para> |
---|
1222 | Reverse photo-electric effect |
---|
1223 | </para></listitem> |
---|
1224 | <listitem><para> |
---|
1225 | Reverse Compton scattering |
---|
1226 | </para></listitem> |
---|
1227 | <listitem><para> |
---|
1228 | Approximated multiple scattering (see comment in <xref linkend="sect.EvtBias.ReverseMC.Limitation.ReverseMS"/>) |
---|
1229 | </para></listitem> |
---|
1230 | </itemizedlist> |
---|
1231 | It is important to note that the electromagnetic reverse processes are cut dependent |
---|
1232 | as their equivalent forward processes. The implementation of the reverse processes is based on |
---|
1233 | the forward processes implemented in the G4 standard electromagnetic package. |
---|
1234 | </para> |
---|
1235 | |
---|
1236 | </sect4> |
---|
1237 | |
---|
1238 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1239 | <sect4 id="sect.EvtBias.ReverseMC.InG4.RemarkOnNbEvents"> |
---|
1240 | <title> |
---|
1241 | Nb of adjoint particle types and nb of G4 events of an adjoint simulation |
---|
1242 | </title> |
---|
1243 | |
---|
1244 | <para> |
---|
1245 | The list of type of adjoint and forward particles that are generated on the adjoint source |
---|
1246 | and considered in the simulation is a function of the adjoint processes declared in the physics list. |
---|
1247 | For example if only the e- and gamma electromagnetic processes are considered, |
---|
1248 | only adjoint e- and adjoint gamma will be considered as primaries. |
---|
1249 | In this case an adjoint event will be divided in four G4 event consisting in the reverse tracking of an adjoint e-, |
---|
1250 | the forward tracking of its equivalent forward e-, the reverse tracking of |
---|
1251 | an adjoint gamma, and the forward tracking of its equivalent forward gamma. |
---|
1252 | In this case a run of 100 adjoint events will consist into 400 Geant4 events. |
---|
1253 | If the proton ionization is also considered adjoint and forward protons are also generated as primaries |
---|
1254 | and 600 Geant4 events are processed for 100 adjoint events. |
---|
1255 | </para> |
---|
1256 | |
---|
1257 | </sect4> |
---|
1258 | |
---|
1259 | </sect3> |
---|
1260 | |
---|
1261 | <!-- ******************* Section (Level#3) ****************** --> |
---|
1262 | <sect3 id="sect.EvtBias.ReverseMC.UserCoding"> |
---|
1263 | <title> |
---|
1264 | How to update a G4 application to use the reverse Monte Carlo mode |
---|
1265 | </title> |
---|
1266 | |
---|
1267 | <para> |
---|
1268 | Some modifications are needed to an existing Geant4 application in order to adapt |
---|
1269 | it for the use of the reverse simulation mode |
---|
1270 | (see also the G4 example <emphasis role="bold">examples/extended/biasing/ReverseMC01</emphasis>). |
---|
1271 | It consists into the: |
---|
1272 | |
---|
1273 | <itemizedlist spacing="compact"> |
---|
1274 | <listitem><para> |
---|
1275 | Creation of the adjoint simulation manager in the main code |
---|
1276 | </para></listitem> |
---|
1277 | <listitem><para> |
---|
1278 | Optional declaration of user actions that will be used during the adjoint tracking phase |
---|
1279 | </para></listitem> |
---|
1280 | <listitem><para> |
---|
1281 | Use of a special physics lists that combine the adjoint and forward processes |
---|
1282 | </para></listitem> |
---|
1283 | <listitem><para> |
---|
1284 | Modification of the user analysis part of the code |
---|
1285 | </para></listitem> |
---|
1286 | </itemizedlist> |
---|
1287 | </para> |
---|
1288 | |
---|
1289 | |
---|
1290 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1291 | <sect4 id="sect.EvtBias.ReverseMC.UserCoding.Main"> |
---|
1292 | <title> |
---|
1293 | Creation of G4AdjointSimManager in the main |
---|
1294 | </title> |
---|
1295 | <para> |
---|
1296 | The class G4AdjointSimManager represents the manager of an adjoint simulation. |
---|
1297 | This static class should be created somewhere in the main code. The way to do that is illustrated below |
---|
1298 | <informalexample> |
---|
1299 | <programlisting> |
---|
1300 | int main(int argc,char** argv) { |
---|
1301 | ... |
---|
1302 | G4AdjointSimManager* theAdjointSimManager = G4AdjointSimManager::GetInstance(); |
---|
1303 | ... |
---|
1304 | } |
---|
1305 | </programlisting> |
---|
1306 | </informalexample> |
---|
1307 | By doing this the G4 application can be run in the reverse MC mode as well |
---|
1308 | as in the forward MC mode. |
---|
1309 | It is important to note that G4AdjointSimManager |
---|
1310 | is not a new G4RunManager and that the creation of G4RunManager in the main and the declaration of the geometry, |
---|
1311 | physics list, and user actions to G4RunManager is still needed. |
---|
1312 | The definition of the adjoint and external sources and the start of an adjoint simulation can be controlled |
---|
1313 | by G4UI commands in the directory <emphasis role="bold">/adjoint</emphasis>. |
---|
1314 | </para> |
---|
1315 | </sect4> |
---|
1316 | |
---|
1317 | |
---|
1318 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1319 | <sect4 id="sect.EvtBias.ReverseMC.UserCoding.AdjointActions"> |
---|
1320 | <title> |
---|
1321 | Optional declaration of adjoint user actions |
---|
1322 | </title> |
---|
1323 | <para> |
---|
1324 | During an adjoint simulation the user stepping, tracking, stacking and event actions declared to G4RunManager |
---|
1325 | are used only during the G4 events dedicated to the forward tracking of normal particles in the sensitive region, |
---|
1326 | while during the events where adjoint particles are tracked backward the following happen concerning these actions: |
---|
1327 | |
---|
1328 | <itemizedlist spacing="compact"> |
---|
1329 | <listitem><para> |
---|
1330 | The user stepping action is replaced by G4AdjointSteppingAction that is reponsible to stop an adjoint track when |
---|
1331 | it reaches the external source, exceed the maximum energy of the external source, or cross the adjoint source surface. |
---|
1332 | If needed the user can declare its own stepping action that will be called by G4AdjointSteppingAction after the |
---|
1333 | check of stopping track conditions. This stepping action can be different that the stepping action used for the |
---|
1334 | forward simulation. It is declared to G4AdjointSimManager by the following lines of code : |
---|
1335 | |
---|
1336 | <informalexample><programlisting> |
---|
1337 | G4AdjointSimManager* theAdjointSimManager = G4AdjointSimManager::GetInstance(); |
---|
1338 | theAdjointSimManager->SetAdjointSteppingAction(aUserDefinedSteppingAction); |
---|
1339 | </programlisting></informalexample> |
---|
1340 | |
---|
1341 | </para></listitem> |
---|
1342 | |
---|
1343 | <listitem><para> |
---|
1344 | No stacking, tracking and event actions are considered by default. If needed the user can declare to G4AdjointSimManager |
---|
1345 | stacking, tracking and event actions that will be used only during the adjoint tracking phase. |
---|
1346 | The following lines of code show how to declare these adjoint actions to G4AdjointSimManager: |
---|
1347 | |
---|
1348 | <informalexample><programlisting> |
---|
1349 | G4AdjointSimManager* theAdjointSimManager = G4AdjointSimManager::GetInstance(); |
---|
1350 | theAdjointSimManager->SetAdjointEventAction(aUserDefinedEventAction); |
---|
1351 | theAdjointSimManager->SetAdjointStackingAction(aUserDefinedStackingAction); |
---|
1352 | theAdjointSimManager->SetAdjointTrackingAction(aUserDefinedTrackingAction); |
---|
1353 | </programlisting></informalexample> |
---|
1354 | |
---|
1355 | </para></listitem> |
---|
1356 | |
---|
1357 | </itemizedlist> |
---|
1358 | By default no user run action is considered in an adjoint simulation but if needed such action can be declared to |
---|
1359 | G4AdjointSimManager as such: |
---|
1360 | <informalexample><programlisting> |
---|
1361 | G4AdjointSimManager* theAdjointSimManager = G4AdjointSimManager::GetInstance(); |
---|
1362 | theAdjointSimManager->SetAdjointRunAction(aUserDefinedRunAction); |
---|
1363 | </programlisting></informalexample> |
---|
1364 | |
---|
1365 | </para> |
---|
1366 | |
---|
1367 | </sect4> |
---|
1368 | |
---|
1369 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1370 | <sect4 id="sect.EvtBias.ReverseMC.UserCoding.PhysicsList"> |
---|
1371 | <title> |
---|
1372 | Physics list for reverse and forward electromagnetic processes |
---|
1373 | </title> |
---|
1374 | |
---|
1375 | <para> |
---|
1376 | To run an adjoint simulation a specific physics list should be used where existing G4 adjoint electromagnetic processes |
---|
1377 | and their forward equivalent have to be declared. |
---|
1378 | An example of such physics list is provided by the class G4AdjointPhysicsLits in the G4 example |
---|
1379 | <emphasis role="bold">extended/biasing/ReverseMC01</emphasis>. |
---|
1380 | </para> |
---|
1381 | |
---|
1382 | </sect4> |
---|
1383 | |
---|
1384 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1385 | <sect4 id="sect.EvtBias.ReverseMC.UserCoding.Analysis"> |
---|
1386 | <title> |
---|
1387 | Modification in the analysis part of the code |
---|
1388 | </title> |
---|
1389 | |
---|
1390 | <para> |
---|
1391 | The user code should be modified to normalize the signals computed during the forward tracking phase to the weight of the last adjoint particle |
---|
1392 | that reaches the external surface. |
---|
1393 | This weight represents the statistical weight that the last full adjoint tracks (from the adjoint source |
---|
1394 | to the external source) would have in a forward simulation. If multiplied by a signal and registered in function of energy |
---|
1395 | and/or direction the simulation results will give an answer matrix of this signal. |
---|
1396 | To normalize it to a given spectrum it has to be furthermore multiplied by a directional differential flux |
---|
1397 | corresponding to this spectrum |
---|
1398 | |
---|
1399 | The weight, direction, position , kinetic energy and type of the last adjoint particle that reaches the |
---|
1400 | external source, and that would represents the primary of a forward simulation, can be get from G4AdjointSimManager by using for example the following line of codes |
---|
1401 | <informalexample><programlisting> |
---|
1402 | |
---|
1403 | G4AdjointSimManager* theAdjointSimManager = G4AdjointSimManager::GetInstance(); |
---|
1404 | G4String particle_name = theAdjointSimManager->GetFwdParticleNameAtEndOfLastAdjointTrack(); |
---|
1405 | G4int PDGEncoding= theAdjointSimManager->GetFwdParticlePDGEncodingAtEndOfLastAdjointTrack(); |
---|
1406 | G4double weight = theAdjointSimManager->GetWeightAtEndOfLastAdjointTrack(); |
---|
1407 | G4double Ekin = theAdjointSimManager->GetEkinAtEndOfLastAdjointTrack(); |
---|
1408 | G4double Ekin_per_nuc=theAdjointSimManager->GetEkinNucAtEndOfLastAdjointTrack(); // in case of ions |
---|
1409 | G4ThreeVector dir = theAdjointSimManager->GetDirectionAtEndOfLastAdjointTrack(); |
---|
1410 | G4ThreeVector pos = theAdjointSimManager->GetPositionAtEndOfLastAdjointTrack(); |
---|
1411 | |
---|
1412 | </programlisting></informalexample> |
---|
1413 | |
---|
1414 | </para> |
---|
1415 | |
---|
1416 | <para> |
---|
1417 | In order to have a code working for both forward and adjoint simulation mode, the extra code needed in user actions or analysis |
---|
1418 | manager for the adjoint |
---|
1419 | simulation mode can be separated to the code needed only for the normal forward simulation by using the following public method |
---|
1420 | of G4AdjointSimManager: |
---|
1421 | <informalexample><programlisting> |
---|
1422 | G4bool GetAdjointSimMode(); |
---|
1423 | </programlisting></informalexample> |
---|
1424 | that returns true if an adjoint simulation is running and false if not. |
---|
1425 | |
---|
1426 | </para> |
---|
1427 | |
---|
1428 | <para> |
---|
1429 | The following code example shows how to normalize a detector signal and compute an answer matrix |
---|
1430 | in the case of an adjoint simulation. |
---|
1431 | <example id="sect.EvtBias.ReverseMC.UserCoding.Analysis_1"> |
---|
1432 | <title> |
---|
1433 | Normalization in the case of an adjoint simulation. The detector signal S computed during the forward tracking |
---|
1434 | phase is normalized to a primary source of e- with a differential directional flux given by the function F. |
---|
1435 | An answer matrix of the signal is also computed. |
---|
1436 | </title> |
---|
1437 | |
---|
1438 | <programlisting> |
---|
1439 | |
---|
1440 | G4double S = ...; // signal in the sensitive volume computed during a forward tracking phase |
---|
1441 | |
---|
1442 | //Normalization of the signal for an adjoint simulation |
---|
1443 | G4AdjointSimManager* theAdjSimManager = G4AdjointSimManager::GetInstance(); |
---|
1444 | if (theAdjSimManager->GetAdjointSimMode()) { |
---|
1445 | G4double normalized_S=0.; //normalized to a given e- primary spectrum |
---|
1446 | G4double S_for_answer_matrix=0.; //for e- answer matrix |
---|
1447 | |
---|
1448 | if (theAdjSimManager->GetFwdParticleNameAtEndOfLastAdjointTrack() == "e-"){ |
---|
1449 | G4double ekin_prim = theAdjSimManager->GetEkinAtEndOfLastAdjointTrack(); |
---|
1450 | G4ThreeVector dir_prim = theAdjointSimManager->GetDirectionAtEndOfLastAdjointTrack(); |
---|
1451 | G4double weight_prim = theAdjSimManager->GetWeightAtEndOfLastAdjointTrack(); |
---|
1452 | S_for_answer_matrix = S*weight_prim; |
---|
1453 | normalized_S = S_for_answer_matrix*F(ekin_prim,dir); //F(ekin_prim,dir_prim) gives the differential directional flux of primary e- |
---|
1454 | } |
---|
1455 | //follows the code where normalized_S and S_for_answer_matrix are registered or whatever |
---|
1456 | .... |
---|
1457 | } |
---|
1458 | |
---|
1459 | //analysis/normalization code for forward simulation |
---|
1460 | else { |
---|
1461 | ... |
---|
1462 | } |
---|
1463 | ... |
---|
1464 | </programlisting> |
---|
1465 | </example> |
---|
1466 | |
---|
1467 | </para> |
---|
1468 | |
---|
1469 | </sect4> |
---|
1470 | |
---|
1471 | </sect3> |
---|
1472 | |
---|
1473 | <!-- ******************* Section (Level#3) ****************** --> |
---|
1474 | <sect3 id="sect.EvtBias.ReverseMC.Control"> |
---|
1475 | <title> |
---|
1476 | Control of an adjoint simulation |
---|
1477 | </title> |
---|
1478 | |
---|
1479 | <para> |
---|
1480 | The G4UI commands in the directory |
---|
1481 | <ulink url="./AllResources/Control/UIcommands/_adjoint_.html">/adjoint</ulink>. |
---|
1482 | allow the user to : |
---|
1483 | <itemizedlist spacing="compact"> |
---|
1484 | <listitem><para> |
---|
1485 | Define the adjoint source where adjoint primaries are generated |
---|
1486 | </para></listitem> |
---|
1487 | <listitem><para> |
---|
1488 | Define the external source till which adjoint particles are tracked |
---|
1489 | </para></listitem> |
---|
1490 | <listitem><para> |
---|
1491 | Start an adjoint simulation |
---|
1492 | </para></listitem> |
---|
1493 | </itemizedlist> |
---|
1494 | </para> |
---|
1495 | </sect3> |
---|
1496 | |
---|
1497 | <!-- ******************* Section (Level#3) ****************** --> |
---|
1498 | <sect3 id="sect.EvtBias.ReverseMC.Limitation"> |
---|
1499 | <title> |
---|
1500 | Known issues in the Reverse MC mode |
---|
1501 | </title> |
---|
1502 | |
---|
1503 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1504 | <sect4 id="sect.EvtBias.ReverseMC.Limitation.Weight"> |
---|
1505 | <title> |
---|
1506 | Occasional wrong high weight in the adjoint simulation |
---|
1507 | </title> |
---|
1508 | |
---|
1509 | <para> |
---|
1510 | In rare cases an adjoint track may get a wrong high weight when reaching the external source. |
---|
1511 | While this happens not often it may corrupt the simulation results significantly. This happens |
---|
1512 | in some tracks where both reverse photo-electric and bremsstrahlung processes take place at low energy. |
---|
1513 | We still need some investigations to remove this problem at the level of physical adjoint/reverse processes. |
---|
1514 | However this problem can be solved at the level of event actions or analysis in the user code by adding a test on the |
---|
1515 | normalized signal during an adjoint simulation. An example of such test has been implemented in the Geant4 example |
---|
1516 | <emphasis role="bold"> extended/biasing/ReverseMC01 </emphasis>. |
---|
1517 | In this implementation an event is rejected when the relative error of the computed normalized energy deposited |
---|
1518 | increases during one event by more than 50% while the computed precision is already below 10%. |
---|
1519 | </para> |
---|
1520 | |
---|
1521 | </sect4> |
---|
1522 | |
---|
1523 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1524 | <sect4 id="sect.EvtBias.ReverseMC.Limitation.ReverseBrem"> |
---|
1525 | <title> |
---|
1526 | Reverse bremsstrahlung |
---|
1527 | </title> |
---|
1528 | |
---|
1529 | <para> |
---|
1530 | A difference between the differential cross sections used in the adjoint and forward bremsstrahlung |
---|
1531 | models is the source of a higher flux of >100 keV gamma in the reverse simulation compared to the forward simulation mode. |
---|
1532 | In principle the adjoint processes/models should make use of the direct differential cross section to sample |
---|
1533 | the adjoint secondaries and compute the adjoint cross section. However due to the way the effective differential cross section is |
---|
1534 | considered in the forward model G4eBremsstrahlungModel this was not possible to achieve for the reverse bremsstrahlung. |
---|
1535 | Indeed the differential cross section used in G4AdjointeBremstrahlungModel is obtained by the numerical derivation |
---|
1536 | over the cut energy of the direct cross section provided by G4eBremsstrahlungModel. |
---|
1537 | This would be a correct procedure if the distribution of secondary in G4eBremsstrahlungModel |
---|
1538 | would match this differential cross section. Unfortunately it is not the case as independent parameterization are used |
---|
1539 | in G4eBremsstrahlungModel for both the cross sections and the sampling of secondaries. (It means that in the forward case |
---|
1540 | if one would integrate the effective differential cross section considered in the simulation we would not find back |
---|
1541 | the used cross section). |
---|
1542 | In the future we plan to correct this problem by using an extra weight correction factor after the occurrence of a reverse |
---|
1543 | bremsstrahlung. This weight factor should be the ratio between the differential CS used in the adjoint simulation and the |
---|
1544 | one effectively used in the forward processes. As it is impossible to have a simple and direct access to the forward differential CS |
---|
1545 | in G4eBremsstrahlungModel we are investigating the feasibility to use the differential CS considered in |
---|
1546 | G4Penelope models. |
---|
1547 | </para> |
---|
1548 | |
---|
1549 | </sect4> |
---|
1550 | |
---|
1551 | <!-- ******************* Section (Level#4) ****************** --> |
---|
1552 | <sect4 id="sect.EvtBias.ReverseMC.Limitation.ReverseMS"> |
---|
1553 | <title> |
---|
1554 | Reverse multiple scattering |
---|
1555 | </title> |
---|
1556 | |
---|
1557 | <para> |
---|
1558 | For the reverse multiple scattering the same model is used than in the forward case. |
---|
1559 | This approximation makes that the discrepancy between the adjoint and forward |
---|
1560 | simulation cases can get to a level of ~ 10-15% relative differences in the test cases that we have considered. |
---|
1561 | In the future we plan to improve the adjoint multiple scattering models by forcing the computation of |
---|
1562 | multiple scattering effect at the end of an adjoint step. |
---|
1563 | </para> |
---|
1564 | |
---|
1565 | </sect4> |
---|
1566 | |
---|
1567 | </sect3> |
---|
1568 | </sect2> |
---|
1569 | </sect1> |
---|