ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/group/trunk/tengDissertation/Appendix.tex
(Generate patch)

Comparing trunk/tengDissertation/Appendix.tex (file contents):
Revision 2828 by tim, Thu Jun 8 20:10:36 2006 UTC vs.
Revision 2840 by tim, Fri Jun 9 02:54:01 2006 UTC

# Line 1 | Line 1
1   \appendix
2   \chapter{\label{chapt:oopse}Object-Oriented Parallel Simulation Engine}
3  
4 < Designing object-oriented software is hard, and designing reusable
5 < object-oriented scientific software is even harder. Absence of
6 < applying modern software development practices is the bottleneck of
7 < Scientific Computing community\cite{Wilson2006}. For instance, in
4 > Absence of applying modern software development practices is the
5 > bottleneck of Scientific Computing community\cite{Wilson2006}. In
6   the last 20 years , there are quite a few MD packages that were
7   developed to solve common MD problems and perform robust simulations
8   . However, many of the codes are legacy programs that are either
# Line 64 | Line 62 | as \texttt{StatProps} (see Sec.~\ref{appendixSection:S
62   program of the package, \texttt{oopse} and it corresponding parallel
63   version \texttt{oopse\_MPI}, as well as other useful utilities, such
64   as \texttt{StatProps} (see Sec.~\ref{appendixSection:StaticProps}),
65 < \texttt{DynamicProps} (see
66 < Sec.~\ref{appendixSection:appendixSection:DynamicProps}),
67 < \texttt{Dump2XYZ} (see
70 < Sec.~\ref{appendixSection:appendixSection:Dump2XYZ}), \texttt{Hydro}
71 < (see Sec.~\ref{appendixSection:appendixSection:hydrodynamics})
65 > \texttt{DynamicProps} (see Sec.~\ref{appendixSection:DynamicProps}),
66 > \texttt{Dump2XYZ} (see Sec.~\ref{appendixSection:Dump2XYZ}),
67 > \texttt{Hydro} (see Sec.~\ref{appendixSection:hydrodynamics})
68   \textit{etc}.
69  
70   \begin{figure}
# Line 113 | Line 109 | OOPSE}\cite{Meineke05} and PROTOMOL\cite{Matthey05} \t
109   As one of the latest advanced techniques emerged from
110   object-oriented community, design patterns were applied in some of
111   the modern scientific software applications, such as JMol, {\sc
112 < OOPSE}\cite{Meineke05} and PROTOMOL\cite{Matthey05} \textit{etc}.
113 < The following sections enumerates some of the patterns used in {\sc
114 < OOPSE}.
112 > OOPSE}\cite{Meineke2005} and PROTOMOL\cite{Matthey2005}
113 > \textit{etc}. The following sections enumerates some of the patterns
114 > used in {\sc OOPSE}.
115  
116   \subsection{\label{appendixSection:singleton}Singleton}
117 +
118   The Singleton pattern not only provides a mechanism to restrict
119   instantiation of a class to one object, but also provides a global
120   point of access to the object. Currently implemented as a global
# Line 127 | Line 124 | static data approach in {\sc OOPSE}. {\tt IntegratorFa
124   pollution.Although the singleton pattern can be implemented in
125   various ways  to account for different aspects of the software
126   designs, such as lifespan control \textit{etc}, we only use the
127 < static data approach in {\sc OOPSE}. {\tt IntegratorFactory} class
128 < is declared as
132 < \begin{lstlisting}[float,caption={[A classic Singleton design pattern implementation(I)] Declaration of {\tt IntegratorFactory} class.},label={appendixScheme:singletonDeclaration}]
127 > static data approach in {\sc OOPSE}. IntegratorFactory class is
128 > declared as
129  
130 + \begin{lstlisting}[float,caption={[A classic Singleton design pattern implementation(I)] The declaration of of simple Singleton pattern.},label={appendixScheme:singletonDeclaration}]
131 +
132   class IntegratorFactory {
133 <  public:
134 <    static IntegratorFactory* getInstance();
135 <    protected:
136 <      IntegratorFactory();
137 <    private:
138 <      static IntegratorFactory* instance_;
133 > public:
134 >  static IntegratorFactory*
135 >  getInstance();
136 > protected:
137 >  IntegratorFactory();
138 > private:
139 >  static IntegratorFactory* instance_;
140   };
141  
142   \end{lstlisting}
143 +
144   The corresponding implementation is
145 \begin{lstlisting}[float,caption={[A classic implementation of Singleton design pattern (II)] Implementation of {\tt IntegratorFactory} class.},label={appendixScheme:singletonImplementation}]
145  
146 + \begin{lstlisting}[float,caption={[A classic implementation of Singleton design pattern (II)] The implementation of simple Singleton pattern.},label={appendixScheme:singletonImplementation}]
147 +
148   IntegratorFactory::instance_ = NULL;
149  
150   IntegratorFactory* getInstance() {
# Line 154 | Line 155 | Since constructor is declared as {\tt protected}, a cl
155   }
156  
157   \end{lstlisting}
157 Since constructor is declared as {\tt protected}, a client can not
158 instantiate {\tt IntegratorFactory} directly. Moreover, since the
159 member function {\tt getInstance} serves as the only entry of access
160 to {\tt IntegratorFactory}, this approach fulfills the basic
161 requirement, a single instance. Another consequence of this approach
162 is the automatic destruction since static data are destroyed upon
163 program termination.
158  
159 + Since constructor is declared as protected, a client can not
160 + instantiate IntegratorFactory directly. Moreover, since the member
161 + function getInstance serves as the only entry of access to
162 + IntegratorFactory, this approach fulfills the basic requirement, a
163 + single instance. Another consequence of this approach is the
164 + automatic destruction since static data are destroyed upon program
165 + termination.
166 +
167   \subsection{\label{appendixSection:factoryMethod}Factory Method}
168  
169   Categoried as a creational pattern, the Factory Method pattern deals
170   with the problem of creating objects without specifying the exact
171   class of object that will be created. Factory Method is typically
172   implemented by delegating the creation operation to the subclasses.
173 + Parameterized Factory pattern where factory method (
174 + createIntegrator member function) creates products based on the
175 + identifier (see List.~\ref{appendixScheme:factoryDeclaration}). If
176 + the identifier has been already registered, the factory method will
177 + invoke the corresponding creator (see List.~\ref{integratorCreator})
178 + which utilizes the modern C++ template technique to avoid excess
179 + subclassing.
180  
181 < Registers a creator with a type identifier. Looks up the type
173 < identifier in the internal map. If it is found, it invokes the
174 < corresponding creator for the type identifier and returns its
175 < result.
176 < \begin{lstlisting}[float,caption={[The implementation of Factory pattern (I)].},label={appendixScheme:factoryDeclaration}]
181 > \begin{lstlisting}[float,caption={[The implementation of Parameterized Factory pattern (I)]Source code of IntegratorFactory class.},label={appendixScheme:factoryDeclaration}]
182  
183   class IntegratorFactory {
184 <  public:
185 <    typedef std::map<string, IntegratorCreator*> CreatorMapType;
184 > public:
185 >  typedef std::map<string, IntegratorCreator*> CreatorMapType;
186  
187 <    bool registerIntegrator(IntegratorCreator* creator);
187 >  bool registerIntegrator(IntegratorCreator* creator) {
188 >    return creatorMap_.insert(creator->getIdent(), creator).second;
189 >  }
190  
191 <    Integrator* createIntegrator(const string& id, SimInfo* info);
191 >  Integrator* createIntegrator(const string& id, SimInfo* info) {
192 >    Integrator* result = NULL;
193 >    CreatorMapType::iterator i = creatorMap_.find(id);
194 >    if (i != creatorMap_.end()) {
195 >      result = (i->second)->create(info);
196 >    }
197 >    return result;
198 >  }
199  
200 <  private:
201 <    CreatorMapType creatorMap_;
200 > private:
201 >  CreatorMapType creatorMap_;
202   };
189
203   \end{lstlisting}
204  
205 < \begin{lstlisting}[float,caption={[The implementation of Factory pattern (II)].},label={appendixScheme:factoryDeclarationImplementation}]
205 > \begin{lstlisting}[float,caption={[The implementation of Parameterized Factory pattern (III)]Source code of creator classes.},label={appendixScheme:integratorCreator}]
206  
194 bool IntegratorFactory::unregisterIntegrator(const string& id) {
195  return creatorMap_.erase(id) == 1;
196 }
197
198 Integrator* IntegratorFactory::createIntegrator(const string& id,
199                                                SimInfo* info) {
200  CreatorMapType::iterator i = creatorMap_.find(id);
201  if (i != creatorMap_.end()) {
202    return (i->second)->create(info);
203  } else {
204    return NULL;
205  }
206 }
207
208 \end{lstlisting}
209
210 \begin{lstlisting}[float,caption={[The implementation of Factory pattern (III)].},label={appendixScheme:integratorCreator}]
211
207   class IntegratorCreator {
208 <  public:
208 > public:
209      IntegratorCreator(const string& ident) : ident_(ident) {}
210  
211      const string& getIdent() const { return ident_; }
212  
213      virtual Integrator* create(SimInfo* info) const = 0;
214  
215 <  private:
215 > private:
216      string ident_;
217   };
218  
219   template<class ConcreteIntegrator>
220   class IntegratorBuilder : public IntegratorCreator {
221 <  public:
222 <    IntegratorBuilder(const string& ident) : IntegratorCreator(ident) {}
223 <    virtual  Integrator* create(SimInfo* info) const {
224 <      return new ConcreteIntegrator(info);
225 <    }
221 > public:
222 >  IntegratorBuilder(const string& ident)
223 >                   : IntegratorCreator(ident) {}
224 >  virtual  Integrator* create(SimInfo* info) const {
225 >    return new ConcreteIntegrator(info);
226 >  }
227   };
228   \end{lstlisting}
229  
230   \subsection{\label{appendixSection:visitorPattern}Visitor}
231  
232 < The purpose of the Visitor Pattern is to encapsulate an operation
233 < that you want to perform on the elements. The operation being
234 < performed on a structure can be switched without changing the
235 < interfaces  of the elements. In other words, one can add virtual
236 < functions into a set of classes without modifying their interfaces.
237 < The UML class diagram of Visitor patten is shown in
238 < Fig.~\ref{appendixFig:visitorUML}. {\tt Dump2XYZ} program in
239 < Sec.~\ref{appendixSection:Dump2XYZ} uses Visitor pattern
240 < extensively.
232 > The visitor pattern is designed to decouple the data structure and
233 > algorithms used upon them by collecting related operation from
234 > element classes into other visitor classes, which is equivalent to
235 > adding virtual functions into a set of classes without modifying
236 > their interfaces. Fig.~\ref{appendixFig:visitorUML} demonstrates the
237 > structure of Visitor pattern which is used extensively in {\tt
238 > Dump2XYZ}. In order to convert an OOPSE dump file, a series of
239 > distinct operations are performed on different StuntDoubles (See the
240 > class hierarchy in Fig.~\ref{oopseFig:hierarchy} and the declaration
241 > in List.~\ref{appendixScheme:element}). Since the hierarchies
242 > remains stable, it is easy to define a visit operation (see
243 > List.~\ref{appendixScheme:visitor}) for each class of StuntDouble.
244 > Note that using Composite pattern\cite{Gamma1994}, CompositVisitor
245 > manages a priority visitor list and handles the execution of every
246 > visitor in the priority list on different StuntDoubles.
247  
248   \begin{figure}
249   \centering
250   \includegraphics[width=\linewidth]{visitor.eps}
251 < \caption[The architecture of {\sc OOPSE}] {Overview of the structure
252 < of {\sc OOPSE}} \label{appendixFig:visitorUML}
251 > \caption[The UML class diagram of Visitor patten] {The UML class
252 > diagram of Visitor patten.} \label{appendixFig:visitorUML}
253   \end{figure}
254  
255 < \begin{lstlisting}[float,caption={[The implementation of Visitor pattern (I)]Source code of the visitor classes.},label={appendixScheme:visitor}]
255 > \begin{figure}
256 > \centering
257 > \includegraphics[width=\linewidth]{hierarchy.eps}
258 > \caption[Class hierarchy for ojects in {\sc OOPSE}]{ A diagram of
259 > the class hierarchy. } \label{oopseFig:hierarchy}
260 > \end{figure}
261  
262 < class BaseVisitor{
263 <  public:
264 <    virtual void visit(Atom* atom);
265 <    virtual void visit(DirectionalAtom* datom);
259 <    virtual void visit(RigidBody* rb);
262 > \begin{lstlisting}[float,caption={[The implementation of Visitor pattern (II)]Source code of the element classes.},label={appendixScheme:element}]
263 >
264 > class StuntDouble { public:
265 >  virtual void accept(BaseVisitor* v) = 0;
266   };
267  
268 + class Atom: public StuntDouble { public:
269 +  virtual void accept{BaseVisitor* v*} {
270 +    v->visit(this);
271 +  }
272 + };
273 +
274 + class DirectionalAtom: public Atom { public:
275 +  virtual void accept{BaseVisitor* v*} {
276 +    v->visit(this);
277 +  }
278 + };
279 +
280 + class RigidBody: public StuntDouble { public:
281 +  virtual void accept{BaseVisitor* v*} {
282 +    v->visit(this);
283 +  }
284 + };
285 +
286   \end{lstlisting}
287  
288 < \begin{lstlisting}[float,caption={[The implementation of Visitor pattern (II)]Source code of the element classes.},label={appendixScheme:element}]
288 > \begin{lstlisting}[float,caption={[The implementation of Visitor pattern (I)]Source code of the visitor classes.},label={appendixScheme:visitor}]
289  
290 < class StuntDouble {
291 <  public:
292 <    virtual void accept(BaseVisitor* v) = 0;
290 > class BaseVisitor{
291 > public:
292 >  virtual void visit(Atom* atom);
293 >  virtual void visit(DirectionalAtom* datom);
294 >  virtual void visit(RigidBody* rb);
295   };
296  
297 < class Atom: public StuntDouble {
298 <  public:
299 <    virtual void accept{BaseVisitor* v*} {
300 <      v->visit(this);
275 <    }
297 > class BaseAtomVisitor:public BaseVisitor{ public:
298 >  virtual void visit(Atom* atom);
299 >  virtual void visit(DirectionalAtom* datom);
300 >  virtual void visit(RigidBody* rb);
301   };
302  
303 < class DirectionalAtom: public Atom {
304 <  public:
305 <    virtual void accept{BaseVisitor* v*} {
306 <      v->visit(this);
282 <    }
303 > class SSDAtomVisitor:public BaseAtomVisitor{ public:
304 >  virtual void visit(Atom* atom);
305 >  virtual void visit(DirectionalAtom* datom);
306 >  virtual void visit(RigidBody* rb);
307   };
308  
309 < class RigidBody: public StuntDouble {
310 <  public:
311 <    virtual void accept{BaseVisitor* v*} {
312 <      v->visit(this);
309 > class CompositeVisitor: public BaseVisitor {
310 > public:
311 >
312 >  typedef list<pair<BaseVisitor*, int> > VistorListType;
313 >  typedef VistorListType::iterator VisitorListIterator;
314 >  virtual void visit(Atom* atom) {
315 >    VisitorListIterator i;
316 >    BaseVisitor* curVisitor;
317 >    for(i = visitorList.begin();i != visitorList.end();++i) {
318 >      atom->accept(*i);
319 >    }
320 >  }
321 >
322 >  virtual void visit(DirectionalAtom* datom) {
323 >    VisitorListIterator i;
324 >    BaseVisitor* curVisitor;
325 >    for(i = visitorList.begin();i != visitorList.end();++i) {
326 >      atom->accept(*i);
327      }
328 +  }
329 +
330 +  virtual void visit(RigidBody* rb) {
331 +    VisitorListIterator i;
332 +    std::vector<Atom*> myAtoms;
333 +    std::vector<Atom*>::iterator ai;
334 +    myAtoms = rb->getAtoms();
335 +    for(i = visitorList.begin();i != visitorList.end();++i) {{
336 +      rb->accept(*i);
337 +      for(ai = myAtoms.begin(); ai != myAtoms.end(); ++ai){
338 +        (*ai)->accept(*i);
339 +    }
340 +  }
341 +
342 +  void addVisitor(BaseVisitor* v, int priority);
343 +
344 +  protected:
345 +    VistorListType visitorList;
346   };
347  
348   \end{lstlisting}
349 +
350   \section{\label{appendixSection:concepts}Concepts}
351  
295 \begin{figure}
296 \centering
297 \includegraphics[width=\linewidth]{heirarchy.eps}
298 \caption[Class heirarchy for StuntDoubles in {\sc OOPSE}]{ The class
299 heirarchy of StuntDoubles in {\sc OOPSE}.}
300 \label{oopseFig:heirarchy}
301 \end{figure}
302
352   OOPSE manipulates both traditional atoms as well as some objects
353   that {\it behave like atoms}.  These objects can be rigid
354   collections of atoms or atoms which have orientational degrees of
355 < freedom.  A diagram of the class heirarchy is illustrated in
356 < Fig.~\ref{oopseFig:heirarchy}.
357 <
358 <
355 > freedom.  A diagram of the class hierarchy is illustrated in
356 > Fig.~\ref{oopseFig:hierarchy}. Every Molecule, Atom and
357 > DirectionalAtom in {\sc OOPSE} have their own names which are
358 > specified in the {\tt .md} file. In contrast, RigidBodies are
359 > denoted by their membership and index inside a particular molecule:
360 > [MoleculeName]\_RB\_[index] (the contents inside the brackets depend
361 > on the specifics of the simulation). The names of rigid bodies are
362 > generated automatically. For example, the name of the first rigid
363 > body in a DMPC molecule is DMPC\_RB\_0.
364   \begin{itemize}
365   \item A {\bf StuntDouble} is {\it any} object that can be manipulated by the
366   integrators and minimizers.
# Line 316 | Line 370 | Every Molecule, Atom and DirectionalAtom in {\sc OOPSE
370   DirectionalAtom}s which behaves as a single unit.
371   \end{itemize}
372  
319 Every Molecule, Atom and DirectionalAtom in {\sc OOPSE} have their
320 own names which are specified in the {\tt .md} file. In contrast,
321 RigidBodies are denoted by their membership and index inside a
322 particular molecule: [MoleculeName]\_RB\_[index] (the contents
323 inside the brackets depend on the specifics of the simulation). The
324 names of rigid bodies are generated automatically. For example, the
325 name of the first rigid body in a DMPC molecule is DMPC\_RB\_0.
326
373   \section{\label{appendixSection:syntax}Syntax of the Select Command}
374  
375 < The most general form of the select command is: {\tt select {\it
376 < expression}}. This expression represents an arbitrary set of
331 < StuntDoubles (Atoms or RigidBodies) in {\sc OOPSE}. Expressions are
332 < composed of either name expressions, index expressions, predefined
333 < sets, user-defined expressions, comparison operators, within
334 < expressions, or logical combinations of the above expression types.
335 < Expressions can be combined using parentheses and the Boolean
336 < operators.
375 > {\sc OOPSE} provides a powerful selection utility to select
376 > StuntDoubles. The most general form of the select command is:
377  
378 + {\tt select {\it expression}}.
379 +
380 + This expression represents an arbitrary set of StuntDoubles (Atoms
381 + or RigidBodies) in {\sc OOPSE}. Expressions are composed of either
382 + name expressions, index expressions, predefined sets, user-defined
383 + expressions, comparison operators, within expressions, or logical
384 + combinations of the above expression types. Expressions can be
385 + combined using parentheses and the Boolean operators.
386 +
387   \subsection{\label{appendixSection:logical}Logical expressions}
388  
389   The logical operators allow complex queries to be constructed out of

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines