2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2011,2012,2013, by the GROMACS development team, led by
5 * David van der Spoel, Berk Hess, Erik Lindahl, and including many
6 * others, as listed in the AUTHORS file in the top-level source
7 * directory and at http://www.gromacs.org.
9 * GROMACS is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public License
11 * as published by the Free Software Foundation; either version 2.1
12 * of the License, or (at your option) any later version.
14 * GROMACS is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with GROMACS; if not, see
21 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
22 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
24 * If you want to redistribute modifications to GROMACS, please
25 * consider that scientific software is very special. Version
26 * control is crucial - bugs must be traceable. We will be happy to
27 * consider code for inclusion in the official distribution, but
28 * derived work must not be called official GROMACS. Details are found
29 * in the README & COPYING files - if they are missing, get the
30 * official version at http://www.gromacs.org.
32 * To help us fund GROMACS development, we humbly ask that you cite
33 * the research papers on the package. Check out http://www.gromacs.org.
35 /*! \defgroup module_trajectoryanalysis Framework for Trajectory Analysis (trajectoryanalysis)
36 * \ingroup group_analysismodules
38 * Provides functionality for implementing trajectory analysis modules.
40 * This module implements a framework for implementing flexible trajectory
41 * analysis routines. It provides a base class for implementing analysis as
42 * reusable modules that can be used from different contexts and can also
43 * support per-frame parallelization. It integrally uses functionality from the
45 * - \ref module_options
46 * - \ref module_analysisdata
47 * - \ref module_selection
49 * The main interface of this module is the gmx::TrajectoryAnalysisModule class.
50 * Analysis modules should derive from this class, and override the necessary
51 * virtual methods to provide the actual initialization and analysis routines.
52 * Classes gmx::TrajectoryAnalysisSettings and gmx::TopologyInformation (in
53 * addition to classes declared in the above-mentioned modules) are used to pass
54 * information to and from these methods. gmx::TrajectoryAnalysisModuleData can
55 * be used in advanced scenarios where the tool requires custom thread-local
56 * data for parallel analysis.
58 * The sequence charts below provides an overview of how the trajectory
59 * analysis modules typically interact with other components.
60 * The first chart provides an overview of the call sequence of the most
61 * important methods in gmx::TrajectoryAnalysisModule.
62 * There is a runner, which is responsible for doing the work that is shared
63 * between all trajectory analysis (such as reading the trajectory and
64 * processing selections). The runner then calls different methods in the
65 * analysis module at appropriate points to perform the module-specific tasks.
66 * The analysis module is responsible for creating and managing
67 * gmx::AnalysisData objects, and the chart shows the most important
68 * interactions with this module as well. Interactions with options (for
69 * command-line option processing) and selections is not shown for brevity: see
70 * \ref module_options for an overview of how options work, and the second
71 * chart for a more detailed view of how selections are accessed from an
75 * module [ URL="\ref gmx::TrajectoryAnalysisModule" ],
76 * data [ label="analysis data", URL="\ref module_analysisdata" ];
78 * runner box module [ label="caller owns runner and module objects" ];
79 * module => data [ label="create (in constructor)" ];
80 * runner => module [ label="initOptions()",
81 * URL="\ref gmx::TrajectoryAnalysisModule::initOptions()" ];
82 * runner => runner [ label="parse user input" ];
83 * runner => module [ label="optionsFinished()",
84 * URL="\ref gmx::TrajectoryAnalysisModule::optionsFinished()" ];
85 * runner => runner [ label="initialize topology\nand selections" ];
86 * runner => module [ label="initAnalysis()",
87 * URL="\ref gmx::TrajectoryAnalysisModule::initAnalysis()" ];
88 * module => data [ label="initialize" ];
89 * runner => runner [ label="read frame 0" ];
90 * runner => module [ label="initAfterFirstFrame()",
91 * URL="\ref gmx::TrajectoryAnalysisModule::initAfterFirstFrame()" ];
92 * --- [ label="loop over frames starts" ];
93 * runner => runner [ label="initialize frame 0" ];
94 * runner => module [ label="analyzeFrame(0)",
95 * URL="\ref gmx::TrajectoryAnalysisModule::analyzeFrame()" ];
96 * module => data [ label="add data",
97 * URL="\ref gmx::AnalysisDataHandle" ];
98 * module => data [ label="finishFrame()",
99 * URL="\ref gmx::AnalysisDataHandle::finishFrame()" ];
100 * data => data [ label="process frame 0" ];
101 * runner => runner [ label="read and initialize frame 1" ];
102 * runner => module [ label="analyzeFrame(1)",
103 * URL="\ref gmx::TrajectoryAnalysisModule::analyzeFrame()" ];
105 * --- [ label="loop over frames ends" ];
106 * runner => module [ label="finishAnalysis()",
107 * URL="\ref gmx::TrajectoryAnalysisModule::finishAnalysis()" ];
108 * module => data [ label="post-process data" ];
109 * runner => module [ label="writeOutput()",
110 * URL="\ref gmx::TrajectoryAnalysisModule::writeOutput()" ];
113 * The second chart below shows the interaction with selections and options
114 * with focus on selection options. The gmx::TrajectoryAnalysisModule object
115 * creates one or more gmx::Selection variables, and uses gmx::SelectionOption
116 * to indicate them as the destination for selections. This happens in
117 * gmx::TrajectoryAnalysisModule::initOptions(). After the options have been
118 * parsed (includes parsing any options present on the command-line or read
119 * from files, but not those provided interactively),
120 * gmx::TrajectoryAnalysisModule::optionsFinished() can adjust the selections
121 * using gmx::SelectionOptionInfo. This is done like this to allow the
122 * analysis module to influence the interactive prompt of selections based on
123 * what command-line options were given. After optionsFinished() returns, the
124 * interactive selection prompt is presented if necessary. After this point,
125 * all access to selections from the analysis module is through the
126 * gmx::Selection variables: the runner is responsible for calling methods in
127 * the selection library, and these methods update the content referenced by
128 * the gmx::Selection variables. See documentation of
129 * gmx::TrajectoryAnalysisModule for details of what the selections contain at
133 * options [ label="Options", URL="\ref module_options" ],
134 * selection [ label="selections", URL="\ref module_selection" ],
135 * module [ label="module", URL="\ref gmx::TrajectoryAnalysisModule" ];
137 * runner box selection [ label="all these objects are owned by the framework" ];
138 * runner => module [ label="initOptions()",
139 * URL="\ref gmx::TrajectoryAnalysisModule::initOptions()" ];
140 * module => options [ label="addOption(SelectionOption)",
141 * URL="\ref gmx::SelectionOption" ];
142 * module => options [ label="addOption() (other options)",
143 * URL="\ref gmx::Options::addOption()" ];
146 * runner => options [ label="parse command-line parameters" ];
147 * options => selection [ label="parse selections" ];
148 * selection -> module [ label="initialize Selection variables",
149 * URL="\ref gmx::Selection" ];
151 * runner => module [ label="optionsFinished()",
152 * URL="\ref gmx::TrajectoryAnalysisModule::optionsFinished()" ];
153 * module => selection [ label="adjust SelectionOptions",
154 * URL="\ref gmx::SelectionOptionInfo" ];
156 * runner => selection [ label="prompt missing selections" ];
157 * selection -> module [ label="initialize Selection variables",
158 * URL="\ref gmx::Selection" ];
159 * runner => selection [ label="compile selections" ];
160 * selection -> module [ label="change content referenced\nby Selection variables" ];
161 * runner => module [ label="initAnalysis()",
162 * URL="\ref gmx::TrajectoryAnalysisModule::initAnalysis()" ];
164 * --- [ label="loop over frames starts" ];
165 * runner => runner [ label="read and initialize frame 0" ];
166 * runner => selection [ label="evaluate selections for frame 0" ];
167 * selection -> module [ label="change content referenced\nby Selection variables" ];
171 * The final chart shows the flow within the frame loop in the case of parallel
172 * (threaded) execution and the interaction with the \ref module_analysisdata
173 * module in this case. Although parallelization has not yet been implemented,
174 * it has influenced the design and needs to be understood if one wants to
175 * write modules that can take advantage of the parallelization once it gets
176 * implemented. The parallelization takes part over frames: analyzing a single
177 * frame is one unit of work. When the frame loop is started,
178 * gmx::TrajectoryAnalysisModule::startFrames() is called for each thread, and
179 * initializes an object that contains thread-local data needed during the
180 * analysis. This includes selection information, gmx::AnalysisDataHandle
181 * objects, and possibly other module-specific variables. Then, the runner
182 * reads the frames in sequence and passes the work into the different threads,
183 * together with the appropriate thread-local data object.
184 * The gmx::TrajectoryAnalysisModule::analyzeFrame() calls are only allowed to modify
185 * the thread-local data object; everything else is read-only. For any output,
186 * they pass the information to gmx::AnalysisData, which takes care of ordering
187 * the data from different frames such that it gets processed in the right
189 * When all frames are analyzed, gmx::TrajectoryAnalysisModule::finishFrames()
190 * is called for each thread-local data object to destroy them and to
191 * accumulate possible results from them into the main
192 * gmx::TrajectoryAnalysisModule object.
193 * Note that in the diagram, some part of the work attributed for the runner
194 * (e.g., evaluating selections) will actually be carried out in the analysis
195 * threads before gmx::TrajectoryAnalysisModule::analyzeFrame() gets called.
198 * module [ label="module object" ],
199 * thread1 [ label="analysis\nthread 1" ],
200 * thread2 [ label="analysis\nthread 2" ],
201 * data [ label="analysis data", URL="\ref module_analysisdata" ];
203 * module box thread2 [ label="single TrajectoryAnalysisModule object",
204 * URL="\ref gmx::TrajectoryAnalysisModule" ];
206 * --- [ label="loop over frames starts" ];
207 * runner => thread1 [ label="startFrames()",
208 * URL="\ref gmx::TrajectoryAnalysisModule::startFrames()" ];
209 * thread1 => data [ label="startData()",
210 * URL="\ref gmx::AnalysisData::startData()" ];
211 * runner << thread1 [ label="pdata1" ];
212 * runner => thread2 [ label="startFrames()",
213 * URL="\ref gmx::TrajectoryAnalysisModule::startFrames()" ];
214 * thread2 => data [ label="startData()",
215 * URL="\ref gmx::AnalysisData::startData()" ];
216 * runner << thread2 [ label="pdata2" ];
218 * runner => runner [ label="initialize frame 0" ];
219 * runner => thread1 [ label="analyzeFrame(0, pdata1)",
220 * URL="\ref gmx::TrajectoryAnalysisModule::analyzeFrame()" ];
221 * runner => runner [ label="read and initialize frame 1" ];
222 * runner => thread2 [ label="analyzeFrame(1, pdata2)",
223 * URL="\ref gmx::TrajectoryAnalysisModule::analyzeFrame()" ];
224 * thread1 => data [ label="add data",
225 * URL="\ref gmx::AnalysisDataHandle" ];
226 * thread2 => data [ label="add data",
227 * URL="\ref gmx::AnalysisDataHandle" ];
228 * thread2 => data [ label="finishFrame(1)",
229 * URL="\ref gmx::AnalysisDataHandle::finishFrame()" ];
230 * runner << thread2 [ label="analyzeFrame() (frame 1)" ];
231 * runner => runner [ label="read and initialize frame 2" ];
232 * runner => thread2 [ label="analyzeFrame(2)",
233 * URL="\ref gmx::TrajectoryAnalysisModule::analyzeFrame()" ];
234 * thread1 => data [ label="finishFrame(0)",
235 * URL="\ref gmx::AnalysisDataHandle::finishFrame()" ];
236 * data => data [ label="process frame 0" ];
237 * data => data [ label="process frame 1" ];
238 * runner << thread1 [ label="analyzeFrame() (frame 0)" ];
240 * runner => thread1 [ label="finishFrames(pdata1)",
241 * URL="\ref gmx::TrajectoryAnalysisModule::finishFrames()" ];
242 * thread1 => data [ label="finishData()",
243 * URL="\ref gmx::AnalysisData::finishData()" ];
244 * thread1 -> module [ label="accumulate results" ];
246 * runner => thread2 [ label="finishFrames(pdata2)",
247 * URL="\ref gmx::TrajectoryAnalysisModule::finishFrames()" ];
248 * thread2 => data [ label="finishData()",
249 * URL="\ref gmx::AnalysisData::finishData()" ];
250 * thread2 -> module [ label="accumulate results" ];
252 * --- [ label="loop over frames ends" ];
256 * In addition to the framework for defining analysis modules, this module also
257 * provides gmx::TrajectoryAnalysisCommandLineRunner, which implements a
258 * command-line program that runs a certain analysis module.
260 * Internally, the module also defines a set of trajectory analysis modules that
261 * can currently be accessed only through gmx::registerTrajectoryAnalysisModules.
263 * For an example of how to implement an analysis tool using the framework, see
266 * \author Teemu Murtola <teemu.murtola@gmail.com>
270 * Public API convenience header for trajectory analysis framework
272 * \author Teemu Murtola <teemu.murtola@gmail.com>
274 * \ingroup module_trajectoryanalysis
276 #ifndef GMX_TRAJECTORYANALYSIS_H
277 #define GMX_TRAJECTORYANALYSIS_H
279 #include "analysisdata.h"
281 #include "selection.h"
283 #include "selection/nbsearch.h"
284 #include "trajectoryanalysis/analysismodule.h"
285 #include "trajectoryanalysis/analysissettings.h"
286 #include "trajectoryanalysis/cmdlinerunner.h"
287 #include "utility/exceptions.h"
288 #include "utility/programinfo.h"
289 #include "utility/stringutil.h"