1mdrun(1) GROMACS suite, VERSION 4.5 mdrun(1)
2
6 mdrun - performs a simulation, do a normal mode analysis or an energy
7 minimization
8 VERSION 4.5
10
12 mdrun -s topol.tpr -o traj.trr -x traj.xtc -cpi state.cpt -cpo
13 state.cpt -c confout.gro -e ener.edr -g md.log -dhdl dhdl.xvg -field
14 field.xvg -table table.xvg -tablep tablep.xvg -tableb table.xvg -rerun
15 rerun.xtc -tpi tpi.xvg -tpid tpidist.xvg -ei sam.edi -eo sam.edo -j
16 wham.gct -jo bam.gct -ffout gct.xvg -devout deviatie.xvg -runav
17 runaver.xvg -px pullx.xvg -pf pullf.xvg -mtx nm.mtx -dn dipole.ndx
18 -[no]h -[no]version -nice int -deffnm string -xvg enum -[no]pd -dd vec‐
19 tor -nt int -npme int -ddorder enum -[no]ddcheck -rdd real -rcon real
20 -dlb enum -dds real -gcom int -[no]v -[no]compact -[no]seppot -pforce
21 real -[no]reprod -cpt real -[no]cpnum -[no]append -maxh real -multi int
22 -replex int -reseed int -[no]ionize
23
25 The mdrun program is the main computational chemistry engine within
26 GROMACS. Obviously, it performs Molecular Dynamics simulations, but it
27 can also perform Stochastic Dynamics, Energy Minimization, test parti‐
28 cle insertion or (re)calculation of energies. Normal mode analysis is
29 another option. In this case mdrun builds a Hessian matrix from single
30 conformation. For usual Normal Modes-like calculations, make sure that
31 the structure provided is properly energy-minimized. The generated
32 matrix can be diagonalized by g_nmeig.
33 The mdrun program reads the run input file ( -s) and distributes the
36 topology over nodes if needed. mdrun produces at least four output
37 files. A single log file ( -g) is written, unless the option -seppot
38 is used, in which case each node writes a log file. The trajectory
39 file ( -o), contains coordinates, velocities and optionally forces.
40 The structure file ( -c) contains the coordinates and velocities of the
41 last step. The energy file ( -e) contains energies, the temperature,
42 pressure, etc, a lot of these things are also printed in the log file.
43 Optionally coordinates can be written to a compressed trajectory file (
44 -x).
45 The option -dhdl is only used when free energy calculation is turned
48 on.
49 When mdrun is started using MPI with more than 1 node, parallelization
52 is used. By default domain decomposition is used, unless the -pd
53 option is set, which selects particle decomposition.
54 With domain decomposition, the spatial decomposition can be set with
57 option -dd. By default mdrun selects a good decomposition. The user
58 only needs to change this when the system is very inhomogeneous.
59 Dynamic load balancing is set with the option -dlb, which can give a
60 significant performance improvement, especially for inhomogeneous sys‐
61 tems. The only disadvantage of dynamic load balancing is that runs are
62 no longer binary reproducible, but in most cases this is not important.
63 By default the dynamic load balancing is automatically turned on when
64 the measured performance loss due to load imbalance is 5% or more. At
65 low parallelization these are the only important options for domain
66 decomposition. At high parallelization the options in the next two
67 sections could be important for increasing the performace.
68 When PME is used with domain decomposition, separate nodes can be
71 assigned to do only the PME mesh calculation; this is computationally
72 more efficient starting at about 12 nodes. The number of PME nodes is
73 set with option -npme, this can not be more than half of the nodes.
74 By default mdrun makes a guess for the number of PME nodes when the
75 number of nodes is larger than 11 or performance wise not compatible
76 with the PME grid x dimension. But the user should optimize npme. Per‐
77 formance statistics on this issue are written at the end of the log
78 file. For good load balancing at high parallelization, the PME grid x
79 and y dimensions should be divisible by the number of PME nodes (the
80 simulation will run correctly also when this is not the case).
81 This section lists all options that affect the domain decomposition.
84 Option -rdd can be used to set the required maximum distance for inter
86 charge-group bonded interactions. Communication for two-body bonded
87 interactions below the non-bonded cut-off distance always comes for
88 free with the non-bonded communication. Atoms beyond the non-bonded
89 cut-off are only communicated when they have missing bonded interac‐
90 tions; this means that the extra cost is minor and nearly indepedent of
91 the value of -rdd. With dynamic load balancing option -rdd also sets
92 the lower limit for the domain decomposition cell sizes. By default
93 -rdd is determined by mdrun based on the initial coordinates. The cho‐
94 sen value will be a balance between interaction range and communication
95 cost.
96 When inter charge-group bonded interactions are beyond the bonded
98 cut-off distance, mdrun terminates with an error message. For pair
99 interactions and tabulated bonds that do not generate exclusions, this
100 check can be turned off with the option -noddcheck.
101 When constraints are present, option -rcon influences the cell size
103 limit as well. Atoms connected by NC constraints, where NC is the
104 LINCS order plus 1, should not be beyond the smallest cell size. A
105 error message is generated when this happens and the user should change
106 the decomposition or decrease the LINCS order and increase the number
107 of LINCS iterations. By default mdrun estimates the minimum cell size
108 required for P-LINCS in a conservative fashion. For high paralleliza‐
109 tion it can be useful to set the distance required for P-LINCS with the
110 option -rcon.
111 The -dds option sets the minimum allowed x, y and/or z scaling of the
113 cells with dynamic load balancing. mdrun will ensure that the cells can
114 scale down by at least this factor. This option is used for the auto‐
115 mated spatial decomposition (when not using -dd) as well as for deter‐
116 mining the number of grid pulses, which in turn sets the minimum
117 allowed cell size. Under certain circumstances the value of -dds might
118 need to be adjusted to account for high or low spatial inhomogeneity of
119 the system.
120 The option -gcom can be used to only do global communication every n
123 steps. This can improve performance for highly parallel simulations
124 where this global communication step becomes the bottleneck. For a
125 global thermostat and/or barostat the temperature and/or pressure will
126 also only be updated every -gcom steps. By default it is set to the
127 minimum of nstcalcenergy and nstlist.
128 With -rerun an input trajectory can be given for which forces and
131 energies will be (re)calculated. Neighbor searching will be performed
132 for every frame, unless nstlist is zero (see the .mdp file).
133 ED (essential dynamics) sampling is switched on by using the -ei flag
136 followed by an .edi file. The .edi file can be produced using
137 options in the essdyn menu of the WHAT IF program. mdrun produces a
138 .edo file that contains projections of positions, velocities and forces
139 onto selected eigenvectors.
140 When user-defined potential functions have been selected in the .mdp
143 file the -table option is used to pass mdrun a formatted table with
144 potential functions. The file is read from either the current directory
145 or from the GMXLIB directory. A number of pre-formatted tables are
146 presented in the GMXLIB dir, for 6-8, 6-9, 6-10, 6-11, 6-12 Lennard
147 Jones potentials with normal Coulomb. When pair interactions are
148 present a separate table for pair interaction functions is read using
149 the -tablep option.
150 When tabulated bonded functions are present in the topology, interac‐
153 tion functions are read using the -tableb option. For each different
154 tabulated interaction type the table file name is modified in a differ‐
155 ent way: before the file extension an underscore is appended, then a b
156 for bonds, an a for angles or a d for dihedrals and finally the table
157 number of the interaction type.
158 The options -px and -pf are used for writing pull COM coordinates and
161 forces when pulling is selected in the .mdp file.
162 With -multi multiple systems are simulated in parallel. As many input
165 files are required as the number of systems. The system number is
166 appended to the run input and each output filename, for instance
167 topol.tpr becomes topol0.tpr, topol1.tpr etc. The number of nodes per
168 system is the total number of nodes divided by the number of systems.
169 One use of this option is for NMR refinement: when distance or orienta‐
170 tion restraints are present these can be ensemble averaged over all the
171 systems.
172 With -replex replica exchange is attempted every given number of
175 steps. The number of replicas is set with the -multi option, see
176 above. All run input files should use a different coupling tempera‐
177 ture, the order of the files is not important. The random seed is set
178 with -reseed. The velocities are scaled and neighbor searching is per‐
179 formed after every exchange.
180 Finally some experimental algorithms can be tested when the appropriate
183 options have been given. Currently under investigation are: polariz‐
184 ability, and X-Ray bombardments.
185 The option -pforce is useful when you suspect a simulation crashes due
188 to too large forces. With this option coordinates and forces of atoms
189 with a force larger than a certain value will be printed to stderr.
190 Checkpoints containing the complete state of the system are written at
193 regular intervals (option -cpt) to the file -cpo, unless option -cpt
194 is set to -1. The previous checkpoint is backed up to state_prev.cpt
195 to make sure that a recent state of the system is always available,
196 even when the simulation is terminated while writing a checkpoint.
197 With -cpnum all checkpoint files are kept and appended with the step
198 number. A simulation can be continued by reading the full state from
199 file with option -cpi. This option is intelligent in the way that if
200 no checkpoint file is found, Gromacs just assumes a normal run and
201 starts from the first step of the tpr file. By default the output will
202 be appending to the existing output files. The checkpoint file contains
203 checksums of all output files, such that you will never loose data when
204 some output files are modified, corrupt or removed. There are three
205 scenarios with -cpi:
206 * no files with matching names are present: new output files are writ‐
208 ten
209 * all files are present with names and checksums matching those stored
211 in the checkpoint file: files are appended
212 * otherwise no files are modified and a fatal error is generated
214 With -noappend new output files are opened and the simulation part
216 number is added to all output file names. Note that in all cases the
217 checkpoint file itself is not renamed and will be overwritten, unless
218 its name does not match the -cpo option.
219 With checkpointing the output is appended to previously written output
222 files, unless -noappend is used or none of the previous output files
223 are present (except for the checkpoint file). The integrity of the
224 files to be appended is verified using checksums which are stored in
225 the checkpoint file. This ensures that output can not be mixed up or
226 corrupted due to file appending. When only some of the previous output
227 files are present, a fatal error is generated and no old output files
228 are modified and no new output files are opened. The result with
229 appending will be the same as from a single run. The contents will be
230 binary identical, unless you use a different number of nodes or dynamic
231 load balancing or the FFT library uses optimizations through timing.
232 With option -maxh a simulation is terminated and a checkpoint file is
235 written at the first neighbor search step where the run time exceeds
236 -maxh*0.99 hours.
237 When mdrun receives a TERM signal, it will set nsteps to the current
240 step plus one. When mdrun receives an INT signal (e.g. when ctrl+C is
241 pressed), it will stop after the next neighbor search step (with
242 nstlist=0 at the next step). In both cases all the usual output will
243 be written to file. When running with MPI, a signal to one of the
244 mdrun processes is sufficient, this signal should not be sent to mpirun
245 or the mdrun process that is the parent of the others.
246 When mdrun is started with MPI, it does not run niced by default.
249
251 -s topol.tpr Input
252 Run input file: tpr tpb tpa
253 -o traj.trr Output
255 Full precision trajectory: trr trj cpt
256 -x traj.xtc Output, Opt.
258 Compressed trajectory (portable xdr format)
259 -cpi state.cpt Input, Opt.
261 Checkpoint file
262 -cpo state.cpt Output, Opt.
264 Checkpoint file
265 -c confout.gro Output
267 Structure file: gro g96 pdb etc.
268 -e ener.edr Output
270 Energy file
271 -g md.log Output
273 Log file
274 -dhdl dhdl.xvg Output, Opt.
276 xvgr/xmgr file
277 -field field.xvg Output, Opt.
279 xvgr/xmgr file
280 -table table.xvg Input, Opt.
282 xvgr/xmgr file
283 -tablep tablep.xvg Input, Opt.
285 xvgr/xmgr file
286 -tableb table.xvg Input, Opt.
288 xvgr/xmgr file
289 -rerun rerun.xtc Input, Opt.
291 Trajectory: xtc trr trj gro g96 pdb cpt
292 -tpi tpi.xvg Output, Opt.
294 xvgr/xmgr file
295 -tpid tpidist.xvg Output, Opt.
297 xvgr/xmgr file
298 -ei sam.edi Input, Opt.
300 ED sampling input
301 -eo sam.edo Output, Opt.
303 ED sampling output
304 -j wham.gct Input, Opt.
306 General coupling stuff
307 -jo bam.gct Output, Opt.
309 General coupling stuff
310 -ffout gct.xvg Output, Opt.
312 xvgr/xmgr file
313 -devout deviatie.xvg Output, Opt.
315 xvgr/xmgr file
316 -runav runaver.xvg Output, Opt.
318 xvgr/xmgr file
319 -px pullx.xvg Output, Opt.
321 xvgr/xmgr file
322 -pf pullf.xvg Output, Opt.
324 xvgr/xmgr file
325 -mtx nm.mtx Output, Opt.
327 Hessian matrix
328 -dn dipole.ndx Output, Opt.
330 Index file
331
334 -[no]hno
335 Print help info and quit
336 -[no]versionno
338 Print version info and quit
339 -nice int 0
341 Set the nicelevel
342 -deffnm string
344 Set the default filename for all file options
345 -xvg enum xmgrace
347 xvg plot formatting: xmgrace, xmgr or none
348 -[no]pdno
350 Use particle decompostion
351 -dd vector 0 0 0
353 Domain decomposition grid, 0 is optimize
354 -nt int 0
356 Number of threads to start (0 is guess)
357 -npme int -1
359 Number of separate nodes to be used for PME, -1 is guess
360 -ddorder enum interleave
362 DD node order: interleave, pp_pme or cartesian
363 -[no]ddcheckyes
365 Check for all bonded interactions with DD
366 -rdd real 0
368 The maximum distance for bonded interactions with DD (nm), 0 is deter‐
369 mine from initial coordinates
370 -rcon real 0
372 Maximum distance for P-LINCS (nm), 0 is estimate
373 -dlb enum auto
375 Dynamic load balancing (with DD): auto, no or yes
376 -dds real 0.8
378 Minimum allowed dlb scaling of the DD cell size
379 -gcom int -1
381 Global communication frequency
382 -[no]vno
384 Be loud and noisy
385 -[no]compactyes
387 Write a compact log file
388 -[no]seppotno
390 Write separate V and dVdl terms for each interaction type and node to
391 the log file(s)
392 -pforce real -1
394 Print all forces larger than this (kJ/mol nm)
395 -[no]reprodno
397 Try to avoid optimizations that affect binary reproducibility
398 -cpt real 15
400 Checkpoint interval (minutes)
401 -[no]cpnumno
403 Keep and number checkpoint files
404 -[no]appendyes
406 Append to previous output files when continuing from checkpoint
407 instead of adding the simulation part number to all file names
408 -maxh real -1
410 Terminate after 0.99 times this time (hours)
411 -multi int 0
413 Do multiple simulations in parallel
414 -replex int 0
416 Attempt replica exchange every steps
417 -reseed int -1
419 Seed for replica exchange, -1 is generate a seed
420 -[no]ionizeno
422 Do a simulation including the effect of an X-Ray bombardment on your
423 system
424
427 gromacs(7)
428 More information about GROMACS is available at <http://www.gro‐
430 macs.org/>.
431 Thu 26 Aug 2010 mdrun(1)