Molecular dynamics parameters (.mdp options)¶
General information¶
Default values are given in parentheses, or listed first among choices. The first option in the list is always the default option. Units are given in square brackets. The difference between a dash and an underscore is ignored.
A sample mdp file is available. This should be appropriate to start a normal simulation. Edit it to suit your specific needs and desires.
Preprocessing¶

include
¶ directories to include in your topology. Format:
I/home/john/mylib I../otherlib

define
¶ defines to pass to the preprocessor, default is no defines. You can use any defines to control options in your customized topology files. Options that act on existing top file mechanisms include
DFLEXIBLE
will use flexible water instead of rigid water into your topology, this can be useful for normal mode analysis.DPOSRES
will trigger the inclusion ofposre.itp
into your topology, used for implementing position restraints.
Run control¶

integrator
¶ (Despite the name, this list includes algorithms that are not actually integrators over time.
integrator=steep
and all entries following it are in this category)
md
¶ A leapfrog algorithm for integrating Newton’s equations of motion.

mdvv
¶ A velocity Verlet algorithm for integrating Newton’s equations of motion. For constant NVE simulations started from corresponding points in the same trajectory, the trajectories are analytically, but not binary, identical to the
integrator=md
leapfrog integrator. The kinetic energy, which is determined from the whole step velocities and is therefore slightly too high. The advantage of this integrator is more accurate, reversible NoseHoover and ParrinelloRahman coupling integration based on Trotter expansion, as well as (slightly too small) full step velocity output. This all comes at the cost off extra computation, especially with constraints and extra communication in parallel. Note that for nearly all production simulations theintegrator=md
integrator is accurate enough.

mdvvavek
¶ A velocity Verlet algorithm identical to
integrator=mdvv
, except that the kinetic energy is determined as the average of the two half step kinetic energies as in theintegrator=md
integrator, and this thus more accurate. With NoseHoover and/or ParrinelloRahman coupling this comes with a slight increase in computational cost.

sd
¶ An accurate and efficient leapfrog stochastic dynamics integrator. With constraints, coordinates needs to be constrained twice per integration step. Depending on the computational cost of the force calculation, this can take a significant part of the simulation time. The temperature for one or more groups of atoms (
tcgrps
) is set withreft
, the inverse friction constant for each group is set withtaut
. The parameterstcoupl
andnsttcouple
are ignored. The random generator is initialized withldseed
. When used as a thermostat, an appropriate value fortaut
is 2 ps, since this results in a friction that is lower than the internal friction of water, while it is high enough to remove excess heat NOTE: temperature deviations decay twice as fast as with a Berendsen thermostat with the sametaut
.

bd
¶ An Euler integrator for Brownian or position Langevin dynamics, the velocity is the force divided by a friction coefficient (
bdfric
) plus random thermal noise (reft
). Whenbdfric
is 0, the friction coefficient for each particle is calculated as mass/taut
, as for the integratorintegrator=sd
. The random generator is initialized withldseed
.

steep
¶ A steepest descent algorithm for energy minimization. The maximum step size is
emstep
, the tolerance isemtol
.

cg
¶ A conjugate gradient algorithm for energy minimization, the tolerance is
emtol
. CG is more efficient when a steepest descent step is done every once in a while, this is determined bynstcgsteep
. For a minimization prior to a normal mode analysis, which requires a very high accuracy, GROMACS should be compiled in double precision.

lbfgs
¶ A quasiNewtonian algorithm for energy minimization according to the lowmemory BroydenFletcherGoldfarbShanno approach. In practice this seems to converge faster than Conjugate Gradients, but due to the correction steps necessary it is not (yet) parallelized.

nm
¶ Normal mode analysis is performed on the structure in the tpr file. GROMACS should be compiled in double precision.

tpi
¶ Test particle insertion. The last molecule in the topology is the test particle. A trajectory must be provided to
mdrun rerun
. This trajectory should not contain the molecule to be inserted. Insertions are performednsteps
times in each frame at random locations and with random orientiations of the molecule. Whennstlist
is larger than one,nstlist
insertions are performed in a sphere with radiusrtpi
around a the same random location using the same pair list. Since pair list construction is expensive, one can perform several extra insertions with the same list almost for free. The random seed is set withldseed
. The temperature for the Boltzmann weighting is set withreft
, this should match the temperature of the simulation of the original trajectory. Dispersion correction is implemented correctly for TPI. All relevant quantities are written to the file specified withmdrun tpi
. The distribution of insertion energies is written to the file specified withmdrun tpid
. No trajectory or energy file is written. Parallel TPI gives identical results to singlenode TPI. For charged molecules, using PME with a fine grid is most accurate and also efficient, since the potential in the system only needs to be calculated once per frame.

tpic
¶ Test particle insertion into a predefined cavity location. The procedure is the same as for
integrator=tpi
, except that one coordinate extra is read from the trajectory, which is used as the insertion location. The molecule to be inserted should be centered at 0,0,0. GROMACS does not do this for you, since for different situations a different way of centering might be optimal. Alsortpi
sets the radius for the sphere around this location. Neighbor searching is done only once per frame,nstlist
is not used. Parallelintegrator=tpic
gives identical results to singlerankintegrator=tpic
.

mimic
¶ Enable MiMiC QM/MM coupling to run hybrid molecular dynamics. Keey in mind that its required to launch CPMD compiled with MiMiC as well. In this mode all options regarding integration (Tcoupling, Pcoupling, timestep and number of steps) are ignored as CPMD will do the integration instead. Options related to forces computation (cutoffs, PME parameters, etc.) are working as usual. Atom selection to define QM atoms is read from
QMMMgrps


tinit
¶ (0) [ps] starting time for your run (only makes sense for timebased integrators)

dt
¶ (0.001) [ps] time step for integration (only makes sense for timebased integrators)

nsteps
¶ (0) maximum number of steps to integrate or minimize, 1 is no maximum

initstep
¶ (0) The starting step. The time at step i in a run is calculated as: t =
tinit
+dt
* (initstep
+ i). The freeenergy lambda is calculated as: lambda =initlambda
+deltalambda
* (initstep
+ i). Also nonequilibrium MD parameters can depend on the step number. Thus for exact restarts or redoing part of a run it might be necessary to setinitstep
to the step number of the restart frame. gmx converttpr does this automatically.

simulationpart
¶ (0) A simulation can consist of multiple parts, each of which has a part number. This option specifies what that number will be, which helps keep track of parts that are logically the same simulation. This option is generally useful to set only when coping with a crashed simulation where files were lost.

commmode
¶ 
Linear
¶ Remove center of mass translational velocity

Angular
¶ Remove center of mass translational and rotational velocity

Linearaccelerationcorrection
¶ Remove center of mass translational velocity. Correct the center of mass position assuming linear acceleration over
nstcomm
steps. This is useful for cases where an acceleration is expected on the center of mass which is nearly constant overnstcomm
steps. This can occur for example when pulling on a group using an absolute reference.

None
¶ No restriction on the center of mass motion


nstcomm
¶ (100) [steps] frequency for center of mass motion removal

commgrps
¶ group(s) for center of mass motion removal, default is the whole system
Langevin dynamics¶
Energy minimization¶

emtol
¶ (10.0) [kJ mol^{1} nm^{1}] the minimization is converged when the maximum force is smaller than this value

emstep
¶ (0.01) [nm] initial stepsize

nstcgsteep
¶ (1000) [steps] frequency of performing 1 steepest descent step while doing conjugate gradient energy minimization.

nbfgscorr
¶ (10) Number of correction steps to use for LBFGS minimization. A higher number is (at least theoretically) more accurate, but slower.
Shell Molecular Dynamics¶
When shells or flexible constraints are present in the system the
positions of the shells and the lengths of the flexible constraints
are optimized at every time step until either the RMS force on the
shells and constraints is less than emtol
, or a maximum number
of iterations niter
has been reached. Minimization is converged
when the maximum force is smaller than emtol
. For shell MD this
value should be 1.0 at most.

niter
¶ (20) maximum number of iterations for optimizing the shell positions and the flexible constraints.

fcstep
¶ (0) [ps^{2}] the step size for optimizing the flexible constraints. Should be chosen as mu/(d2V/dq2) where mu is the reduced mass of two particles in a flexible constraint and d2V/dq2 is the second derivative of the potential in the constraint direction. Hopefully this number does not differ too much between the flexible constraints, as the number of iterations and thus the runtime is very sensitive to fcstep. Try several values!
Test particle insertion¶

rtpi
¶ (0.05) [nm] the test particle insertion radius, see integrators
integrator=tpi
andintegrator=tpic
Output control¶

nstxout
¶ (0) [steps] number of steps that elapse between writing coordinates to the output trajectory file (trr), the last coordinates are always written unless 0, which means coordinates are not written into the trajectory file.

nstvout
¶ (0) [steps] number of steps that elapse between writing velocities to the output trajectory file (trr), the last velocities are always written unless 0, which means velocities are not written into the trajectory file.

nstfout
¶ (0) [steps] number of steps that elapse between writing forces to the output trajectory file (trr), the last forces are always written, unless 0, which means forces are not written into the trajectory file.

nstlog
¶ (1000) [steps] number of steps that elapse between writing energies to the log file, the last energies are always written.

nstcalcenergy
¶ (100) number of steps that elapse between calculating the energies, 0 is never. This option is only relevant with dynamics. This option affects the performance in parallel simulations, because calculating energies requires global communication between all processes which can become a bottleneck at high parallelization.

nstenergy
¶ (1000) [steps] number of steps that elapse between writing energies to energy file, the last energies are always written, should be a multiple of
nstcalcenergy
. Note that the exact sums and fluctuations over all MD steps modulonstcalcenergy
are stored in the energy file, so gmx energy can report exact energy averages and fluctuations also whennstenergy
> 1

nstxoutcompressed
¶ (0) [steps] number of steps that elapse between writing position coordinates using lossy compression (xtc file), 0 for not writing compressed coordinates output.

compressedxprecision
¶ (1000) [real] precision with which to write to the compressed trajectory file

compressedxgrps
¶ group(s) to write to the compressed trajectory file, by default the whole system is written (if
nstxoutcompressed
> 0)

energygrps
¶ group(s) for which to write to write shortranged nonbonded potential energies to the energy file (not supported on GPUs)
Neighbor searching¶

cutoffscheme
¶ 
Verlet
¶ Generate a pair list with buffering. The buffer size is automatically set based on
verletbuffertolerance
, unless this is set to 1, in which caserlist
will be used.

group
¶ Generate a pair list for groups of atoms, corresponding to the charge groups in the topology. This option is no longer supported.


nstlist
¶  [steps]

>0
¶ Frequency to update the neighbor list. When dynamics and
verletbuffertolerance
set,nstlist
is actually a minimum value and gmx mdrun might increase it, unless it is set to 1. With parallel simulations and/or nonbonded force calculation on the GPU, a value of 20 or 40 often gives the best performance.

0
¶ The neighbor list is only constructed once and never updated. This is mainly useful for vacuum simulations in which all particles see each other. But vacuum simulations are (temporarily) not supported.

<0
¶ Unused.

pbc
¶ 
xyz
¶ Use periodic boundary conditions in all directions.

no
¶ Use no periodic boundary conditions, ignore the box. To simulate without cutoffs, set all cutoffs and
nstlist
to 0. For best performance without cutoffs on a single MPI rank, setnstlist
to zero andnstype=simple
.

xy
¶ Use periodic boundary conditions in x and y directions only. This works only with
nstype=grid
and can be used in combination with walls. Without walls or with only one wall the system size is infinite in the z direction. Therefore pressure coupling or Ewald summation methods can not be used. These disadvantages do not apply when two walls are used.


periodicmolecules
¶ 
no
¶ molecules are finite, fast molecular PBC can be used

yes
¶ for systems with molecules that couple to themselves through the periodic boundary conditions, this requires a slower PBC algorithm and molecules are not made whole in the output


verletbuffertolerance
¶ (0.005) [kJ mol^{1} ps^{1}]
Used when performing a simulation with dynamics. This sets the maximum allowed error for pair interactions per particle caused by the Verlet buffer, which indirectly sets
rlist
. As bothnstlist
and the Verlet buffer size are fixed (for performance reasons), particle pairs not in the pair list can occasionally get within the cutoff distance duringnstlist
1 steps. This causes very small jumps in the energy. In a constanttemperature ensemble, these very small energy jumps can be estimated for a given cutoff andrlist
. The estimate assumes a homogeneous particle distribution, hence the errors might be slightly underestimated for multiphase systems. (See the reference manual for details). For longer pairlist lifetime (nstlist
1) *dt
the buffer is overestimated, because the interactions between particles are ignored. Combined with cancellation of errors, the actual drift of the total energy is usually one to two orders of magnitude smaller. Note that the generated buffer size takes into account that the GROMACS pairlist setup leads to a reduction in the drift by a factor 10, compared to a simple particlepair based list. Without dynamics (energy minimization etc.), the buffer is 5% of the cutoff. For NVE simulations the initial temperature is used, unless this is zero, in which case a buffer of 10% is used. For NVE simulations the tolerance usually needs to be lowered to achieve proper energy conservation on the nanosecond time scale. To override the automated buffer setting, useverletbuffertolerance
=1 and setrlist
manually.

rlist
¶ (1) [nm] Cutoff distance for the shortrange neighbor list. With dynamics, this is by default set by the
verletbuffertolerance
option and the value ofrlist
is ignored. Without dynamics, this is by default set to the maximum cutoff plus 5% buffer, except for test particle insertion, where the buffer is managed exactly and automatically. For NVE simulations, where the automated setting is not possible, the advised procedure is to run gmx grompp with an NVT setup with the expected temperature and copy the resulting value ofrlist
to the NVE setup.
Electrostatics¶

coulombtype
¶ 
Cutoff
¶ Plain cutoff with pair list radius
rlist
and Coulomb cutoffrcoulomb
, whererlist
>=rcoulomb
.

Ewald
¶ Classical Ewald sum electrostatics. The realspace cutoff
rcoulomb
should be equal torlist
. Use e.g.rlist
=0.9,rcoulomb
=0.9. The highest magnitude of wave vectors used in reciprocal space is controlled byfourierspacing
. The relative accuracy of direct/reciprocal space is controlled byewaldrtol
.NOTE: Ewald scales as O(N^{3/2}) and is thus extremely slow for large systems. It is included mainly for reference  in most cases PME will perform much better.

PME
¶ Fast smooth ParticleMesh Ewald (SPME) electrostatics. Direct space is similar to the Ewald sum, while the reciprocal part is performed with FFTs. Grid dimensions are controlled with
fourierspacing
and the interpolation order withpmeorder
. With a grid spacing of 0.1 nm and cubic interpolation the electrostatic forces have an accuracy of 23*10^{4}. Since the error from the vdwcutoff is larger than this you might try 0.15 nm. When running in parallel the interpolation parallelizes better than the FFT, so try decreasing grid dimensions while increasing interpolation.

P3MAD
¶ ParticleParticle ParticleMesh algorithm with analytical derivative for for long range electrostatic interactions. The method and code is identical to SPME, except that the influence function is optimized for the grid. This gives a slight increase in accuracy.

ReactionField
¶ Reaction field electrostatics with Coulomb cutoff
rcoulomb
, whererlist
>=rvdw
. The dielectric constant beyond the cutoff isepsilonrf
. The dielectric constant can be set to infinity by settingepsilonrf
=0.

User
¶ Currently unsupported. gmx mdrun will now expect to find a file
table.xvg
with userdefined potential functions for repulsion, dispersion and Coulomb. When pair interactions are present, gmx mdrun also expects to find a filetablep.xvg
for the pair interactions. When the same interactions should be used for nonbonded and pair interactions the user can specify the same file name for both table files. These files should contain 7 columns: thex
value,f(x)
,f'(x)
,g(x)
,g'(x)
,h(x)
,h'(x)
, wheref(x)
is the Coulomb function,g(x)
the dispersion function andh(x)
the repulsion function. Whenvdwtype
is not set to User the values forg
,g'
,h
andh'
are ignored. For the nonbonded interactionsx
values should run from 0 to the largest cutoff distance +tableextension
and should be uniformly spaced. For the pair interactions the table length in the file will be used. The optimal spacing, which is used for nonuser tables, is0.002 nm
when you run in mixed precision or0.0005 nm
when you run in double precision. The function value atx=0
is not important. More information is in the printed manual.

PMESwitch
¶ Currently unsupported. A combination of PME and a switch function for the directspace part (see above).
rcoulomb
is allowed to be smaller thanrlist
.

PMEUser
¶ Currently unsupported. A combination of PME and user tables (see above).
rcoulomb
is allowed to be smaller thanrlist
. The PME mesh contribution is subtracted from the user table by gmx mdrun. Because of this subtraction the user tables should contain about 10 decimal places.

PMEUserSwitch
¶ Currently unsupported. A combination of PMEUser and a switching function (see above). The switching function is applied to final particleparticle interaction, i.e. both to the user supplied function and the PME Mesh correction part.


coulombmodifier
¶ 
Potentialshift
¶ Shift the Coulomb potential by a constant such that it is zero at the cutoff. This makes the potential the integral of the force. Note that this does not affect the forces or the sampling.

None
¶ Use an unmodified Coulomb potential. This can be useful when comparing energies with those computed with other software.


rcoulombswitch
¶ (0) [nm] where to start switching the Coulomb potential, only relevant when force or potential switching is used

rcoulomb
¶ (1) [nm] The distance for the Coulomb cutoff. Note that with PME this value can be increased by the PME tuning in gmx mdrun along with the PME grid spacing.

epsilonr
¶ (1) The relative dielectric constant. A value of 0 means infinity.

epsilonrf
¶ (0) The relative dielectric constant of the reaction field. This is only used with reactionfield electrostatics. A value of 0 means infinity.
Van der Waals¶

vdwtype
¶ 

PME
¶ Fast smooth Particlemesh Ewald (SPME) for VdW interactions. The grid dimensions are controlled with
fourierspacing
in the same way as for electrostatics, and the interpolation order is controlled withpmeorder
. The relative accuracy of direct/reciprocal space is controlled byewaldrtollj
, and the specific combination rules that are to be used by the reciprocal routine are set usingljpmecombrule
.

Shift
¶ This functionality is deprecated and replaced by using
vdwtype=Cutoff
withvdwmodifier=Forceswitch
. The LJ (not Buckingham) potential is decreased over the whole range and the forces decay smoothly to zero betweenrvdwswitch
andrvdw
.

Switch
¶ This functionality is deprecated and replaced by using
vdwtype=Cutoff
withvdwmodifier=Potentialswitch
. The LJ (not Buckingham) potential is normal out torvdwswitch
, after which it is switched off to reach zero atrvdw
. Both the potential and force functions are continuously smooth, but be aware that all switch functions will give rise to a bulge (increase) in the force (since we are switching the potential).

User
¶ Currently unsupported. See user for
coulombtype
. The function value at zero is not important. When you want to use LJ correction, make sure thatrvdw
corresponds to the cutoff in the userdefined function. Whencoulombtype
is not set to User the values for thef
andf'
columns are ignored.


vdwmodifier
¶ 
Potentialshift
¶ Shift the Van der Waals potential by a constant such that it is zero at the cutoff. This makes the potential the integral of the force. Note that this does not affect the forces or the sampling.

None
¶ Use an unmodified Van der Waals potential. This can be useful when comparing energies with those computed with other software.

Forceswitch
¶ Smoothly switches the forces to zero between
rvdwswitch
andrvdw
. This shifts the potential shift over the whole range and switches it to zero at the cutoff. Note that this is more expensive to calculate than a plain cutoff and it is not required for energy conservation, since Potentialshift conserves energy just as well.

Potentialswitch
¶ Smoothly switches the potential to zero between
rvdwswitch
andrvdw
. Note that this introduces articifically large forces in the switching region and is much more expensive to calculate. This option should only be used if the force field you are using requires this.


rvdwswitch
¶ (0) [nm] where to start switching the LJ force and possibly the potential, only relevant when force or potential switching is used

rvdw
¶ (1) [nm] distance for the LJ or Buckingham cutoff
Tables¶

tableextension
¶ (1) [nm] Extension of the nonbonded potential lookup tables beyond the largest cutoff distance. With actual nonbonded interactions the tables are never accessed beyond the cutoff. But a longer table length might be needed for the 14 interactions, which are always tabulated irrespective of the use of tables for the nonbonded interactions.

energygrptable
¶ Currently unsupported. When user tables are used for electrostatics and/or VdW, here one can give pairs of energy groups for which seperate user tables should be used. The two energy groups will be appended to the table file name, in order of their definition in
energygrps
, seperated by underscores. For example, ifenergygrps = Na Cl Sol
andenergygrptable = Na Na Na Cl
, gmx mdrun will readtable_Na_Na.xvg
andtable_Na_Cl.xvg
in addition to the normaltable.xvg
which will be used for all other energy group pairs.
Ewald¶

fourierspacing
¶ (0.12) [nm] For ordinary Ewald, the ratio of the box dimensions and the spacing determines a lower bound for the number of wave vectors to use in each (signed) direction. For PME and P3M, that ratio determines a lower bound for the number of Fourierspace grid points that will be used along that axis. In all cases, the number for each direction can be overridden by entering a nonzero value for that
fouriernx
direction. For optimizing the relative load of the particleparticle interactions and the mesh part of PME, it is useful to know that the accuracy of the electrostatics remains nearly constant when the Coulomb cutoff and the PME grid spacing are scaled by the same factor. Note that this spacing can be scaled up along withrcoulomb
by the PME tuning in gmx mdrun.

fouriernx
¶

fourierny
¶

fouriernz
¶ (0) Highest magnitude of wave vectors in reciprocal space when using Ewald. Grid size when using PME or P3M. These values override
fourierspacing
per direction. The best choice is powers of 2, 3, 5 and 7. Avoid large primes. Note that these grid sizes can be reduced along with scaling uprcoulomb
by the PME tuning in gmx mdrun.

pmeorder
¶ (4) Interpolation order for PME. 4 equals cubic interpolation. You might try 6/8/10 when running in parallel and simultaneously decrease grid dimension.

ewaldrtol
¶ (10^{5}) The relative strength of the Ewaldshifted direct potential at
rcoulomb
is given byewaldrtol
. Decreasing this will give a more accurate direct sum, but then you need more wave vectors for the reciprocal sum.

ewaldrtollj
¶ (10^{3}) When doing PME for VdWinteractions,
ewaldrtollj
is used to control the relative strength of the dispersion potential atrvdw
in the same way asewaldrtol
controls the electrostatic potential.

ljpmecombrule
¶ (Geometric) The combination rules used to combine VdWparameters in the reciprocal part of LJPME. Geometric rules are much faster than LorentzBerthelot and usually the recommended choice, even when the rest of the force field uses the LorentzBerthelot rules.

Geometric
¶ Apply geometric combination rules

LorentzBerthelot
¶ Apply LorentzBerthelot combination rules


ewaldgeometry
¶ 
3d
¶ The Ewald sum is performed in all three dimensions.

3dc
¶ The reciprocal sum is still performed in 3D, but a force and potential correction applied in the
z
dimension to produce a pseudo2D summation. If your system has a slab geometry in thexy
plane you can try to increase thez
dimension of the box (a box height of 3 times the slab height is usually ok) and use this option.


epsilonsurface
¶ (0) This controls the dipole correction to the Ewald summation in 3D. The default value of zero means it is turned off. Turn it on by setting it to the value of the relative permittivity of the imaginary surface around your infinite system. Be careful  you shouldn’t use this if you have free mobile charges in your system. This value does not affect the slab 3DC variant of the long range corrections.
Temperature coupling¶

tcoupl
¶ 
no
¶ No temperature coupling.

berendsen
¶ Temperature coupling with a Berendsen thermostat to a bath with temperature
reft
, with time constanttaut
. Several groups can be coupled separately, these are specified in thetcgrps
field separated by spaces.

nosehoover
¶ Temperature coupling using a NoseHoover extended ensemble. The reference temperature and coupling groups are selected as above, but in this case
taut
controls the period of the temperature fluctuations at equilibrium, which is slightly different from a relaxation time. For NVT simulations the conserved energy quantity is written to the energy and log files.

andersen
¶ Temperature coupling by randomizing a fraction of the particle velocities at each timestep. Reference temperature and coupling groups are selected as above.
taut
is the average time between randomization of each molecule. Inhibits particle dynamics somewhat, but little or no ergodicity issues. Currently only implemented with velocity Verlet, and not implemented with constraints.

andersenmassive
¶ Temperature coupling by randomizing velocities of all particles at infrequent timesteps. Reference temperature and coupling groups are selected as above.
taut
is the time between randomization of all molecules. Inhibits particle dynamics somewhat, but little or no ergodicity issues. Currently only implemented with velocity Verlet.

vrescale
¶ Temperature coupling using velocity rescaling with a stochastic term (JCP 126, 014101). This thermostat is similar to Berendsen coupling, with the same scaling using
taut
, but the stochastic term ensures that a proper canonical ensemble is generated. The random seed is set withldseed
. This thermostat works correctly even fortaut
=0. For NVT simulations the conserved energy quantity is written to the energy and log file.


nsttcouple
¶ (1) The frequency for coupling the temperature. The default value of 1 sets
nsttcouple
equal tonstlist
, unlessnstlist
<=0, then a value of 10 is used. For velocity Verlet integratorsnsttcouple
is set to 1.

nhchainlength
¶ (10) The number of chained NoseHoover thermostats for velocity Verlet integrators, the leapfrog
integrator=md
integrator only supports 1. Data for the NH chain variables is not printed to the edr file by default, but can be turned on with theprintnosehooverchainvariables
option.

printnosehooverchainvariables
¶ 
no
¶ Do not store NoseHoover chain variables in the energy file.

yes
¶ Store all positions and velocities of the NoseHoover chain in the energy file.


tcgrps
¶ groups to couple to separate temperature baths
Pressure coupling¶

pcoupl
¶ 
no
¶ No pressure coupling. This means a fixed box size.

Berendsen
¶ Exponential relaxation pressure coupling with time constant
taup
. The box is scaled everynstpcouple
steps. It has been argued that this does not yield a correct thermodynamic ensemble, but it is the most efficient way to scale a box at the beginning of a run.

ParrinelloRahman
¶ Extendedensemble pressure coupling where the box vectors are subject to an equation of motion. The equation of motion for the atoms is coupled to this. No instantaneous scaling takes place. As for NoseHoover temperature coupling the time constant
taup
is the period of pressure fluctuations at equilibrium. This is probably a better method when you want to apply pressure scaling during data collection, but beware that you can get very large oscillations if you are starting from a different pressure. For simulations where the exact fluctations of the NPT ensemble are important, or if the pressure coupling time is very short it may not be appropriate, as the previous time step pressure is used in some steps of the GROMACS implementation for the current time step pressure.

MTTK
¶ MartynaTuckermanTobiasKlein implementation, only useable with
integrator=mdvv
orintegrator=mdvvavek
, very similar to ParrinelloRahman. As for NoseHoover temperature coupling the time constanttaup
is the period of pressure fluctuations at equilibrium. This is probably a better method when you want to apply pressure scaling during data collection, but beware that you can get very large oscillations if you are starting from a different pressure. Currently (as of version 5.1), it only supports isotropic scaling, and only works without constraints.


pcoupltype
¶ Specifies the kind of isotropy of the pressure coupling used. Each kind takes one or more values for
compressibility
andrefp
. Only a single value is permitted fortaup
.
isotropic
¶ Isotropic pressure coupling with time constant
taup
. One value each forcompressibility
andrefp
is required.

semiisotropic
¶ Pressure coupling which is isotropic in the
x
andy
direction, but different in thez
direction. This can be useful for membrane simulations. Two values each forcompressibility
andrefp
are required, forx/y
andz
directions respectively.

anisotropic
¶ Same as before, but 6 values are needed for
xx
,yy
,zz
,xy/yx
,xz/zx
andyz/zy
components, respectively. When the offdiagonal compressibilities are set to zero, a rectangular box will stay rectangular. Beware that anisotropic scaling can lead to extreme deformation of the simulation box.

surfacetension
¶ Surface tension coupling for surfaces parallel to the xyplane. Uses normal pressure coupling for the
z
direction, while the surface tension is coupled to thex/y
dimensions of the box. The firstrefp
value is the reference surface tension times the number of surfacesbar nm
, the second value is the referencez
pressurebar
. The twocompressibility
values are the compressibility in thex/y
andz
direction respectively. The value for thez
compressibility should be reasonably accurate since it influences the convergence of the surfacetension, it can also be set to zero to have a box with constant height.


nstpcouple
¶ (1) The frequency for coupling the pressure. The default value of 1 sets
nstpcouple
equal tonstlist
, unlessnstlist
<=0, then a value of 10 is used. For velocity Verlet integratorsnstpcouple
is set to 1.

taup
¶ (1) [ps] The time constant for pressure coupling (one value for all directions).

compressibility
¶ [bar^{1}] The compressibility (NOTE: this is now really in bar^{1}) For water at 1 atm and 300 K the compressibility is 4.5e5 bar^{1}. The number of required values is implied by
pcoupltype
.

refp
¶ [bar] The reference pressure for coupling. The number of required values is implied by
pcoupltype
.

refcoordscaling
¶ 
no
¶ The reference coordinates for position restraints are not modified. Note that with this option the virial and pressure might be ill defined, see here for more details.

all
¶ The reference coordinates are scaled with the scaling matrix of the pressure coupling.

com
¶ Scale the center of mass of the reference coordinates with the scaling matrix of the pressure coupling. The vectors of each reference coordinate to the center of mass are not scaled. Only one COM is used, even when there are multiple molecules with position restraints. For calculating the COM of the reference coordinates in the starting configuration, periodic boundary conditions are not taken into account. Note that with this option the virial and pressure might be ill defined, see here for more details.

Simulated annealing¶
Simulated annealing is controlled separately for each temperature group in GROMACS. The reference temperature is a piecewise linear function, but you can use an arbitrary number of points for each group, and choose either a single sequence or a periodic behaviour for each group. The actual annealing is performed by dynamically changing the reference temperature used in the thermostat algorithm selected, so remember that the system will usually not instantaneously reach the reference temperature!

annealing
¶ Type of annealing for each temperature group

no
¶ No simulated annealing  just couple to reference temperature value.

single
¶ A single sequence of annealing points. If your simulation is longer than the time of the last point, the temperature will be coupled to this constant value after the annealing sequence has reached the last time point.

periodic
¶ The annealing will start over at the first reference point once the last reference time is reached. This is repeated until the simulation ends.


annealingnpoints
¶ A list with the number of annealing reference/control points used for each temperature group. Use 0 for groups that are not annealed. The number of entries should equal the number of temperature groups.

annealingtime
¶ List of times at the annealing reference/control points for each group. If you are using periodic annealing, the times will be used modulo the last value, i.e. if the values are 0, 5, 10, and 15, the coupling will restart at the 0ps value after 15ps, 30ps, 45ps, etc. The number of entries should equal the sum of the numbers given in
annealingnpoints
.

annealingtemp
¶ List of temperatures at the annealing reference/control points for each group. The number of entries should equal the sum of the numbers given in
annealingnpoints
.
Confused? OK, let’s use an example. Assume you have two temperature
groups, set the group selections to annealing = single periodic
,
the number of points of each group to annealingnpoints = 3 4
, the
times to annealingtime = 0 3 6 0 2 4 6
and finally temperatures
to annealingtemp = 298 280 270 298 320 320 298
. The first group
will be coupled to 298K at 0ps, but the reference temperature will
drop linearly to reach 280K at 3ps, and then linearly between 280K and
270K from 3ps to 6ps. After this is stays constant, at 270K. The
second group is coupled to 298K at 0ps, it increases linearly to 320K
at 2ps, where it stays constant until 4ps. Between 4ps and 6ps it
decreases to 298K, and then it starts over with the same pattern
again, i.e. rising linearly from 298K to 320K between 6ps and
8ps. Check the summary printed by gmx grompp if you are unsure!
Velocity generation¶

genvel
¶ 
no
¶ Do not generate velocities. The velocities are set to zero when there are no velocities in the input structure file.

yes
¶ Generate velocities in gmx grompp according to a Maxwell distribution at temperature
gentemp
, with random seedgenseed
. This is only meaningful withintegrator=md
.


gentemp
¶ (300) [K] temperature for Maxwell distribution
Bonds¶

constraints
¶ Controls which bonds in the topology will be converted to rigid holonomic constraints. Note that typical rigid water models do not have bonds, but rather a specialized
[settles]
directive, so are not affected by this keyword.
none
¶ No bonds converted to constraints.

hbonds
¶ Convert the bonds with Hatoms to constraints.

allbonds
¶ Convert all bonds to constraints.

hangles
¶ Convert all bonds to constraints and convert the angles that involve Hatoms to bondconstraints.

allangles
¶ Convert all bonds to constraints and all angles to bondconstraints.


constraintalgorithm
¶ Chooses which solver satisfies any nonSETTLE holonomic constraints.

LINCS
¶ LINear Constraint Solver. With domain decomposition the parallel version PLINCS is used. The accuracy in set with
lincsorder
, which sets the number of matrices in the expansion for the matrix inversion. After the matrix inversion correction the algorithm does an iterative correction to compensate for lengthening due to rotation. The number of such iterations can be controlled withlincsiter
. The root mean square relative constraint deviation is printed to the log file everynstlog
steps. If a bond rotates more thanlincswarnangle
in one step, a warning will be printed both to the log file and tostderr
. LINCS should not be used with coupled angle constraints.

SHAKE
¶ SHAKE is slightly slower and less stable than LINCS, but does work with angle constraints. The relative tolerance is set with
shaketol
, 0.0001 is a good value for “normal” MD. SHAKE does not support constraints between atoms on different decomposition domains, so it can only be used with domain decomposition when socalled updategroups are used, which is usally the case when only bonds involving hydrogens are constrained. SHAKE can not be used with energy minimization.


continuation
¶ This option was formerly known as
unconstrainedstart
.
no
¶ apply constraints to the start configuration and reset shells

yes
¶ do not apply constraints to the start configuration and do not reset shells, useful for exact coninuation and reruns


shaketol
¶ (0.0001) relative tolerance for SHAKE

lincsorder
¶ (4) Highest order in the expansion of the constraint coupling matrix. When constraints form triangles, an additional expansion of the same order is applied on top of the normal expansion only for the couplings within such triangles. For “normal” MD simulations an order of 4 usually suffices, 6 is needed for large timesteps with virtual sites or BD. For accurate energy minimization an order of 8 or more might be required. With domain decomposition, the cell size is limited by the distance spanned by
lincsorder
+1 constraints. When one wants to scale further than this limit, one can decreaselincsorder
and increaselincsiter
, since the accuracy does not deteriorate when (1+lincsiter
)*lincsorder
remains constant.

lincsiter
¶ (1) Number of iterations to correct for rotational lengthening in LINCS. For normal runs a single step is sufficient, but for NVE runs where you want to conserve energy accurately or for accurate energy minimization you might want to increase it to 2.

lincswarnangle
¶ (30) [deg] maximum angle that a bond can rotate before LINCS will complain
Energy group exclusions¶

energygrpexcl
¶ Pairs of energy groups for which all nonbonded interactions are excluded. An example: if you have two energy groups
Protein
andSOL
, specifyingenergygrpexcl = Protein Protein SOL SOL
would give only the nonbonded interactions between the protein and the solvent. This is especially useful for speeding up energy calculations withmdrun rerun
and for excluding interactions within frozen groups.
Walls¶

nwall
¶ (0) When set to 1 there is a wall at
z=0
, when set to 2 there is also a wall atz=zbox
. Walls can only be used withpbc
=xy
. When set to 2, pressure coupling and Ewald summation can be used (it is usually best to use semiisotropic pressure coupling with thex/y
compressibility set to 0, as otherwise the surface area will change). Walls interact wit the rest of the system through an optionalwallatomtype
. Energy groupswall0
andwall1
(fornwall
=2) are added automatically to monitor the interaction of energy groups with each wall. The center of mass motion removal will be turned off in thez
direction.

wallatomtype
¶ the atom type name in the force field for each wall. By (for example) defining a special wall atom type in the topology with its own combination rules, this allows for independent tuning of the interaction of each atomtype with the walls.

walltype
¶ 
93
¶ LJ integrated over the volume behind the wall: 93 potential

104
¶ LJ integrated over the wall surface: 104 potential

126
¶ direct LJ potential with the
z
distance from the wall


table
¶ user defined potentials indexed with the
z
distance from the wall, the tables are read analogously to theenergygrptable
option, where the first name is for a “normal” energy group and the second name iswall0
orwall1
, only the dispersion and repulsion columns are used

wallrlinpot
¶ (1) [nm] Below this distance from the wall the potential is continued linearly and thus the force is constant. Setting this option to a postive value is especially useful for equilibration when some atoms are beyond a wall. When the value is <=0 (<0 for
walltype
=table), a fatal error is generated when atoms are beyond a wall.

walldensity
¶ [nm^{3}] / [nm^{2}] the number density of the atoms for each wall for wall types 93 and 104

wallewaldzfac
¶ (3) The scaling factor for the third box vector for Ewald summation only, the minimum is 2. Ewald summation can only be used with
nwall
=2, where one should useewaldgeometry
=3dc
. The empty layer in the box serves to decrease the unphysical Coulomb interaction between periodic images.
COM pulling¶
Note that where pulling coordinates are applicable, there can be more
than one (set with pullncoords
) and multiple related mdp
variables will exist accordingly. Documentation references to things
like pullcoord1vec
should be understood to apply to to the
applicable pulling coordinate, eg. the second pull coordinate is described by
pullcoord2vec, pullcoord2k, and so on.

pull
¶ 
no
¶ No center of mass pulling. All the following pull options will be ignored (and if present in the mdp file, they unfortunately generate warnings)

yes
¶ Center of mass pulling will be applied on 1 or more groups using 1 or more pull coordinates.


pullcylinderr
¶ (1.5) [nm] the radius of the cylinder for
pullcoord1geometry=cylinder

pullconstrtol
¶ (10^{6}) the relative constraint tolerance for constraint pulling

pullprintcom
¶ 
no
¶ do not print the COM for any group

yes
¶ print the COM of all groups for all pull coordinates


pullprintrefvalue
¶ 
no
¶ do not print the reference value for each pull coordinate

yes
¶ print the reference value for each pull coordinate


pullprintcomponents
¶ 
no
¶ only print the distance for each pull coordinate

yes
¶ print the distance and Cartesian components selected in
pullcoord1dim


pullnstxout
¶ (50) frequency for writing out the COMs of all the pull group (0 is never)

pullnstfout
¶ (50) frequency for writing out the force of all the pulled group (0 is never)

pullpbcrefprevstepcom
¶ 
no
¶ Use the reference atom (
pullgroup1pbcatom
) for the treatment of periodic boundary conditions.

yes
¶ Use the COM of the previous step as reference for the treatment of periodic boundary conditions. The reference is initialized using the reference atom (
pullgroup1pbcatom
), which should be located centrally in the group. Using the COM from the previous step can be useful if one or more pull groups are large.


pullxoutaverage
¶ 
no
¶ Write the instantaneous coordinates for all the pulled groups.

yes
¶ Write the average coordinates (since last output) for all the pulled groups. N.b., some analysis tools might expect instantaneous pull output.


pullfoutaverage
¶ 
no
¶ Write the instantaneous force for all the pulled groups.

yes
¶ Write the average force (since last output) for all the pulled groups. N.b., some analysis tools might expect instantaneous pull output.


pullngroups
¶ (1) The number of pull groups, not including the absolute reference group, when used. Pull groups can be reused in multiple pull coordinates. Below only the pull options for group 1 are given, further groups simply increase the group index number.

pullncoords
¶ (1) The number of pull coordinates. Below only the pull options for coordinate 1 are given, further coordinates simply increase the coordinate index number.

pullgroup1name
¶ The name of the pull group, is looked up in the index file or in the default groups to obtain the atoms involved.

pullgroup1weights
¶ Optional relative weights which are multiplied with the masses of the atoms to give the total weight for the COM. The number should be 0, meaning all 1, or the number of atoms in the pull group.

pullgroup1pbcatom
¶ (0) The reference atom for the treatment of periodic boundary conditions inside the group (this has no effect on the treatment of the pbc between groups). This option is only important when the diameter of the pull group is larger than half the shortest box vector. For determining the COM, all atoms in the group are put at their periodic image which is closest to
pullgroup1pbcatom
. A value of 0 means that the middle atom (number wise) is used, which is only safe for small groups. gmx grompp checks that the maximum distance from the reference atom (specifically chosen, or not) to the other atoms in the group is not too large. This parameter is not used withpullcoord1geometry
cylinder. A value of 1 turns on cosine weighting, which is useful for a group of molecules in a periodic system, e.g. a water slab (see Engin et al. J. Chem. Phys. B 2010).

pullcoord1type
¶ 
umbrella
¶ Center of mass pulling using an umbrella potential between the reference group and one or more groups.

constraint
¶ Center of mass pulling using a constraint between the reference group and one or more groups. The setup is identical to the option umbrella, except for the fact that a rigid constraint is applied instead of a harmonic potential.

constantforce
¶ Center of mass pulling using a linear potential and therefore a constant force. For this option there is no reference position and therefore the parameters
pullcoord1init
andpullcoord1rate
are not used.

flatbottom
¶ At distances above
pullcoord1init
a harmonic potential is applied, otherwise no potential is applied.

flatbottomhigh
¶ At distances below
pullcoord1init
a harmonic potential is applied, otherwise no potential is applied.

externalpotential
¶ An external potential that needs to be provided by another module.


pullcoord1potentialprovider
¶ The name of the external module that provides the potential for the case where
pullcoord1type
is externalpotential.

pullcoord1geometry
¶ 
distance
¶ Pull along the vector connecting the two groups. Components can be selected with
pullcoord1dim
.

direction
¶ Pull in the direction of
pullcoord1vec
.

directionperiodic
¶ As
pullcoord1geometry=direction
, but does not apply periodic box vector corrections to keep the distance within half the box length. This is (only) useful for pushing groups apart by more than half the box length by continuously changing the reference location using a pull rate. With this geometry the box should not be dynamic (e.g. no pressure scaling) in the pull dimensions and the pull force is not added to the virial.

directionrelative
¶ As
pullcoord1geometry=direction
, but the pull vector is the vector that points from the COM of a third to the COM of a fourth pull group. This means that 4 groups need to be supplied inpullcoord1groups
. Note that the pull force will give rise to a torque on the pull vector, which is turn leads to forces perpendicular to the pull vector on the two groups defining the vector. If you want a pull group to move between the two groups defining the vector, simply use the union of these two groups as the reference group.

cylinder
¶ Designed for pulling with respect to a layer where the reference COM is given by a local cylindrical part of the reference group. The pulling is in the direction of
pullcoord1vec
. From the first of the two groups inpullcoord1groups
a cylinder is selected around the axis going through the COM of the second group with directionpullcoord1vec
with radiuspullcylinderr
. Weights of the atoms decrease continously to zero as the radial distance goes from 0 topullcylinderr
(mass weighting is also used). The radial dependence gives rise to radial forces on both pull groups. Note that the radius should be smaller than half the box size. For tilted cylinders they should be even smaller than half the box size since the distance of an atom in the reference group from the COM of the pull group has both a radial and an axial component. This geometry is not supported with constraint pulling.

angle
¶ Pull along an angle defined by four groups. The angle is defined as the angle between two vectors: the vector connecting the COM of the first group to the COM of the second group and the vector connecting the COM of the third group to the COM of the fourth group.

angleaxis
¶ As
pullcoord1geometry=angle
but the second vector is given bypullcoord1vec
. Thus, only the two groups that define the first vector need to be given.

dihedral
¶ Pull along a dihedral angle defined by six groups. These pairwise define three vectors: the vector connecting the COM of group 1 to the COM of group 2, the COM of group 3 to the COM of group 4, and the COM of group 5 to the COM group 6. The dihedral angle is then defined as the angle between two planes: the plane spanned by the the two first vectors and the plane spanned the two last vectors.


pullcoord1groups
¶ The group indices on which this pull coordinate will operate. The number of group indices required is geometry dependent. The first index can be 0, in which case an absolute reference of
pullcoord1origin
is used. With an absolute reference the system is no longer translation invariant and one should think about what to do with the center of mass motion.

pullcoord1dim
¶ (Y Y Y) Selects the dimensions that this pull coordinate acts on and that are printed to the output files when
pullprintcomponents
=pullcoord1start=yes
. Withpullcoord1geometry
=pullcoord1geometry=distance
, only Cartesian components set to Y contribute to the distance. Thus setting this to Y Y N results in a distance in the x/y plane. With other geometries all dimensions with nonzero entries inpullcoord1vec
should be set to Y, the values for other dimensions only affect the output.

pullcoord1origin
¶ (0.0 0.0 0.0) The pull reference position for use with an absolute reference.

pullcoord1vec
¶ (0.0 0.0 0.0) The pull direction. gmx grompp normalizes the vector.

pullcoord1start
¶ 
no
¶ do not modify
pullcoord1init

yes
¶ add the COM distance of the starting conformation to
pullcoord1init


pullcoord1init
¶ (0.0) [nm] or [deg] The reference distance or reference angle at t=0.

pullcoord1rate
¶ (0) [nm/ps] or [deg/ps] The rate of change of the reference position or reference angle.

pullcoord1k
¶ (0) [kJ mol^{1} nm^{2}] or [kJ mol^{1} nm^{1}] or [kJ mol^{1} rad^{2}] or [kJ mol^{1} rad^{1}] The force constant. For umbrella pulling this is the harmonic force constant in kJ mol^{1} nm^{2} (or kJ mol^{1} rad^{2} for angles). For constant force pulling this is the force constant of the linear potential, and thus the negative (!) of the constant force in kJ mol^{1} nm^{1} (or kJ mol^{1} rad^{1} for angles). Note that for angles the force constant is expressed in terms of radians (while
pullcoord1init
andpullcoord1rate
are expressed in degrees).

pullcoord1kB
¶ (pullk1) [kJ mol^{1} nm^{2}] or [kJ mol^{1} nm^{1}] or [kJ mol^{1} rad^{2}] or [kJ mol^{1} rad^{1}] As
pullcoord1k
, but for state B. This is only used whenfreeenergy
is turned on. The force constant is then (1  lambda) *pullcoord1k
+ lambda *pullcoord1kB
.
AWH adaptive biasing¶

awh
¶ 
no
¶ No biasing.

yes
¶ Adaptively bias a reaction coordinate using the AWH method and estimate the corresponding PMF. The PMF and other AWH data are written to energy file at an interval set by
awhnstout
and can be extracted with thegmx awh
tool. The AWH coordinate can be multidimensional and is defined by mapping each dimension to a pull coordinate index. This is only allowed ifpullcoord1type=externalpotential
andpullcoord1potentialprovider
=awh
for the concerned pull coordinate indices. Pull geometry ‘directionperiodic’ is not supported by AWH.


awhpotential
¶ 
convolved
¶ The applied biasing potential is the convolution of the bias function and a set of harmonic umbrella potentials (see
awhpotential=umbrella
below). This results in a smooth potential function and force. The resolution of the potential is set by the force constant of each umbrella, seeawh1dim1forceconstant
.

umbrella
¶ The potential bias is applied by controlling the position of an harmonic potential using MonteCarlo sampling. The force constant is set with
awh1dim1forceconstant
. The umbrella location is sampled using MonteCarlo everyawhnstsample
steps. There are no advantages to using an umbrella. This option is mainly for comparison and testing purposes.

AWH will not share biases across simulations started with gmx mdrun option
multidir
. The biases will be independent.
With gmx mdrun and option
multidir
the bias and PMF estimates for biases withawh1sharegroup
>0 will be shared across simulations with the biases with the sameawh1sharegroup
value. The simulations should have the same AWH settings for sharing to make sense. gmx mdrun will check whether the simulations are technically compatible for sharing, but the user should check that bias sharing physically makes sense.

awhseed
¶ (1) Random seed for MonteCarlo sampling the umbrella position, where 1 indicates to generate a seed. Only used with
awhpotential=umbrella
.

awhnstout
¶ (100000) Number of steps between printing AWH data to the energy file, should be a multiple of
nstenergy
.

awhnstsample
¶ (10) Number of steps between sampling of the coordinate value. This sampling is the basis for updating the bias and estimating the PMF and other AWH observables.

awhnsamplesupdate
¶ (10) The number of coordinate samples used for each AWH update. The update interval in steps is
awhnstsample
times this value.

awhnbias
¶ (1) The number of biases, each acting on its own coordinate. The following options should be specified for each bias although below only the options for bias number 1 is shown. Options for other bias indices are obtained by replacing ‘1’ by the bias index.

awh1errorinit
¶ (10.0) [kJ mol^{1}] Estimated initial average error of the PMF for this bias. This value together with the given diffusion constant(s)
awh1dim1diffusion
determine the initial biasing rate. The error is obviously not known a priori. Only a rough estimate ofawh1errorinit
is needed however. As a general guideline, leaveawh1errorinit
to its default value when starting a new simulation. On the other hand, when there is a priori knowledge of the PMF (e.g. when an initial PMF estimate is provided, see theawh1userdata
option) thenawh1errorinit
should reflect that knowledge.

awh1growth
¶ 
explinear
¶
Each bias keeps a reference weight histogram for the coordinate samples. Its size sets the magnitude of the bias function and free energy estimate updates (few samples corresponds to large updates and vice versa). Thus, its growth rate sets the maximum convergence rate. By default, there is an initial stage in which the histogram grows close to exponentially (but slower than the sampling rate). In the final stage that follows, the growth rate is linear and equal to the sampling rate (set by
awhnstsample
). The initial stage is typically necessary for efficient convergence when starting a new simulation where high free energy barriers have not yet been flattened by the bias.
linear
¶
As
awh1growth=explinear
but skip the initial stage. This may be useful if there is a priori knowledge (seeawh1errorinit
) which eliminates the need for an initial stage. This is also the setting compatible withawh1target=localboltzmann
.

awh1equilibratehistogram
¶ 
no
¶ Do not equilibrate histogram.

yes
¶ Before entering the initial stage (see
awh1growth=explinear
), make sure the histogram of sampled weights is following the target distribution closely enough (specifically, at least 80% of the target region needs to have a local relative error of less than 20%). This option would typically only be used whenawh1sharegroup
> 0 and the initial configurations poorly represent the target distribution.


awh1target
¶ 
constant
¶ The bias is tuned towards a constant (uniform) coordinate distribution in the defined sampling interval (defined by [
awh1dim1start
,awh1dim1end
]).

cutoff
¶ Similar to
awh1target=constant
, but the target distribution is proportional to 1/(1 + exp(F awh1target=cutoff
)), where F is the free energy relative to the estimated global minimum. This provides a smooth switch of a flat target distribution in regions with free energy lower than the cutoff to a Boltzmann distribution in regions with free energy higher than the cutoff.

boltzmann
¶ The target distribution is a Boltzmann distribtution with a scaled beta (inverse temperature) factor given by
awh1targetbetascaling
. E.g., a value of 0.1 would give the same coordinate distribution as sampling with a simulation temperature scaled by 10.

localboltzmann
¶ Same target distribution and use of
awh1targetbetascaling
but the convergence towards the target distribution is inherently local i.e., the rate of change of the bias only depends on the local sampling. This local convergence property is only compatible withawh1growth=linear
, since forawh1growth=explinear
histograms are globally rescaled in the initial stage.


awh1targetbetascaling
¶ (0) For
awh1target=boltzmann
andawh1target=localboltzmann
it is the unitless beta scaling factor taking values in (0,1).

awh1targetcutoff
¶ (0) [kJ mol^{1}] For
awh1target=cutoff
this is the cutoff, should be > 0.

awh1userdata
¶ 
no
¶ Initialize the PMF and target distribution with default values.

yes
¶ Initialize the PMF and target distribution with user provided data. For
awhnbias
= 1, gmx mdrun will expect a fileawhinit.xvg
to be present in the run directory. For multiple biases, gmx mdrun expects filesawhinit1.xvg
,awhinit2.xvg
, etc. The file name can be changed with theawh
option. The firstawh1ndim
columns of each input file should contain the coordinate values, such that each row defines a point in coordinate space. Columnawh1ndim
+ 1 should contain the PMF value for each point. The target distribution column can either follow the PMF (columnawh1ndim
+ 2) or be in the same column as written by gmx awh.

Do not share the bias.
Share the bias and PMF estimates within and/or between simulations. Within a simulation, the bias will be shared between biases that have the same
awh1sharegroup
index (note that the current code does not support this). Withawhsharemultisim=yes
and gmx mdrun optionmultidir
the bias will also be shared across simulations. Sharing may increase convergence initially, although the starting configurations can be critical, especially when sharing between many biases. Currently, positive group values should start at 1 and increase by 1 for each subsequent bias that is shared.

awh1ndim
¶ (1) [integer] Number of dimensions of the coordinate, each dimension maps to 1 pull coordinate. The following options should be specified for each such dimension. Below only the options for dimension number 1 is shown. Options for other dimension indices are obtained by replacing ‘1’ by the dimension index.

awh1dim1coordprovider
¶ 
pull
¶ The module providing the reaction coordinate for this dimension. Currently AWH can only act on pull coordinates.


awh1dim1coordindex
¶ (1) Index of the pull coordinate defining this coordinate dimension.

awh1dim1forceconstant
¶ (0) [kJ mol^{1} nm^{2}] or [kJ mol^{1} rad^{2}] Force constant for the (convolved) umbrella potential(s) along this coordinate dimension.

awh1dim1start
¶ (0.0) [nm] or [rad] Start value of the sampling interval along this dimension. The range of allowed values depends on the relevant pull geometry (see
pullcoord1geometry
). For dihedral geometriesawh1dim1start
greater thanawh1dim1end
is allowed. The interval will then wrap around from +period/2 to period/2. For the direction geometry, the dimension is made periodic when the direction is along a box vector and covers more than 95% of the box length. Note that one should not apply pressure coupling along a periodic dimension.

awh1dim1end
¶ (0.0) [nm] or [rad] End value defining the sampling interval together with
awh1dim1start
.

awh1dim1diffusion
¶ (10^{5}) [nm^{2}/ps] or [rad^{2}/ps] Estimated diffusion constant for this coordinate dimension determining the initial biasing rate. This needs only be a rough estimate and should not critically affect the results unless it is set to something very low, leading to slow convergence, or very high, forcing the system far from equilibrium. Not setting this value explicitly generates a warning.

awh1dim1coverdiameter
¶ (0.0) [nm] or [rad] Diameter that needs to be sampled by a single simulation around a coordinate value before the point is considered covered in the initial stage (see
awh1growth=explinear
). A value > 0 ensures that for each covering there is a continuous transition of this diameter across each coordinate value. This is trivially true for independent simulations but not for for multiple biassharing simulations (awh1sharegroup
>0). For a diameter = 0, covering occurs as soon as the simulations have sampled the whole interval, which for many sharing simulations does not guarantee transitions across free energy barriers. On the other hand, when the diameter >= the sampling interval length, covering occurs when a single simulation has independently sampled the whole interval.
Enforced rotation¶
These mdp parameters can be used enforce the rotation of a group of atoms, e.g. a protein subunit. The reference manual describes in detail 13 different potentials that can be used to achieve such a rotation.

rotation
¶ 
no
¶ No enforced rotation will be applied. All enforced rotation options will be ignored (and if present in the mdp file, they unfortunately generate warnings).

yes
¶ Apply the rotation potential specified by
rottype0
to the group of atoms given under therotgroup0
option.


rotngroups
¶ (1) Number of rotation groups.

rotgroup0
¶ Name of rotation group 0 in the index file.

rottype0
¶ (iso) Type of rotation potential that is applied to rotation group 0. Can be of of the following:
iso
,isopf
,pm
,pmpf
,rm
,rmpf
,rm2
,rm2pf
,flex
,flext
,flex2
, orflex2t
.

rotmassw0
¶ (no) Use mass weighted rotation group positions.

rotvec0
¶ (1.0 0.0 0.0) Rotation vector, will get normalized.

rotpivot0
¶ (0.0 0.0 0.0) [nm] Pivot point for the potentials
iso
,pm
,rm
, andrm2
.

rotrate0
¶ (0) [degree ps^{1}] Reference rotation rate of group 0.

rotk0
¶ (0) [kJ mol^{1} nm^{2}] Force constant for group 0.

rotslabdist0
¶ (1.5) [nm] Slab distance, if a flexible axis rotation type was chosen.

rotmingauss0
¶ (0.001) Minimum value (cutoff) of Gaussian function for the force to be evaluated (for the flexible axis potentials).

roteps0
¶ (0.0001) [nm^{2}] Value of additive constant epsilon for
rm2*
andflex2*
potentials.

rotfitmethod0
¶ (rmsd) Fitting method when determining the actual angle of a rotation group (can be one of
rmsd
,norm
, orpotential
).

rotpotfitnsteps0
¶ (21) For fit type
potential
, the number of angular positions around the reference angle for which the rotation potential is evaluated.

rotpotfitstep0
¶ (0.25) For fit type
potential
, the distance in degrees between two angular positions.

rotnstrout
¶ (100) Output frequency (in steps) for the angle of the rotation group, as well as for the torque and the rotation potential energy.

rotnstsout
¶ (1000) Output frequency for perslab data of the flexible axis potentials, i.e. angles, torques and slab centers.
NMR refinement¶

disre
¶ 
no
¶ ignore distance restraint information in topology file

simple
¶ simple (permolecule) distance restraints.

ensemble
¶ distance restraints over an ensemble of molecules in one simulation box. Normally, one would perform ensemble averaging over multiple simulations, using
mdrun multidir
. The environment variableGMX_DISRE_ENSEMBLE_SIZE
sets the number of systems within each ensemble (usually equal to the number of directories supplied tomdrun multidir
).


disremixed
¶ 
no
¶ the violation used in the calculation of the restraint force is the timeaveraged violation

yes
¶ the violation used in the calculation of the restraint force is the square root of the product of the timeaveraged violation and the instantaneous violation


disrefc
¶ (1000) [kJ mol^{1} nm^{2}] force constant for distance restraints, which is multiplied by a (possibly) different factor for each restraint given in the
fac
column of the interaction in the topology file.

disretau
¶ (0) [ps] time constant for distance restraints running average. A value of zero turns off time averaging.

nstdisreout
¶ (100) [steps] period between steps when the running timeaveraged and instantaneous distances of all atom pairs involved in restraints are written to the energy file (can make the energy file very large)

orire
¶ 
no
¶ ignore orientation restraint information in topology file

yes
¶ use orientation restraints, ensemble averaging can be performed with
mdrun multidir


orirefc
¶ (0) [kJ mol^{1}] force constant for orientation restraints, which is multiplied by a (possibly) different weight factor for each restraint, can be set to zero to obtain the orientations from a free simulation

oriretau
¶ (0) [ps] time constant for orientation restraints running average. A value of zero turns off time averaging.

orirefitgrp
¶ fit group for orientation restraining. This group of atoms is used to determine the rotation R of the system with respect to the reference orientation. The reference orientation is the starting conformation of the first subsystem. For a protein, backbone is a reasonable choice

nstorireout
¶ (100) [steps] period between steps when the running timeaveraged and instantaneous orientations for all restraints, and the molecular order tensor are written to the energy file (can make the energy file very large)
Free energy calculations¶

freeenergy
¶ 
no
¶ Only use topology A.

yes
¶ Interpolate between topology A (lambda=0) to topology B (lambda=1) and write the derivative of the Hamiltonian with respect to lambda (as specified with
dhdlderivatives
), or the Hamiltonian differences with respect to other lambda values (as specified with foreign lambda) to the energy file and/or todhdl.xvg
, where they can be processed by, for example gmx bar. The potentials, bondlengths and angles are interpolated linearly as described in the manual. Whenscalpha
is larger than zero, softcore potentials are used for the LJ and Coulomb interactions.


expanded
¶ Turns on expanded ensemble simulation, where the alchemical state becomes a dynamic variable, allowing jumping between different Hamiltonians. See the expanded ensemble options for controlling how expanded ensemble simulations are performed. The different Hamiltonians used in expanded ensemble simulations are defined by the other free energy options.

initlambda
¶ (1) starting value for lambda (float). Generally, this should only be used with slow growth (i.e. nonzero
deltalambda
). In other cases,initlambdastate
should be specified instead. Must be greater than or equal to 0.

deltalambda
¶ (0) increment per time step for lambda

initlambdastate
¶ (1) starting value for the lambda state (integer). Specifies which columm of the lambda vector (
coullambdas
,vdwlambdas
,bondedlambdas
,restraintlambdas
,masslambdas
,temperaturelambdas
,feplambdas
) should be used. This is a zerobased index:initlambdastate
0 means the first column, and so on.

feplambdas
¶ [array] Zero, one or more lambda values for which Delta H values will be determined and written to dhdl.xvg every
nstdhdl
steps. Values must be between 0 and 1. Free energy differences between different lambda values can then be determined with gmx bar.feplambdas
is different from the other lambdas keywords because all components of the lambda vector that are not specified will usefeplambdas
(includingrestraintlambdas
and therefore the pull code restraints).

coullambdas
¶ [array] Zero, one or more lambda values for which Delta H values will be determined and written to dhdl.xvg every
nstdhdl
steps. Values must be between 0 and 1. Only the electrostatic interactions are controlled with this component of the lambda vector (and only if the lambda=0 and lambda=1 states have differing electrostatic interactions).

vdwlambdas
¶ [array] Zero, one or more lambda values for which Delta H values will be determined and written to dhdl.xvg every
nstdhdl
steps. Values must be between 0 and 1. Only the van der Waals interactions are controlled with this component of the lambda vector.

bondedlambdas
¶ [array] Zero, one or more lambda values for which Delta H values will be determined and written to dhdl.xvg every
nstdhdl
steps. Values must be between 0 and 1. Only the bonded interactions are controlled with this component of the lambda vector.

restraintlambdas
¶ [array] Zero, one or more lambda values for which Delta H values will be determined and written to dhdl.xvg every
nstdhdl
steps. Values must be between 0 and 1. Only the restraint interactions: dihedral restraints, and the pull code restraints are controlled with this component of the lambda vector.

masslambdas
¶ [array] Zero, one or more lambda values for which Delta H values will be determined and written to dhdl.xvg every
nstdhdl
steps. Values must be between 0 and 1. Only the particle masses are controlled with this component of the lambda vector.

temperaturelambdas
¶ [array] Zero, one or more lambda values for which Delta H values will be determined and written to dhdl.xvg every
nstdhdl
steps. Values must be between 0 and 1. Only the temperatures controlled with this component of the lambda vector. Note that these lambdas should not be used for replica exchange, only for simulated tempering.

calclambdaneighbors
¶ (1) Controls the number of lambda values for which Delta H values will be calculated and written out, if
initlambdastate
has been set. A positive value will limit the number of lambda points calculated to only the nth neighbors ofinitlambdastate
: for example, ifinitlambdastate
is 5 and this parameter has a value of 2, energies for lambda points 37 will be calculated and writen out. A value of 1 means all lambda points will be written out. For normal BAR such as with gmx bar, a value of 1 is sufficient, while for MBAR 1 should be used.

scalpha
¶ (0) the softcore alpha parameter, a value of 0 results in linear interpolation of the LJ and Coulomb interactions

scrpower
¶ (6) power 6 for the radial term in the softcore equation.
(48) (deprecated) power 48 for the radial term in the softcore equation. Note that scalpha should generally be much lower (between 0.001 and 0.003).

sccoul
¶ (no) Whether to apply the softcore free energy interaction transformation to the Columbic interaction of a molecule. Default is no, as it is generally more efficient to turn off the Coulomic interactions linearly before turning off the van der Waals interactions. Note that it is only taken into account when lambda states are used, not with
couplelambda0
/couplelambda1
, and you can still turn off softcore interactions by settingscalpha
to 0.

scpower
¶ (0) the power for lambda in the softcore function, only the values 1 and 2 are supported

scsigma
¶ (0.3) [nm] the softcore sigma for particles which have a C6 or C12 parameter equal to zero or a sigma smaller than
scsigma

couplemoltype
¶ Here one can supply a molecule type (as defined in the topology) for calculating solvation or coupling free energies. There is a special option
system
that couples all molecule types in the system. This can be useful for equilibrating a system starting from (nearly) random coordinates.freeenergy
has to be turned on. The Van der Waals interactions and/or charges in this molecule type can be turned on or off between lambda=0 and lambda=1, depending on the settings ofcouplelambda0
andcouplelambda1
. If you want to decouple one of several copies of a molecule, you need to copy and rename the molecule definition in the topology.

couplelambda0
¶ 
vdwq
¶ all interactions are on at lambda=0

vdw
¶ the charges are zero (no Coulomb interactions) at lambda=0

q
¶ the Van der Waals interactions are turned at lambda=0; softcore interactions will be required to avoid singularities

none
¶ the Van der Waals interactions are turned off and the charges are zero at lambda=0; softcore interactions will be required to avoid singularities.


couplelambda1
¶ analogous to
couplelambda1
, but for lambda=1

coupleintramol
¶ 
no
¶ All intramolecular nonbonded interactions for moleculetype
couplemoltype
are replaced by exclusions and explicit pair interactions. In this manner the decoupled state of the molecule corresponds to the proper vacuum state without periodicity effects.

yes
¶ The intramolecular Van der Waals and Coulomb interactions are also turned on/off. This can be useful for partitioning freeenergies of relatively large molecules, where the intramolecular nonbonded interactions might lead to kinetically trapped vacuum conformations. The 14 pair interactions are not turned off.


nstdhdl
¶ (100) the frequency for writing dH/dlambda and possibly Delta H to dhdl.xvg, 0 means no ouput, should be a multiple of
nstcalcenergy
.

dhdlderivatives
¶ (yes)
If yes (the default), the derivatives of the Hamiltonian with respect to lambda at each
nstdhdl
step are written out. These values are needed for interpolation of linear energy differences with gmx bar (although the same can also be achieved with the right foreign lambda setting, that may not be as flexible), or with thermodynamic integration

dhdlprintenergy
¶ (no)
Include either the total or the potential energy in the dhdl file. Options are ‘no’, ‘potential’, or ‘total’. This information is needed for later free energy analysis if the states of interest are at different temperatures. If all states are at the same temperature, this information is not needed. ‘potential’ is useful in case one is using
mdrun rerun
to generate thedhdl.xvg
file. When rerunning from an existing trajectory, the kinetic energy will often not be correct, and thus one must compute the residual free energy from the potential alone, with the kinetic energy component computed analytically.

separatedhdlfile
¶ 
yes
¶ The free energy values that are calculated (as specified with the foreign lambda and
dhdlderivatives
settings) are written out to a separate file, with the default namedhdl.xvg
. This file can be used directly with gmx bar.

no
¶ The free energy values are written out to the energy output file (
ener.edr
, in accumulated blocks at everynstenergy
steps), where they can be extracted with gmx energy or used directly with gmx bar.


dhhistsize
¶ (0) If nonzero, specifies the size of the histogram into which the Delta H values (specified with foreign lambda) and the derivative dH/dl values are binned, and written to ener.edr. This can be used to save disk space while calculating free energy differences. One histogram gets written for each foreign lambda and two for the dH/dl, at every
nstenergy
step. Be aware that incorrect histogram settings (too small size or too wide bins) can introduce errors. Do not use histograms unless you’re certain you need it.

dhhistspacing
¶ (0.1) Specifies the bin width of the histograms, in energy units. Used in conjunction with
dhhistsize
. This size limits the accuracy with which free energies can be calculated. Do not use histograms unless you’re certain you need it.
Expanded Ensemble calculations¶

nstexpanded
¶ The number of integration steps beween attempted moves changing the system Hamiltonian in expanded ensemble simulations. Must be a multiple of
nstcalcenergy
, but can be greater or less thannstdhdl
.

lmcstats
¶ 
no
¶ No Monte Carlo in state space is performed.

metropolistransition
¶ Uses the Metropolis weights to update the expanded ensemble weight of each state. Min{1,exp((beta_new u_new  beta_old u_old)}

barkertransition
¶ Uses the Barker transition critera to update the expanded ensemble weight of each state i, defined by exp(beta_new u_new)/(exp(beta_new u_new)+exp(beta_old u_old))

wanglandau
¶ Uses the WangLandau algorithm (in state space, not energy space) to update the expanded ensemble weights.

minvariance
¶ Uses the minimum variance updating method of Escobedo et al. to update the expanded ensemble weights. Weights will not be the free energies, but will rather emphasize states that need more sampling to give even uncertainty.


lmcmcmove
¶ 
no
¶ No Monte Carlo in state space is performed.

metropolistransition
¶ Randomly chooses a new state up or down, then uses the Metropolis critera to decide whether to accept or reject: Min{1,exp((beta_new u_new  beta_old u_old)}

barkertransition
¶ Randomly chooses a new state up or down, then uses the Barker transition critera to decide whether to accept or reject: exp(beta_new u_new)/(exp(beta_new u_new)+exp(beta_old u_old))

gibbs
¶ Uses the conditional weights of the state given the coordinate (exp(beta_i u_i) / sum_k exp(beta_i u_i) to decide which state to move to.

metropolizedgibbs
¶ Uses the conditional weights of the state given the coordinate (exp(beta_i u_i) / sum_k exp(beta_i u_i) to decide which state to move to, EXCLUDING the current state, then uses a rejection step to ensure detailed balance. Always more efficient that Gibbs, though only marginally so in many situations, such as when only the nearest neighbors have decent phase space overlap.


lmcseed
¶ (1) random seed to use for Monte Carlo moves in state space. When
lmcseed
is set to 1, a pseudo random seed is us

mctemperature
¶ Temperature used for acceptance/rejection for Monte Carlo moves. If not specified, the temperature of the simulation specified in the first group of
reft
is used.

wlratio
¶ (0.8) The cutoff for the histogram of state occupancies to be reset, and the free energy incrementor to be changed from delta to delta *
wlscale
. If we define the Nratio = (number of samples at each histogram) / (average number of samples at each histogram).wlratio
of 0.8 means that means that the histogram is only considered flat if all Nratio > 0.8 AND simultaneously all 1/Nratio > 0.8.

wlscale
¶ (0.8) Each time the histogram is considered flat, then the current value of the WangLandau incrementor for the free energies is multiplied by
wlscale
. Value must be between 0 and 1.

initwldelta
¶ (1.0) The initial value of the WangLandau incrementor in kT. Some value near 1 kT is usually most efficient, though sometimes a value of 23 in units of kT works better if the free energy differences are large.

wloneovert
¶ (no) Set WangLandau incrementor to scale with 1/(simulation time) in the large sample limit. There is significant evidence that the standard WangLandau algorithms in state space presented here result in free energies getting ‘burned in’ to incorrect values that depend on the initial state. when
wloneovert
is true, then when the incrementor becomes less than 1/N, where N is the mumber of samples collected (and thus proportional to the data collection time, hence ‘1 over t’), then the WangLambda incrementor is set to 1/N, decreasing every step. Once this occurs,wlratio
is ignored, but the weights will still stop updating when the equilibration criteria set inlmcweightsequil
is achieved.

lmcrepeats
¶ (1) Controls the number of times that each Monte Carlo swap type is performed each iteration. In the limit of large numbers of Monte Carlo repeats, then all methods converge to Gibbs sampling. The value will generally not need to be different from 1.

lmcgibbsdelta
¶ (1) Limit Gibbs sampling to selected numbers of neighboring states. For Gibbs sampling, it is sometimes inefficient to perform Gibbs sampling over all of the states that are defined. A positive value of
lmcgibbsdelta
means that only states plus or minuslmcgibbsdelta
are considered in exchanges up and down. A value of 1 means that all states are considered. For less than 100 states, it is probably not that expensive to include all states.

lmcforcednstart
¶ (0) Force initial state space sampling to generate weights. In order to come up with reasonable initial weights, this setting allows the simulation to drive from the initial to the final lambda state, with
lmcforcednstart
steps at each state before moving on to the next lambda state. Iflmcforcednstart
is sufficiently long (thousands of steps, perhaps), then the weights will be close to correct. However, in most cases, it is probably better to simply run the standard weight equilibration algorithms.

nsttransitionmatrix
¶ (1) Frequency of outputting the expanded ensemble transition matrix. A negative number means it will only be printed at the end of the simulation.

symmetrizedtransitionmatrix
¶ (no) Whether to symmetrize the empirical transition matrix. In the infinite limit the matrix will be symmetric, but will diverge with statistical noise for short timescales. Forced symmetrization, by using the matrix T_sym = 1/2 (T + transpose(T)), removes problems like the existence of (small magnitude) negative eigenvalues.

mininumvarmin
¶ (100) The minvariance strategy (option of
lmcstats
is only valid for larger number of samples, and can get stuck if too few samples are used at each state.mininumvarmin
is the minimum number of samples that each state that are allowed before the minvariance strategy is activated if selected.

initlambdaweights
¶ The initial weights (free energies) used for the expanded ensemble states. Default is a vector of zero weights. format is similar to the lambda vector settings in
feplambdas
, except the weights can be any floating point number. Units are kT. Its length must match the lambda vector lengths.

lmcweightsequil
¶ 
no
¶ Expanded ensemble weights continue to be updated throughout the simulation.

yes
¶ The input expanded ensemble weights are treated as equilibrated, and are not updated throughout the simulation.

wldelta
¶ Expanded ensemble weight updating is stopped when the WangLandau incrementor falls below this value.

numberalllambda
¶ Expanded ensemble weight updating is stopped when the number of samples at all of the lambda states is greater than this value.

numbersteps
¶ Expanded ensemble weight updating is stopped when the number of steps is greater than the level specified by this value.

numbersamples
¶ Expanded ensemble weight updating is stopped when the number of total samples across all lambda states is greater than the level specified by this value.

countratio
¶ Expanded ensemble weight updating is stopped when the ratio of samples at the least sampled lambda state and most sampled lambda state greater than this value.


simulatedtempering
¶ (no) Turn simulated tempering on or off. Simulated tempering is implemented as expanded ensemble sampling with different temperatures instead of different Hamiltonians.

simtemplow
¶ (300) [K] Low temperature for simulated tempering.

simtemphigh
¶ (300) [K] High temperature for simulated tempering.

simulatedtemperingscaling
¶ Controls the way that the temperatures at intermediate lambdas are calculated from the
temperaturelambdas
part of the lambda vector.
linear
¶ Linearly interpolates the temperatures using the values of
temperaturelambdas
, i.e. ifsimtemplow
=300,simtemphigh
=400, then lambda=0.5 correspond to a temperature of 350. A nonlinear set of temperatures can always be implemented with uneven spacing in lambda.

geometric
¶ Interpolates temperatures geometrically between
simtemplow
andsimtemphigh
. The i:th state has temperaturesimtemplow
* (simtemphigh
/simtemplow
) raised to the power of (i/(ntemps1)). This should give roughly equal exchange for constant heat capacity, though of course things simulations that involve protein folding have very high heat capacity peaks.

exponential
¶ Interpolates temperatures exponentially between
simtemplow
andsimtemphigh
. The i:th state has temperaturesimtemplow
+ (simtemphigh
simtemplow
)*((exp(temperaturelambdas
(i))1)/(exp(1.0)i)).

Nonequilibrium MD¶

accgrps
¶ groups for constant acceleration (e.g.
Protein Sol
) all atoms in groups Protein and Sol will experience constant acceleration as specified in theaccelerate
line

accelerate
¶ (0) [nm ps^{2}] acceleration for
accgrps
; x, y and z for each group (e.g.0.1 0.0 0.0 0.1 0.0 0.0
means that first group has constant acceleration of 0.1 nm ps^{2} in X direction, second group the opposite).

freezegrps
¶ Groups that are to be frozen (i.e. their X, Y, and/or Z position will not be updated; e.g.
Lipid SOL
).freezedim
specifies for which dimension(s) the freezing applies. To avoid spurious contributions to the virial and pressure due to large forces between completely frozen atoms you need to use energy group exclusions, this also saves computing time. Note that coordinates of frozen atoms are not scaled by pressurecoupling algorithms.

freezedim
¶ dimensions for which groups in
freezegrps
should be frozen, specifyY
orN
for X, Y and Z and for each group (e.g.Y Y N N N N
means that particles in the first group can move only in Z direction. The particles in the second group can move in any direction).

cosacceleration
¶ (0) [nm ps^{2}] the amplitude of the acceleration profile for calculating the viscosity. The acceleration is in the Xdirection and the magnitude is
cosacceleration
cos(2 pi z/boxheight). Two terms are added to the energy file: the amplitude of the velocity profile and 1/viscosity.

deform
¶ (0 0 0 0 0 0) [nm ps^{1}] The velocities of deformation for the box elements: a(x) b(y) c(z) b(x) c(x) c(y). Each step the box elements for which
deform
is nonzero are calculated as: box(ts)+(tts)*deform, offdiagonal elements are corrected for periodicity. The coordinates are transformed accordingly. Frozen degrees of freedom are (purposely) also transformed. The time ts is set to t at the first step and at steps at which x and v are written to trajectory to ensure exact restarts. Deformation can be used together with semiisotropic or anisotropic pressure coupling when the appropriate compressibilities are set to zero. The diagonal elements can be used to strain a solid. The offdiagonal elements can be used to shear a solid or a liquid.
Electric fields¶

electricfieldx
¶

electricfieldy
¶

electricfieldz
¶ Here you can specify an electric field that optionally can be alternating and pulsed. The general expression for the field has the form of a gaussian laser pulse:
\[E(t) = E_0 \exp\left[\frac{(tt_0)^2}{2\sigma^2}\right]\cos\left[\omega (tt_0)\right]\]For example, the four parameters for direction x are set in the fields of
electricfieldx
(and similar forelectricfieldy
andelectricfieldz
) likeelectricfieldx = E0 omega t0 sigma
with units (respectively) V nm^{1}, ps^{1}, ps, ps.
In the special case that
sigma = 0
, the exponential term is omitted and only the cosine term is used. If alsoomega = 0
a static electric field is applied.Read more at Electric fields and in ref. 146.
Mixed quantum/classical molecular dynamics¶

QMMM
¶ 
no
¶ No QM/MM.

yes
¶ Do a QM/MM simulation. Several groups can be described at different QM levels separately. These are specified in the
QMMMgrps
field separated by spaces. The level of ab initio theory at which the groups are described is specified byQMmethod
andQMbasis
Fields. Describing the groups at different levels of theory is only possible with the ONIOM QM/MM scheme, specified byQMMMscheme
.


QMMMgrps
¶ groups to be descibed at the QM level (works also in case of MiMiC QM/MM)

QMMMscheme
¶ 
normal
¶ normal QM/MM. There can only be one
QMMMgrps
that is modelled at theQMmethod
andQMbasis
level of ab initio theory. The rest of the system is described at the MM level. The QM and MM subsystems interact as follows: MM point charges are included in the QM oneelectron hamiltonian and all LennardJones interactions are described at the MM level.


QMmethod
¶ (RHF) Method used to compute the energy and gradients on the QM atoms. Available methods are AM1, PM3, RHF, UHF, DFT, B3LYP, MP2, CASSCF, and MMVB. For CASSCF, the number of electrons and orbitals included in the active space is specified by
CASelectrons
andCASorbitals
.

QMbasis
¶ (STO3G) Basis set used to expand the electronic wavefuntion. Only Gaussian basis sets are currently available, i.e.
STO3G, 321G, 321G*, 321+G*, 621G, 631G, 631G*, 631+G*,
and6311G
.

QMcharge
¶ (0) [integer] The total charge in
e
of theQMMMgrps
. In case there are more than oneQMMMgrps
, the total charge of each ONIOM layer needs to be specified separately.

QMmult
¶ (1) [integer] The multiplicity of the
QMMMgrps
. In case there are more than oneQMMMgrps
, the multiplicity of each ONIOM layer needs to be specified separately.

CASorbitals
¶ (0) [integer] The number of orbitals to be included in the active space when doing a CASSCF computation.

CASelectrons
¶ (0) [integer] The number of electrons to be included in the active space when doing a CASSCF computation.

SH
¶ 
no
¶ No surface hopping. The system is always in the electronic groundstate.

yes
¶ Do a QM/MM MD simulation on the excited statepotential energy surface and enforce a diabatic hop to the groundstate when the system hits the conical intersection hyperline in the course the simulation. This option only works in combination with the CASSCF method.

Computational Electrophysiology¶
Use these options to switch on and control ion/water position exchanges in “Computational Electrophysiology” simulation setups. (See the reference manual for details).

swapcoords
¶ 
no
¶ Do not enable ion/water position exchanges.

X ; Y ; Z
¶ Allow for ion/water position exchanges along the chosen direction. In a typical setup with the membranes parallel to the xy plane, ion/water pairs need to be exchanged in Z direction to sustain the requested ion concentrations in the compartments.


swapfrequency
¶ (1) The swap attempt frequency, i.e. every how many time steps the ion counts per compartment are determined and exchanges made if necessary. Normally it is not necessary to check at every time step. For typical Computational Electrophysiology setups, a value of about 100 is sufficient and yields a negligible performance impact.

splitgroup0
¶ Name of the index group of the membraneembedded part of channel #0. The center of mass of these atoms defines one of the compartment boundaries and should be chosen such that it is near the center of the membrane.

splitgroup1
¶ Channel #1 defines the position of the other compartment boundary.

masswsplit0
¶ (no) Defines whether or not massweighting is used to calculate the split group center.

no
¶ Use the geometrical center.

yes
¶ Use the center of mass.


masswsplit1
¶ (no) As above, but for splitgroup #1.

solventgroup
¶ Name of the index group of solvent molecules.

couplsteps
¶ (10) Average the number of ions per compartment over these many swap attempt steps. This can be used to prevent that ions near a compartment boundary (diffusing through a channel, e.g.) lead to unwanted back and forth swaps.

iontypes
¶ (1) The number of different ion types to be controlled. These are during the simulation exchanged with solvent molecules to reach the desired reference numbers.

iontype0name
¶ Name of the first ion type.

iontype0inA
¶ (1) Requested (=reference) number of ions of type 0 in compartment A. The default value of 1 means: use the number of ions as found in time step 0 as reference value.

iontype0inB
¶ (1) Reference number of ions of type 0 for compartment B.

bulkoffsetA
¶ (0.0) Offset of the first swap layer from the compartment A midplane. By default (i.e. bulk offset = 0.0), ion/water exchanges happen between layers at maximum distance (= bulk concentration) to the split group layers. However, an offset b (1.0 < b < +1.0) can be specified to offset the bulk layer from the middle at 0.0 towards one of the compartmentpartitioning layers (at +/ 1.0).

bulkoffsetB
¶ (0.0) Offset of the other swap layer from the compartment B midplane.

threshold
¶ (1) Only swap ions if threshold difference to requested count is reached.

cyl0r
¶ (2.0) [nm] Radius of the split cylinder #0. Two split cylinders (mimicking the channel pores) can optionally be defined relative to the center of the split group. With the help of these cylinders it can be counted which ions have passed which channel. The split cylinder definition has no impact on whether or not ion/water swaps are done.

cyl0up
¶ (1.0) [nm] Upper extension of the split cylinder #0.

cyl0down
¶ (1.0) [nm] Lower extension of the split cylinder #0.

cyl1r
¶ (2.0) [nm] Radius of the split cylinder #1.

cyl1up
¶ (1.0) [nm] Upper extension of the split cylinder #1.

cyl1down
¶ (1.0) [nm] Lower extension of the split cylinder #1.
Densityguided simulations¶
These options enable and control the calculation and application of additional forces that are derived from threedimensional densities, e.g., from cryo electronmicroscopy experiments. (See the reference manual for details)

densityguidedsimulationactive
¶ (no) Activate densityguided simulations.

densityguidedsimulationgroup
¶ (protein) The atoms that are subject to the forces from the densityguided simulation and contribute to the simulated density.

densityguidedsimulationsimilaritymeasure
¶ (innerproduct) Similarity measure between the density that is calculated from the atom positions and the reference density.

innerproduct
¶ Takes the sum of the product of reference density and simulated density voxel values.

relativeentropy
¶ Uses the negative relative entropy (or KullbackLeibler divergence) between reference density and simulated density as similarity measure. Negative density values are ignored.


densityguidedsimulationatomspreadingweight
¶ (unity) Determines the multiplication factor for the Gaussian kernel when spreading atoms on the grid.

unity
¶ Every atom in the density fitting group is assigned the same unit factor.

mass
¶ Atoms contribute to the simulated density proportional to their mass.

charge
¶ Atoms contribute to the simulated density proportional to their charge.


densityguidedsimulationforceconstant
¶ (1e+09) [kJ mol^{1}] The scaling factor for densityguided simulation forces. May also be negative.

densityguidedsimulationgaussiantransformspreadingwidth
¶ (0.2) [nm] The Gaussian RMS width for the spread kernel for the simulated density.

densityguidedsimulationgaussiantransformspreadingrangeinmultiplesofwidth
¶ (4) The range after which the gaussian is cut off in multiples of the Gaussian RMS width described above.

densityguidedsimulationreferencedensityfilename
¶ (reference.mrc) Reference density file name using an absolute path or a path relative to the to the folder from which gmx mdrun is called.

densityguidedsimulationnst
¶ (1) Interval in steps at which the density fitting forces are evaluated and applied. The forces are scaled by this number when applied (See the reference manual for details).

densityguidedsimulationnormalizedensities
¶ (true) Normalize the sum of density voxel values to one for the reference density as well as the simulated density.

densityguidedsimulationadaptiveforcescaling
¶ (false) Adapt the force constant to ensure a steady increase in similarity between simulated and reference density.

true
¶ Use adaptive force scaling.


densityguidedsimulationadaptiveforcescalingtimeconstant
¶ (4) [ps] Couple force constant to increase in similarity with reference density with this time constant. Larger times result in looser coupling.
User defined thingies¶

user1grps
¶

user2grps
¶

userint1 (0)
¶

userint2 (0)
¶

userint3 (0)
¶

userint4 (0)
¶

userreal1 (0)
¶

userreal2 (0)
¶

userreal3 (0)
¶

userreal4 (0)
¶ These you can use if you modify code. You can pass integers and reals and groups to your subroutine. Check the inputrec definition in
src/gromacs/mdtypes/inputrec.h