SourceForge.net Logo
MCCCS Towhee (towhee_input Version 7.2.x)

 

 

Overview
    This section covers the variables that are set in the towhee_input file Version 7.2.x. Each variable is listed along with its type (logical, character, integer, or double precision). towhee_input is the main input file for Towhee and is generally the only file that needs to be edited on a regular basis. It has a regimented style to the input. The variables are described here in the order they appear in this file. Please look at one of the example files (available with the code package) for the precise file format.

    Note that for each variable listed below you must include the name of the variable on the previous line. Any spaces or tabs prior to the first non-white-space character of the variable name are ignored when the code checks for variable name matches.

    All Towhee parameter files (including towhee_input) allow internal comments for user convenience. Such comments are lines which begin with the '#' character; the entire line is then ignored.

    Some variables are optional. If the variable name is not included in towhee_input it will be set to the specified default value. Optional input file values are marked as optional parameter.
Bug reports and feature enhancements for 7.2.x versions
  • 7.2.0: Rebuilt the Arbitrary Trial Distribution setup for Angle B terms. This resolved a bug related to Angle B distributions that had flat portions of the distribution (typically, a range of zero probability for angle potentials that contain a nonbonded term). Arbitrary trial distributions for the angle B term that use a Gaussian distribution now only use a random number to select a peak from the distribution if there is more than one peak to choose from. Previously, a random number was generated no matter the number of peaks. This means the new code will not follow the exact same random number path as previously, but otherwise this is simply a time saving improvement. Rebuilt configurational-bias to properly handle flexible all-atom three membered rings. Previous versions were not properly accounting for all of the bending angles and torsional interactions related to the three membered ring generation. Added the MMFF94 Ethyleneoxide example to test flexible three membered all-atom rings and the buffered 14-7 potential. Revisited the MMFF94 force field implementation and discovered several mistakes in the previous implementation of that force field. This implementation has not yet been validated for the entire test suite provided in the original papers, but has now at least passed a few of those test cases. Resolved a bug that occured when peforming CBMC growth of ring structures with a forcefield that includes bond-bond terms. Resolved a bug where regrowing cyclic molecules in a very small box would sometimes result in a failure in the biasing term due to the method used to set the range of the biasing interaction Gaussian for ring-closure. Combined the old volnvt.F and volnpt.F files into the new volumemoves.F and began streamlining those subroutines. Removed an unneeded random number call from the volumes moves so this changed the random number sequence (and therefore results) of the examples. Moved the FENE_Hexamer example from the Canonical Ensemble directory to the more appropriate Gibbs_Ensemble directory. Added some of the variables required for an additional thermodyanmic pressure calculation, but that option is not yet fully functional is not suggested for use.
towhee_input file differences from version 7.2.x
  • Added two new variables to control configurational-bias Monte Carlo move options: itbang and orderbang.
  • Added new features related to the computation of pressure during a simulation. The old pressurefreq variable was split into pressure_virial_freq and pressure_thermo_freq. This additional thermodynamic pressure calculation is not fully functional in version 7.2.0 and should be set to 0.
Quick links

Links to other towhee_input information pages, described elsewhere on this page:

  • Nonbonded potential types are described in the classical_potential section.
  • Information used to describe molecule-specific forcefield information is in the input_style section.
  • Various constants declared at compile-time found in preproc.h.

Variable explanations for towhee_input
    inputformat (character string)
    • 'Towhee' : reads in the input variables following the format for Towhee. This format is described in this file.
    • 'LAMMPS' : reads in the input variables from the lammps_input and lammps_data files. Outputs files suitable for use with Towhee.
    random_number_generator (character string of size 20) optional parameter
    • 'DX-1597-2-7' : (default value) uses the DX-1597-2-7 pseudorandom number generator described in Deng 2005 with a period of approximately 1014903.
    • 'KISS99' : uses a version of the Keep it Simple Stupid pseudorandom number generator originally posted to an internet newsgroup by G. Marsaglia. This has a period of approximately 109.
    • 'MRG32k3a' : uses the MRG32k3a pseudorandom number generator described in l'Ecuyer 1999 with a period of approximately 1057.
    • 'RANLUX' : uses the RANLUX psuedorandom number generator as described by Lüscher 1994 and implemented by James 1994 with a period of approximately 10171.
      When using this setting the following optional parameter is available
        random_luxlevel (integer) optional parameter
          An integer value that defines the luxury level used by the pseudorandom number generator. See James 1994 for more information about the luxury level.
        • 0: Does not discard any pseudorandom numbers from the sequence.
        • 1: Periodically discards 24 pseudorandom numbers from the sequence.
        • 2: Periodically discards 73 pseudorandom numbers from the sequence.
        • 3: (default value) Periodically discards 199 pseudorandom numbers from the sequence.
        • 4: Periodically discards 365 pseudorandom numbers from the sequence.
    random_seed (integer) optional parameter
    • The 32 bit integer seed that is used to initialize the psuedorandom number generator when not performing a restart. Must be positive. Default value is 1302002.
    random_allow_restart (logical) optional parameter
    • .TRUE. (default value) if you want to restart the pseudorandom number generator from the sequence of integers in towhee_initial if possible. This is done when the simulation is restarting from a towhee_initial file generated by Version 5.0.0 or later. If this is not possible then Towhee will initialize the pseudorandom number generator using the random_seed.
    • .FALSE. if you always want to initialize the pseudorandom number generator using the random_seed.
    ensemble (character string of size 30)
    • 'npt' : Isobaric-Isothermal Ensemble. The volume moves for each simulation box are performed in an exchange with an external pressure bath set at a specified pressure. The total number of molecules and the temperature are conserved. See McDonald 1972 for the initial Monte Carlo molecular simulation work using the isobaric-isothermal ensemble. Combining this ensemble with any of the two-box molecule transfer moves turns it into the NPT-Gibbs ensemble. See Panagiotopoulos 1987, Panagiotopoulos et al. 1988, and Smit et al. 1989 for the initial Monte Carlo molecular simulation work using the Gibbs ensemble.
    • 'nvt' : Canonical Ensemble. The total volume of the system, The total number of each type of molecule in the system, and the temperature are conserved. This is the prototypical ensemble for Monte Carlo molecular simulation and was first used in that context by Metropolis et al. 1953. Combining this ensemble with any of the two-box molecule transfer moves turns it into the NVT-Gibbs ensemble. See Panagiotopoulos 1987, Panagiotopoulos et al. 1988, and Smit et al. 1989 for the initial Monte Carlo molecular simulation work using the Gibbs ensemble.
    • 'uvt' : Grand Canonical Ensemble. The total volume of the system is conserved. The total number of molecules in the system equilibrates with an external ideal gas bath set at a specified chemical potential.
    temperature (double precision)
    • The temperature in Kelvin.

    The variable in this subsection is only included in the input file if ensemble is set to 'npt'
    pressure (double precision)
    • The external pressure in kPa.
    End of the subsection only included if ensemble is 'npt'

    The variable in this subsection is only included in the input file if ensemble is set to 'uvt'
    tmmc_flag (logical) optional parameter
    • .TRUE. if you want to perform transition matrix Monte Carlo (TMMC) on a single component system in the grand canonical ensemble. This will require other variables to be specified below.
    • .FALSE. if you do not want to perform transition matrix Monte Carlo (TMMC). This is the default value if the parameter is not specified.
    End of the subsection only included if ensemble is 'uvt'

    nmolty (integer)
    • The total number of molecule types in the simulation. This must be less then or equal to NTMAX.
    nmolectyp (integer) [one value for each molecule type]
    • The number of molecules of each type (listed sequentially on a single line). For the constant N ensembles (nvt, npt) this is the actual number of molecules in the simulation. For the constant chemical potential ensembles (uvt) this is the maximum number of molecules allowed in the simulation.

    The variable in this subsection is only included in the input file if ensemble is set to 'uvt'
    chempot (double precision)
    • The real chemical potential (this includes intramolecular portions and is identical to the CB chemical potential output by the code) for molecules of each type (listed sequentially on a single line). The units are in Kelvin (identical to the output CB chemical potential).
    End of the subsection only included if ensemble is 'uvt'

    numboxes (integer)
    • The number of simulation boxes in the system. This value must be less than or equal to MAXBOX Note that many of the variables below depend upon numboxes as information is required based on the number of simulation boxes (such as box lengths) and some Monte Carlo moves are only valid for multiple box ensembles.
    stepstyle (character string of length 10)
    • 'cycles' : Run a Monte Carlo simulation for nstep Monte Carlo cycles. A cycle is equal to N Monte Carlo moves, where N is the number of molecules in the system.
    • 'moves' : Run a Monte Carlo simulation for nstep Monte Carlo moves.
    nstep (integer)
    • The number of Monte Carlo steps to perform where each step is either a full Monte Carlo cycles (if stepstyle is 'cycles') or a single move (if stepstyle is 'moves')
    controlstyle (character*20) optional parameter
    • 'equilibration' : designed for use when you are first setting up a simulation and still need to rapidly equilibrate all of the maximum update values. This option sets the following parameters based upon the value of nstep.
      • printfreq = nstep/10
      • blocksize = nstep/5
      • moviefreq = nstep/5
      • backupfreq = nstep/5
      • runoutput = 'full'
      • pdb_output_freq = nstep/5
      • loutdft = .FALSE.
      • loutlammps = .FALSE.
      • loutdlpoly = .FALSE.
      • louthist = .FALSE.
      • pressure_virial_freq = 10
      • pressure_thermo_freq = 0
      • trmaxdispfreq = nstep/100
      • volmaxdispfreq = nstep/100
    • 'production' : designed for use during production runs. This option sets the following parameters based upon the value of nstep.
      • printfreq = nstep/10
      • blocksize = nstep/5
      • moviefreq = nstep/20
      • backupfreq = nstep/5
      • runoutput = 'full'
      • pdb_output_freq = nstep/5
      • loutdft = .FALSE.
      • loutlammps = .FALSE.
      • loutdlpoly = .FALSE.
      • louthist = .FALSE.
      • pressure_virial_freq = 10
      • pressure_thermo_freq = 10
      • trmaxdispfreq = nstep/10
      • volmaxdispfreq = nstep/10
    • 'manual' : This is the default setting if no other value of controlstyle is specified. When using this option you must also include the following variables
        printfreq (integer) optional parameter
        • The step frequency for outputting information about the system to standard output (fortran unit 6). The information is the number of Monte Carlo steps performed thus far during the run, the total energy in each box, the x-box length of each box, the pressure of each box, and the number of molecules of each type in each box. If printfreq = 0, or if it is not specified, there will be no output.

        Variables in this subsection only need to be specified if tmmc_flag is .TRUE..
        n_tmmc_min (integer) optional
        • Minimum number of molecules allowed in the simulation box. By default, this is set to 0. Specifying a non-zero value is useful when sampling specific molecule number ranges. The maximum number is set to nmolty.
        weight_freq (integer)
        • The step frequency for updating the biasing function for transition-matrix Monte Carlo.
        c_matrix_freq (integer)
        • The step frequency for writing to file the collection matrix for TMMC. A new file indicating the step number will be created.
        run_name (character)
        • File prefix for TMMC output. The current estimate of the natural logarithm of the particle number probability distribution is written to a file "run_name.tmmc_weights.dat" every time the biasing function is updated. Collection matrix info is written to a file "run_name.c.stepnumber.dat". Accumulated semigrand potential energies are written to a file "run_name.vsg.stepnumber". This information is useful if one is interested in the average potential energy as a function density.
        in_c_flag (logical) optional parameter
        • By default, set to .FALSE. Set this variable to .TRUE. if an initial collection matrix is to be used to initiate the TMMC simulation. In the future, this will be modified so that either a collection matrix or biasing function can be read in. This option can be used to for continuing a TMMC simulation.
        in_cfile (character) optional parameter
        • Name of the collection matrix input file to be read in. Only needed if in_c_flag is .TRUE..
        End of variables subsection when tmmc_flag is .TRUE..

        blocksize (integer) optional parameter
        • The size of the blocks for computing block averages. If you want this to be meaningful then blocksize should divide cleanly into nstep. The quantities that are averaged (in each simulation box) are the specific density, the pressure, all of the energy terms, the chemical potential of each molecule type, number density of each molecule type, and the mole fractions. If blocksize=0, or if it is not specified, no block output takes place.
        moviefreq (integer) optional parameter
        • The step frequency for outputting the system conformations to the towhee_movie file. This file is analyzed after the run using the analyze_movie.F routine to compute a variety of distribution functions. This file can get pretty big if you output frequently so be careful if you have a limited amount of hard disk space available. If moviefreq=0, or if it is not specified, this file is not written.
        backupfreq (integer) optional parameter
        • The step frequency for writing a file named towhee_backup that is suitable for use as a restart file. It overwrites the previous version of towhee_backup each time so it does not take up much space. Typically I set backupfreq so that I get around 10 backups during a run. For more information about restart files look at the manual entries for towhee_initial, towhee_backup, and towhee_final. If backupfreq=0, or if it is not specified, this file is not written.
        restartfreq (integer) optional parameter
        • The step frequency for writing a file named towhee_restart_NNN that is suitable for use as a restart file, where NNN represents the step number at which this file is written. This file is identical to towhee_backup (see backupfreq above), except that it is not overwritten by subsequent restart files, and allows for restarts from multiple points along a run. If restartfreq=0, or if it is not specified, this file is not written.
        runoutput (character*20)
        • 'full' : if you want information about the individual blocks of the block averages and information about the maximum displacement updates.
        • 'blocks' : if you want information about the individual blocks of the block averages and don't want information about the maximum displacement updates.
        • 'updates' : if you don't want information about the individual blocks of the block averages and do want information about the maximum displacement updates.
        • 'none' : if you don't want information about the individual blocks of the block averages or information about the maximum displacement updates.
        pdb_output_freq (integer) optional parameter
        • The step frequency for outputting a snapshot of the simulation to a pdb file named box_xx_step_yyyyyyyyyyyyyy.pdb where xx is the box number converted into a 2 character string and yyyyyyyyyyyyyy is the step number converted into a 14 character string. If you do not wish to output any pdb files then you can set pdb_output_freq to 0 to disable this feature.
        loutdft (logical) optional parameter
        • .TRUE. if you wish to output files for use with the Tramonto classical density functional theory code. This outputs dft_surfaces.dat and dft_decode.dat. See the Tramonto manual for information about these files.
        • .FALSE. (default value) if you do not want to output dft files.
        loutlammps (logical) optional parameter
        • .TRUE. if you wish to output files for use with the LAMMPS massively parallel molecular dynamics code. This outputs lammps_input and lammps_data# where the number is each of the simulation box numbers. See the LAMMPS manual for more information on how to read in these files.
        • .FALSE. (default value) if you do not want to output LAMMPS files.
        loutdlpoly (logical) optional parameter
        • .TRUE. if you wish to output files for use with the DL_POLY molecular dynamics code. This outputs CONFIG# and FIELD# where the number is each of the simulation box numbers. See the DL_POLY manual for more information about these files.
        • .FALSE. (default value) if you do not want to output DL_POLY files.

        The variables in this subsection are only included if ensemble is 'uvt'
        louthist (logical) optional parameter
        • .FALSE. (default value) will not output files for histogram reweighting. No additional variables are required for this setting.
        • .TRUE. will output files used for histogram reweighting. When set to this value you must also include the following additional variables
            hist_label (integer)
            • An integer that is turned into a character string that creates the X portion of the named towhee_hisXY.dat file that is used to output information for histogram reweighting.
            hist_suffix (character*1)
            • A single character that creates the Y portion of the named towhee_hisXY.dat file that is used to output information for histogram reweighting.
            hist_nequil (integer)
            • The number of steps (cycles or moves) that are disregarded for the purposes of outputting histogram information. Set to 0 if you wish to use all of the steps for computing the histogram, and set to a positive number in order to discard those initial steps from the histogram.
            histcalcfreq (integer)
            • The step frequency for computing the information needed for histogram reweighting.
            histdumpfreq (integer)
            • The step frequency for outputting the information needed for histogram reweighting to the various towhee_histogram files. The ratio of histdumpfreq/histcalcfreq must be less than NDUMPHIST.
        End of the subsection only included if ensemble is 'uvt'

        pressure_virial_freq (integer) optional parameter
        • The step frequency for computing the pressure using a deterministic algorithm that is based upon the interaction of pairs of molecules in each simulation box. This is performed using the molecular version of the pressure virial (for continuous potentials) or the radial pressure (for discontinuous potentials). See the Algorithms: Pressure manual page for more information about the pressure calculation methods implemented into Towhee. Computing the pressure using these algorithms is a relatively expensive task (compared to the cost of a Monte Carlo move), especially for large systems. Setting pressure_virial_freq to zero to disables this calculation. If the parameter is not specified, the default value is 0.
        pressure_thermo_freq (integer) optional parameter
        • The step frequency for computing the pressure by sampling the average energy change/volume change ratio to compute the "thermodynamic" pressure each simulation box. Setting pressure_thermo_freq to zero to disables this calculation. If the parameter is not specified, the default value is 0.
        pressure_thermo_style (character*20) optional parameter
        • The style of attempted volume change to use when computing the thermodynamic pressure via additional ghost volume changes.
          • 'range': attempts a ghost volume change of size Volume * Uniform(pressure_thermo_range(0),pressure_thermo_range(1)).
          • 'move data': (default value) attempts volume changes in the exact same manner as performing an actual volume move.
        pressure_thermo_range (double precision) optional parameter
        • The range for the attempted ghost volume changes used when computing the thermodynamic pressure. Default values are (-0.01d0,0.01d0).
        trmaxdispfreq (integer) optional parameter
        • The step frequency for updating the maximum translational (atom and center-of-mass) and rotational displacements. They are adjusted to try and achieve the target acceptance rates (see tatraa, tatrac, and tarot). It is a good idea to do this fairly frequently at the start of the simulation (every step or every 10 steps) in order to get good values for the maximum displacements. Once the acceptance rates are near their desired values then reset trmaxdispfreq to do 10 updates during a run. If trmaxdispfreq is 0, or if it is not specified, no updates take place.
        volmaxdispfreq (integer) optional parameter
        • The step frequency for updating the maximum volume displacements. They are adjusted to try and achieve the target acceptance rates (see tavol). It is a good idea to do this fairly frequently at the start of the simulation (every few steps) in order to get good values for the maximum displacements. Once the acceptance rates are near their desired values I typically set volmaxdisp to do 10 updates during a run. If volmaxdispfreq is set to 0, or if it is not specified, no such updates take place.
    chempotperstep (integer) optional parameter
    • The number of additional trial insertions to perform in each box for at the end of every Monte Carlo step (listed sequentially for each molecule type on a single line). This allows the measurement of chemical potential in ensembles that do not have an insertion and deletion move (such as canonical and isobaric-isothermal). Default value is 0 for all molecule types
    potentialstyle
    • 'internal' : uses energy routines internal to the Towhee software package to describe the energies between atoms. When using this option you must include the following variables.
        ffnumber (integer)
        • 1 or more: reads the force field information from this number of file(s) listed in the ff_filename.
        ff_filename (formatted character*70) [one line for each force field]
        • A list of the filenames (one per line) that contain the force field information. On most systems you can just list this directory and then end the line. However, if you have trouble then adding sufficient blank spaces to the end of the line to get up to 70 characters could resolve your problem.
        classical_potential (character*30)
          The setting for this variable controls the nonbonded potential type. Depending on the setting there are a number of other variables that are also required. Please see the classical_potential web page for more information.
        electrostatic_form (character*50)
        • 'none' : no electrostatic potential
        • 'coulomb' : use Coulomb's law for the electrostatic potential. This option requires the following variables.
            coulombstyle (character*20)
            • 'ewald_fixed_kmax' if you want to use point charges with an Ewald sum that utilizes a constant number of inverse space vectors (kmax) and a variable electrostatic cutoff (rcelect) equal to half the current box length. When using this option you also need the following variables.
                kalp (double precision)
                • Value used in the Ewald sum to compute alpha. The actual Ewald sum alpha term is equal to kalp divided by the shortest box length. The recommended value for kalp is 5.6.
                kmax (integer)
                • Maximum number of inverse space vectors to use in any dimension for the Ewald sum. Recommended value of this parameter is 5. If you want to set this to a larger value to may have to increase VECTORMAX.
                dielect (double precision)
                • The dielectric constant used when computing coulombic interactions. Generally this should be set to 1.0 as the solvated system will act as the screening that the dielectric constant is intended to represent. If you are performing a simulation without any solvent (for example a protein without the water) you might want to set this value to the dielectric constant of the missing solvent.
            • 'ewald_fixed_cutoff' if you want to use point charges with an Ewald sum that utilizes a constant electrostatic cutoff (rcelect) and adjusts the number of inverse space vectors (kmax) according to the following heuristic.
              alpha = ( 1.35 - 0.15 log[ewald_prec]) / rcelect
              kmax = ( alpha * Max[box length] / π) * (-log[ewald_prec])0.5
              When using this option you will also need to list the following variables.
                ewald_prec (double precision)
                • Controls the precision of the Ewald summation technique. The smaller the value, the better the results (and the more expensive the simulation). The recommended value of 1d-4 is generally adequate, while a value of 1d-5 is very good (but more expensive).
                rcelect (double precision)
                • The cutoff for electrostatic interactions computed in the "real space" portion of the Ewald sum. Decreasing this value means less work in the "real space", but correspondingly more work in the "inverse space". Setting this equal to the general nonbonded cutoff (see rcut in classical_potential) is recommended.
                dielect (double precision)
                • The dielectric constant used when computing coulombic interactions. Generally this should be set to 1.0 as the solvated system acts as the screening that the dielectric constant is intended to represent. If you are performing a simulation without any solvent (for example a protein without the water) you could set this value to the dielectric constant of the missing solvent.
            • 'minimum image' uses the minimum image convention to compute the coulombic interactions between all pairs of atoms in a system. Please note that this option is implemented mainly to compute single-molecule energies for test systems and is not suggested for routine use in periodic systems. For a discussion of the perils of using cut-off methods for Coulombic interactions please see Hummer et al. 1997. When using this option you must list the following variable.
                dielect (double precision)
                • The dielectric constant used when computing coulombic interactions. Generally this should be set to 1.0 as the solvated system acts as the screening that the dielectric constant is intended to represent. If you are performing a simulation without any solvent (for example a protein without the water) you could set this value to the dielectric constant of the missing solvent.
        max_bond_length (double precision) optional parameter
        • The maximum allowed value for any bond length in the simulation. Any bond lengths greater than this value are given an infinite energy regardless of the actual form of the bond length potential. This value is also used in order to normalize the bond length distribution in a configurational-bias move and therefore will affect the measured chemical potential. If no value is specified for this optional parameter then the default value of 3.0 is used instead.
        nfield (integer) optional parameter
          The number of external fields to apply in the simulation. These fields can take on a variety of forms, but are always applied relative to a plane in one of the simulation boxes. Typical uses are for simulating the effect of a rigid surface without having to treat the surface atoms explicitly. If this value is not specified then it is set to the default value of 0. For nonzero values of nfield, the following variables are required for for each field.
          fieldtype (character*20)
          • 'Hard Ball' : Places a hard ball of a specified diameter at a particular location in one of the boxes. This ball interacts with each atom in the simulation and excludes all atom centers from its radius. When using this option you must also specify the following variables.
              hrbbox (integer)
              • The number of the simulation box which contains this hard ball. Must range from 1 to numboxes.
              hrbpos (double precision)
              • Position of the center of the hard ball. Must be between 0.0 and the box length of each of the 3 axes. All three coordinates are listed on the same line.
              hrbrad (double precision)
              • Radius of the hard ball. The ball is felt through the periodic boundaries.
              hrbfac (double precision)
              • Factor used to rescale the size of the hard ball. A value of 1.0 will use the hrbrad value without any adjustment. This factor is multiplied by the hrbrad value to compute the true radius of the hard ball.
              hrb_energy_type (character*11)
              • 'infinite' : for a hard overlap with the ball of infinite energy
              • 'finite' : for a specfic finite penalty for each atom that is inside of the hard ball. When using this option you also need the following variable.
          • 'Hard Wall' : Places a hard wall of a specified diameter in one of the boxes. This wall interacts with each atom in the simulation. When using this option you must also specify the following variables.
              hrdbox (integer)
              • The number of the simulation box which contains this hard wall. Must range from 1 to numboxes.
              hrdxyz (character*1)
              • 'x' : hard wall is perpendicular to the x-axis (in the yz plane)
              • 'y' : hard wall is perpendicular to the y-axis (in the xz plane)
              • 'z' : hard wall is perpendicular to the z-axis (in the xy plane)
              hrdcen (double precision)
              • Position of the center of the hard wall. Must be between 0.0 and the box length of the axis that is perpendicular to the wall.
              hrdrad (double precision)
              • Radius of the hard wall. The wall is felt through the periodic boundaries.
              hrd_repulsion_style (character*11)
              • 'centers' : the center of each atom in the system is the portion that interacts with the hard wall region. In other words, the hard wall interacts with the atoms as if they were point particles.
              • 'hard radii' : the atoms are considered to have a radius that is determined by their nonbonded parameters. The distance between the surface of the hard wall and the closest approach of the atomic radius is used to determine the interactions between the wall and the atoms. This option is not currently supported for all choices of the classical_potential.
              hrd_energy_type (character*11)
              • 'infinite' : any molecule inside of the hard wall has an infinite energy (hard overlap).
              • 'finite' : any molecule inside of the hard wall has a finite energy that is specified by the hrd_wall_energy variable. If you use this option you must include the following variable
                  hrd_wall_energy (double precision)
                  • The energy given to any atom that is inside of the hard wall (in Kelvin). This option is designed to enable equilibrium of a hard wall system as this provides an incentive for molecules to leave the hard wall area without causing a simulation ending overlap.
          • 'Harmonic Attractor' : Uses a harmonic potential to root certain atoms to a defined point or their initial positions.
              hafbox (integer)
              • The number of the simulation box in which this harmonic attractor is employed. Must range from 1 to numboxes.
              hafk (double precision)
              • The force constant for the harmonic potential.
              hafentries (integer)
              • The number of types or elements to which this field is applied.
              hafrefpos (character*7)
              • 'Global' : Uses a global set of coordinates for each atom. When using this option you also need to include the following variable.
                  hafglobxyz (double precision array)
                  • The x,y, and z coordinates of the global position. This should be entered all on the same line just separated by spaces.
              • 'Initial' : Uses the initial coordinates of each atom.
              hafkey (character*7)
              • 'Element' : Allows the user to chose to apply this field to a specific group of atoms which are all the same type of element. The following variables must be included for each entry.
              • 'FFtype' : Allows the user to chose to apply this field to a specific group of atoms which are all the same nonbond type. The following variables must be included for each entry.
          • 'Hooper Umbrella' : Places a Hooper Umbrella field (see Hooper et al. 2000) in a simulation box. This is a 4th power energy function based on displacement along a single axis.
            v(d) = umba * [ (d - umbcenter) / umbcenter ]
            With this option you must also specify the following variables.
              umbbox (integer)
              • The number of the simulation box which contains this Umbrella field. Must range from 1 to numboxes.
              umbxyz (character*1)
              • 'x' : Field is perpendicular to the x-axis (in the yz plane)
              • 'y' : Field is perpendicular to the y-axis (in the xz plane)
              • 'z' : Field is perpendicular to the z-axis (in the xy plane)
              umbcenter (double precision)
              • The zero energy point of the field, listed as a distance in Angstroms along the axis specified in umbxyz
              umba (double precision)
              • The energy constant in units of K/kB.
          • 'LJ 9-3 Wall' : Places a 9-3 Lennard-Jones wall in one of the simulation boxes. The wall potential has the following functional form.
            v(d) = [ 2/3 π εwf σwf3 ρwall ] * [ 2/15 (σwf/d)9 - (σwf/d)3 ]
            With this option you must also specify the following variables.
              ljfbox (integer)
              • The number of the simulation box which contains this Lennard-Jones wall. Must range from 1 to numboxes.
              ljfxyz (character*1)
              • 'x' : Lennard-Jones wall is perpendicular to the x-axis (in the yz plane)
              • 'y' : Lennard-Jones wall is perpendicular to the y-axis (in the xz plane)
              • 'z' : Lennard-Jones wall is perpendicular to the z-axis (in the xy plane)
              ljfcen (double precision)
              • Position of the center of the Lennard-Jones wall. Must be between 0.0 and the box length of the axis that is perpendicular to the wall.
              ljfdir (integer)
              • -1: Atoms only interact with the "left" face of this wall. This does not extend through the periodic boundary.
              • 1: Atoms only interact with the "right" face of this wall. This does not extend through the periodic boundary.
              ljfcut (double precision)
              • The distance beyond which the wall-atom interactions are not computed and assumed to be zero.
              ljfshift (logical)
              • T: if you want to shift the lj wall potential to be zero at the cutoff.
              • F: if you do not want to shift the potential.
              ljfrho (double precision)
              • The number density of atoms in the integrated wall potential (units of atoms per cubic Angstrom).
              ljfntypes (integer)
              • The number of atom types in the system that interact with the wall. Any atom type not listed here will have zero interaction with the wall. For each value of ljfntypes you must list the following variables.
                  ljfname (character*6)
                  • The name of the atom. This must match up with the atom names listed in the inpust_style=2 portion of each molecule that is interacting with this wall. If you are not using that input_style this will still work except you will need to know the atom names in the appropriate towhee_ff_* files.
                  ljfsig (double precision)
                  • σ parameter for the interaction between this atom and the wall atoms. Units are Angstroms.
                  ljfeps (double precision)
                  • ε parameter for the interaction between this atom and the wall atoms. Units are K/kB.
          • 'Steele Wall' : Places a 10-4 Lennard-Jones wall in one of the simulation boxes. The wall potential has the following functional form.
            v(z) = εw [ 2/5 (σsf/z)10 - (σsf/z)4 - σsf4 / [ 3 Δ ( z + 0.61 Δ )3 ] ]
            where
            εw = 2 π εsf ρs σsf2 Δ
            This potential is attributed to Steele 1973, but I found that reference a bit confusing so I implemented the equations as presented in Lastoskie et al. 1993 and the variable names here follow the notation in that paper.
            With this option you must also specify the following variables.
              steele box (integer)
              • The number of the simulation box which contains this Steele wall. Must range from 1 to numboxes.
              steele xyz (character*1)
              • 'x' : wall is perpendicular to the x-axis (in the yz plane)
              • 'y' : wall is perpendicular to the y-axis (in the xz plane)
              • 'z' : wall is perpendicular to the z-axis (in the xy plane)
              steele surface (double precision)
              • Position of the surface of the wall. Must be between 0.0 and the box length of the axis that is perpendicular to the wall.
              steele dir (integer)
              • -1: Atoms only interact with the "left/bottom" face of this wall. This does not extend through the periodic boundary.
              • 1: Atoms only interact with the "right/top" face of this wall. This does not extend through the periodic boundary.
              steele cutoff (double precision)
              • The distance beyond which the wall-atom interactions are not computed and assumed to be zero.
              steele shift (logical)
              • T: if you want to shift the wall potential to be zero at the cutoff.
              • F: if you do not want to shift the potential.
              steele delta (double precision)
              • Δ parameter for the spacing between the layers in the solid. Units are in Angstroms.
              steele rho_s (double precision)
              • ρs parameter for the density of the atom in the solid. Units are in atoms per cubic Angstrom.
              steele ntype (integer)
              • The number of atom types in the system that interact with the wall. Any atom type not listed here will have zero interaction with the wall. For each type you must list the following variables.
                  steele name (character*6)
                  • The name of the atom. This must match up with the atom names listed in the input_style=2 portion of each molecule that is interacting with this wall. If you are not using that input_style this will still work except you will need to know the atom names in the appropriate towhee_ff_whatever files.
                  sigma_sf (double precision)
                  • σ parameter for the interaction between this atom and the wall atoms. Units are Angstroms.
                  epsilon_sf (double precision)
                  • ε parameter for the interaction between this atom and the wall atoms. Units are K/kB.
    • 'external' : calls an external code (compiled as a library) that computes the total energy of the system. When using this option you need to include the following variable.
        external_code (character*20)
        • 'KIM' : uses the Knowledgebase of Interatomic Models to compute the energy. Use of this option requires special compilation to include the KIM library. Please see the Code Manual for detailed instructions on how to compile Towhee to include this KIM libraries. When using this option you must also include the following variables.
        • 'lcao' : uses the lcao code of Peter Schultz, also known as Quest or SeqQuest, to compute the quantum mechanical energy. Use of this option requires the placement of several libraries into the /towheebase/lib directory (use of symbolic links is also acceptable in that directory). The code must be compiled with the '--enable-lcao' option in order to link in these libraries. When using this option you must also include the following variables.
    solvation_style (character*20) optional parameter
    • 'none' : no additional solvation potential for this system (default value).
    • 'internal' : use one of the solvation potentials internal to Towhee. This option requires the following variable.
    • 'external' : use a solvation potential via an external program that has been compiled as a library and linked into Towhee.
        solvation_type (character*20)
        • 'tramonto' : solvation using the classical density functional theory code Tramonto to compute a solvation free energy.
    linit (logical)
    • .TRUE. if you are starting the simulation and wish to generate the positions of all of the atoms, assign initial box lengths and maximum displacements.
    • .FALSE. if you want to continue the simulation by reading in box lengths, maximum displacements, and coordinates from towhee_initial.
    initboxtype (character*20)
    • 'unit cell' : generates an initial structure by duplicating a unit cell. Reads information from the towhee_cell file and uses that to create an initial structure.
    • 'dimensions' : the dimensions of the initial boxes are entered in order to construct the initial boxes.
    • 'number density' : the total number density of molecules in the initial boxes are entered in order to compute the initial sizes of the boxes. This option generates cubic boxes.
    initstyle (character*20)
    This variable is only required for initboxtype settings of 'dimensions' or 'number density'
      One line for each simulation box in the system. Each line contains a value for each molecule type.
    • 'full cbmc' : A template for this molecule type is created using configurational-bias. This template is then replicated throughout the simulation box to generate an initial configuration.
    • 'template' : A template for this molecule type is read from towhee_template. This template is then replicated throughout the simulation box to generate an initial configuration.
    • 'coords' : The coordinates for each atom are read from towhee_coords. This is useful if you are starting from a different file format (such as pdb), or have another code for building an initial configuration.
    • 'nanotube' : The coordinates for each atom are read from towhee_nanotube. This file is generated by the Towhee code if you use the input_style for carbon nanotubes. This template is then replicated throughout the simulation box to generate an initial conformation.
    • 'helix cbmc' : The molecule is generated by placing some of the backbone atoms onto a helix and then growing the rest of the atoms using CBMC. Any molecule initialized using this style must have the following information listed subsequent to the initstyle variables.
        helix_moltyp (integer)
          An integer corresponding to the molecule type that had an initstyle variable set to 4 in one of the simulation boxes. These must be listed in consecutive order.
        helix_radius (double precision)
          The radius of the helix (units of Angstroms).
        helix_angle (double precision)
          The pitch angle the helix makes with respect to the z-axis (units of degrees).
        helix_keytype (character*10)
        • 'element' compares the helix_keyname with the character*2 variable element that contains the 2 letter elemental code for each atom.
        • 'nbname' compares the helix_keyname with the character*10 variable nbname that contains the 10 character code for each atom type. This is the same variable that is used when inputting the atom names with the Atom-based connectivity map (input_style=2).
        • 'pdbname' compares the helix_keyname with the character*4 variable pdbname that contains the 4 character code used in the pdb format output. This is most suitable for use with the Polypeptide builder (input_style=1) or the Nucleic acid builder (input_style=4).
        helix_keyname (character*10)
          The key for finding matches of the atom with the data structures for the molecule that is being grown as a helix. You need to choose an atom name that only appears in the backbone (e.g. 'P' for Charmm27 nucleic acids when using the element keytype, or ' CA ' for the C alpha backbone carbon of a polypeptide when using the pdbname keytype).
        helix_conlen (double precision)
          The distance between consecutive helix_element atoms (units of Angstroms).
        helix_phase (double precision)
          The initial angle of the helical chain (units of degrees). Normally, this has little effect as it is just a rotation about the z-axis, but if you are trying to set up two complementary nucleic acid chains to form a double helix then you would want their phase angles to differ by 180 degrees.
    • 'partial cbmc' : The molecule is constructed in two steps. First, a partial list of atom positions, amino acid 3-letter codes, and 4-letter atom identifiers (following the pdb standard) is read from towhee_partial and matched up against the expected amino acid and atom codes from the molecule template listed in towhee_input. Then configurational-bias is used to grow all of the missing atoms to create the initial structure.
    initlattice (character*20)
    This variable is only required for initboxtype settings of 'dimensions' or 'number density'.
      One line for each simulation box in the system. Each line contains a value for each molecule type.
    • 'center' : puts the center of mass of the molecule in the very center of the box. This is a very poor idea if you have multiples of the same molecule type in a box as they will all be right on top of each other. However, it is useful if you are trying to get multiple different molecule types in an initial configuration where one molecule is inside of another (multi-walled nanotubes is a good example).
    • 'none' : place the molecules exactly where their template indicates. This option makes the most sense for initstyle options like 'coords' where you want to just read positions from a file.
    • 'simple cubic' : places the first atom of the molecule onto a simple cubic lattice and the rest of the atoms in the molecule are placed relative to that atom. If you are not using an initstyle of 'coords' then this is probably the option you want to use to generate an initial structure.
    initmol (integer)
    This variable is only required for initboxtype settings of 'dimensions' or 'number density'
    • The initial number of each type of molecule in each box (one line per box).
    inix, iniy, iniz (integer)
    • The initial number of molecules (for initboxtype settings of 'dimensions' or 'number density') or of duplicated unit cells (for initboxtype setting of 'unit cell') in each direction in each box. The product of inix*iniy*iniz must be greater than or equal to the initial number of molecules in that box (for initboxtype settings of 'dimensions' or 'number density'). While these are labeled x, y, and z they actually correspond to the three coordinate vectors (truly x, y, and z for a rectangular box).
    hmatrix (double precision)
    This variable is only required for initboxtype settings of 'dimensions'
    • The initial box dimensions (Angstroms) for the three vectors that describe the simulation box. There are nine entries (3 for each of the 3 vectors) in total for each simulation box. These are listed one vector at a time, with the three numbers which make up each vector listed on the same line. Note that the coordinate system you choose does not have to be orthogonal, but it must follow the right hand rule. The three vectors must also all be at least 45 degrees apart. Note that if you wish to use a rectangular box then only the diagonal elements of hmatrix will be non-zero, and these will be equal to the box lengths in the x, y, and z dimensions.
    box_number_density (double precision)
    This variable is only required for initboxtype settings of 'number density'
    • The initial total number density of molecules in each simulation box. Listed as a single value per line with one line for each box as specified by numboxes. Units are molecules per nm3.

  • Note: the pm* variables are used to determine which move type to perform every time we want to do a Monte Carlo move. A move is selected by choosing a pseudorandom number between 0.0 and 1.0 and then going down the list of pm* until you find one which has a value higher than the pseudorandom number. At least one of the variables must be set to 1.0. A similar procedure is performed when we want to determine which boxes or molecule types to perform the selected move upon. These are done using the pm**pr and pm**mt arrays.

    Starting with Towhee release 4.15.2, all leading pm* probabilities are optional; if such a probability is not specified, then the rest of the variables in that section must likewise be left out, and the probability of that move is zero. As an example, if the variable pmback is not given, then pmbkmt must not be specified, and the Configurational-Bias Protein Backbone Regrowth move will never be performed.

  • Isotropic Volume Move: These variables are only included for the following cases
    • ensemble is 'npt'
    • ensemble is 'nvt' and numboxes is 2 or more.
    pmvol (double precision) optional parameter
    • Probability of performing a volume move. If (ensemble is 'npt') then a single box is selected and it exchanges volume with an external pressure bath (see pressure). If (ensemble = 'nvt' and numboxes > 1) a pair of boxes are selected and volume is exchanged between them.
      pmvlpr (double precision)
      • Probability of performing a volume move on a particular box, or box pair. All of these variables are listed on a single line. If (ensemble = 'npt') then a value of pmvlpr is listed for each box. If (ensemble = 'nvt') then a value is listed for each pair of simulation boxes where the pairs are ordered (1,2), (1,3), ... (1,numboxes), (2,3), ... (numboxes-1,numboxes).
      rmvol (double precision) [a single value regardless of the actual number of box pairs]
      • The initial volume maximum displacement. If this is an isobaric-isothermal ensemble (ensemble = 'npt') then this is the initial maximum volume displacement (cubic Angstroms) in each box. If this is the canonical Gibbs ensemble (ensemble = 'nvt' and numboxes > 1 ) then this is the maximum displacement (logarithmic space) for each pair of boxes. As the simulation progresses, these values will be updated for each box, or each pair of boxes (see iratv).
      tavol (double precision)
      • The target acceptance rate for the volume move. Must be a value between 0.0 and 1.0. The volume displacement (rmvol) is periodically adjusted (see iratv) to yield this acceptance rate. I typically use a value of 0.5, though some researchers prefer smaller values.

    Anisotropic Volume Move: These variables are only included for the following cases
    • ensemble is 'npt'
    • ensemble is 'nvt' and numboxes is 2 or more.
    pmcell (double precision) optional parameter
    • Probability of performing a unit cell adjustment move. If (ensemble = 'npt' ) then a single box is selected and a single hmatrix element is changed. This results in a volume exchange with a fictional external pressure bath (see pressure). If (ensemble = 'nvt' and numboxes > 1) a pair of boxes are selected. One of the boxes is then selected according to the pmcellpt variable and a single hmatrix element is changed in that box. This results in a change of volume for the first box which is countered by isotropically changing the volume in the second box.
      pmcellpr (double precision)
      • Probability of performing a unit cell adjustment move on a particular box, or box pair. All of these variables are listed on a single line If (ensemble = 'npt') then a value of pmcellpr is listed for each box. If (ensemble = 'nvt') then a value is listed for each pair of simulation boxes where the pairs are ordered (1,2), (1,3), ... (1,numboxes), (2,3), ... (numboxes-1,numboxes).
      pmcellpt (double precision)
      • Probability of selecting the first box of the pair as the box to perform the non-isotropic volume move upon, while its partner undergoes an isotropic volume move. This variable is only meaningful if (ensemble = 'nvt'). Note that you can choose to perform the non-isotropic volume move always on the same box and this might be useful if you are doing a solid-vapor equilibria calculation.
      rmcell (double precision)
      • The initial unit cell adjustment maximum displacement. In all cases, this is the maximum amount (in Angstroms) that a single element of the hmatrix can change in a single unit cell move. Note, the in the canonical Gibbs ensemble case it is possible for the isotropic box to undergo an hmatrix change that is larger than this value as that box simply makes up for the volume change caused by the non-isotropic adjustment in the first box. As the simulation progresses, these values are updated for each box with a frequency controlled by iratv.
      tacell (double precision)
      • The target acceptance rate for the unit cell adjustment move. Must be a value between 0.0 and 1.0. The unit cell displacement (rmcell) is periodically adjusted (see iratv) to yield this acceptance rate. I typically use a value of 0.5.

    Rotational-bias 2 box molecule Transfer Move: These variables are only included if numboxes is greater than or equal to 2
    pm2boxrbswap (double precision) optional parameter
    • Probability of performing a rotational-bias interbox molecule transfer move. This move takes a molecule out of one box and tries to place it in another box. The molecule is grown using nch_nb_one attempted different orientations and position (of the center-of-mass) for the new molecule.
      pm2rbswmt (double precision)
      • Probability of performing a rotational-bias interbox molecule transfer move on each type of molecule in the system.
      pm2rbswpr (double precision)
      • Probability of performing a rotational-bias interbox molecule transfer move between each pair of boxes in the system. The box pairs are ordered (1,2), (1,3), ... (1,numboxes), (2,3), ... (numboxes-1,numboxes).

    Configurational-bias 2 box molecule Transfer Move: These variables are only included if numboxes is greater than or equal to 2
    pm2boxcbswap (double precision) optional parameter
      pm2cbswmt (double precision)
      • Probability of performing a configurational-bias interbox molecule transfer move on each type of molecule in the system.
      pm2cbswpr (double precision)
      • Probability of performing a configurational-bias interbox molecule transfer move between each pair of boxes in the system. The box pairs are ordered (1,2), (1,3), ... (1,numboxes), (2,3), ... (numboxes-1,numboxes).

    Configurational-bias grand-canonical insertion/deletion Move: These variables are only included if ensemble is 'uvt'
    pmuvtcbswap (double precision) optional parameter
    • Probability of performing a grand-canonical configurational-bias insertion or deletion move.
      pmuvtcbmt (double precision)
      • Probability of performing a grand-canonical configurational-bias insertion or deletion move on each type of molecule in the system.

    Configurational-bias single box molecule Reinsertion Move
    pm1boxcbswap (double precision) optional parameter
    • Probability of performing an intrabox configurational-bias molecule transfer move. This move takes a molecule out of one box and tries to place it back into the same box. The molecule is grown using coupled-decoupled configurational-bias Monte Carlo.
      pm1cbswmt (double precision)
      • Probability of performing an intrabox configurational-bias molecule transfer move on each type of molecule in the system.

    Intrabox two molecule switch based upon the center of mass positions
    pm1boxcomswitch (double precision) optional parameter
    • Probability of performing an intrabox exchange of the center of mass of two molecules of different types.
      pm1comswbox (double precision)
      • Probability of performing a center of mass switch move in each simulation box. List one value for each simulation box. At least one of the boxes must have a value of 1.0d0.
      pm1comswpair (double precision)
      • Probability of performing a center of mass switch move on each pair of molecule types in the simulation. List one value for each pair of molecule types in the simulation (nmolty*(nmolty-1)/2). At least one of the pairs must have a value of 1.0d0.

    Aggregation Volume Bias Move Type 1
    pmavb1 (double precision) optional parameter
    • Probability of performing an aggregation volume bias move of type 1, as described in Chen and Siepmann 2000. This is useful for forming and destroying clusters in simulations with molecules that tend to aggregate together.
      pmavb1in (double precision)
      • Probability of trying to move a molecule into an inner region for aggregation volume bias move of type 1. Must be strictly between 0.0 and 1.0.
      pmavb1mt (double precision)
      • Probability of performing an aggregation volume bias move of type 1 where a molecule of a certain type is moved. This is an array with one element for each molecule type.
      pmavb1ct (double precision)
      • Probability of performing an aggregation volume bias move of type 1 where the molecule target is of a certain type. The molecule that is moved is chosen according to pmavb1mt and then the type of molecule that is used as a reference for determining the inner and outer regions is found using this variable. This is a two dimensional array and uses one line of text for each type of molecule in the system.
      avb1rad (double precision)
      • The radius used to define the inner and outer volumes in the aggregation volume bias move of type 1. The distance is specified in Angstroms and must be greater than zero, but less than or equal to rcut.

    Aggregation Volume Bias Move Type 2
    pmavb2 (double precision) optional parameter
    • Probability of performing an aggregation volume bias move of type 2, as described in Chen and Siepmann 2001. This is useful for forming and destroying clusters in simulations with molecules that tend to aggregate together.
      pmavb2in (double precision)
      • Probability of trying to move a molecule into an inner region for aggregation volume bias move of type 2. Must be strictly between 0.0 and 1.0.
      pmavb2mt (double precision)
      • Probability of performing an aggregation volume bias move of type 2 where a molecule of a certain type is moved. This is an array with one element for each molecule type.
      pmavb2ct (double precision)
      • Probability of performing an aggregation volume bias move of type 2 where the molecule target is of a certain type. The molecule that is moved is chosen according to pmavb2mt and then the type of molecule that is used as a reference for determining the inner and outer regions is found using this variable. This is a two dimensional array and uses one line of text for each type of molecule in the system.
      avb2rad (double precision)
      • The radius used to define the inner and outer volumes in the aggregation volume bias move of type 2. The distance is specified in Angstroms and must be greater than zero, but less than or equal to rcut.

    Aggregation Volume Bias Move Type 3
    pmavb3 (double precision) optional parameter
    • Probability of performing an aggregation volume bias move of type 3, as described in Chen and Siepmann 2001. This is useful for transferring molecules between clusters.
      pmavb3mt (double precision)
      • Probability of performing an aggregation volume bias move of type 3 where a molecule of a certain type is moved. This is an array with one element for each molecule type.
      pmavb3ct (double precision)
      • Probability of performing an aggregation volume bias move of type 3 where the molecule target is of a certain type. The molecule that is moved is chosen according to pmavb3mt and then the type of molecule that is used as a reference for determining the inner and outer regions is found using this variable. This is a two dimensional array and uses one line of text for each type of molecule in the system.
      avb3rad (double precision)
      • The radius used to define the inner and outer volumes in the aggregation volume bias move of type 3. The distance is specified in Angstroms and must be greater than zero, but less than or equal to rcut.

    Configurational-Bias Partial Molecule Regrowth
    pmcb (double precision) optional parameter
    • Probability of performing a molecule regrowth move on a molecule without regard to which box the molecule is currently located in. This move chooses a molecule of the appropriate type at random, selects an atom of the molecule at random, and then regrows the molecule either entirely (if a random number < pmall) or in all directions except for one. The molecule is regrown using configurational-bias.
      pmcbmt (double precision)
      • Probability of performing a molecule regrowth on each type of molecule in the system.
      pmall (double precision)
      • pmall is the probability that a molecule regrowth move will regrow the entire molecule. This is listed for each molecule type in the simulation.

    Configurational-Bias Protein Backbone Regrowth
    pmback (double precision) optional parameter
    • Probability of performing configurational-bias fixed-endpoint regrowth of a portion of the protein backbone. This selects an atom along the peptide backbone, chooses another backbone atom that is connected by three bonds to the first atom, and then regrows all of the atoms in between these two atoms.

    Configurational-Bias Peptide side-chain Regrowth
    pmcbside (double precision) optional parameter
    • Probability of performing a configurational-bias regrowth move on a single side-chain of peptide. This works by compiling a list of atoms that have a pdbname of ' CA ' and selecting one of those at the origin of a side-chain regrowth. This is functional for the normal amino acids, plus proline and disulfide bonded cysteines.

    Torsional Pivot Move
    pmpivot (double precision) optional parameter
    • Probability of performing a pivot move about a random bond in the molecule. This move chooses a bond that is not entirely contained in a single ring structure, and has at least one bond emanating from each end, and then rotates one side of the molecule about that bond.

    Concerted Rotation Move on a non-peptide backbone
    pmconrot (double precision) optional parameter
    • Probability of performing a concerted rotation move for a sequence of 9 atoms in a molecule.

    Concerted Rotation Move over a 3 peptides backbone sequence
    pmcrback (double precision) optional parameter
    • Probability of performing a concerted rotation move on a sequence of three peptides in a polypeptide. This move only works for polypeptides.
      pmcrbmt (double precision)
      • Probability of performing a protein backbone concerted rotation move on each type of molecule in the system.

    Plane Shift Move
    pmplane (double precision) optional parameter
    • Probability of performing a plane shift move. This move displaces all of the molecules whose center of mass lies in a plane of width planewidth. A new trial position for the center of the plane of atoms is generated uniformly across the available plane.
      pmplanebox (double precision)
      • Probability of performing a plane shift in each of the simulation boxes. List one value for each simulation box. At least one of the boxes must have a value of 1.0d0.
      planewidth (double precision)
      • The width of the plane for the plane shift move. Any molecule whose center of mass is within a plane of this thickness (whose position is chosen uniformly along one axis) will move during the plane shift move. The value of planewidth must be greater than 0.0d0 and less than the shortest box length.

    Row Shift Move
    pmrow (double precision) optional parameter
    • Probability of performing a row shift move. This move displaces all of the molecules whose center of mass lies in a row of diameter rowwidth. A new trial position for the center of the row of atoms is generated uniformly across the available row.
      pmrowbox (double precision)
      • Probability of performing a row shift in each of the simulation boxes. List one value for each simulation box. At least one of the boxes must have a value of 1.0d0.
      rowwidth (double precision)
      • The width of the plan for the row shift move. Any molecule whose center of mass is within a row of this thickness (whose position is chosen uniformly along one axis) will move during the row shift move. The value of rowwidth must be greater than 0.0d0 and less than the shortest box length.

    Intramolecular Single Atom Translation Move
    pmtraat (double precision) optional parameter
    • Probability of performing a single-atom translation move on a molecule without regard to which box the molecule is currently located in. This move chooses a molecule of the appropriate type at random, selects an atom of the molecule at random, selects a vector on a unit sphere at random, and then attempts to displace the atom a random distance between -rmtraa and +rmtraa in that direction.
      pmtamt (double precision)
      • Probability of performing a single-atom translation move on each type of molecule in the system.
      rmtraa (double precision)
      • The initial Atom-translation maximum displacement (Angstroms) for all molecules types in all boxes. As the simulation progresses, these values are updated to yield the desired acceptance rate for each molecule type in each box (see trmaxdispfreq).
      tatraa (double precision)
      • The target acceptance rate for the atom translation move. Must be a value between 0.0 and 1.0. The maximum atom translational displacement (rmtraa) is periodically adjusted (see trmaxdispfreq) to yield this acceptance rate. I typically use a value of 0.5, though some researchers prefer smaller values.

    Composite Move
    pmcomposite (double precision) optional parameter
    • Probability of performing a composite move that consists of a random center-of-mass translation and random rotation. This move is essentially a concatenation of the Center-of-Mass Molecule Translation Move and the Rotation about the Center-of-Mass Move.
      pmcomt (double precision)
      • Probability of performing a composite move on each type of molecule in the system.
      rmcomtra (double precision)
      • The molecular translation displacement (angstroms) for all molecule types in all boxes. This parameter is essentially identical to the rmtrac parameter for translations of molecular COM.
      rmcomrot (double precision)
      • The molecular rotation maximum displacement (radians) for all molecule types in all boxes. This parameter is essentially identical to the rmrot parameter for rotations about COM.

    Center-of-Mass Molecule Translation Move
    pmtracm (double precision) optional parameter
    • Probability of performing a center-of-mass translation move on a molecule without regard to which box the molecule is currently located in. This move chooses a molecule of the appropriate type at random, chooses a vector on a unit sphere at random, and then attempts to displace the entire molecule a random distance between -rmtrac and +rmtrac in that direction.
      pmtcmt (double precision)
      • Probability of performing a center-of-mass translation move on each type of molecule in the system.
      rmtrac (double precision)
      • The initial Center-of-mass translation maximum displacement (Angstroms) for all molecule types in all boxes. As the simulation progresses, these values are updated to yield the desired acceptance rate for each molecule type in each box (see trmaxdispfreq).
      tatrac (double precision)
      • The target acceptance rate for the center-of-mass translation move. Must be a value between 0.0 and 1.0. The maximum center-of-mass translational displacement (rmtrac) is periodically adjusted (see trmaxdispfreq) to yield this acceptance rate. I typically use a value of 0.5, though some researchers prefer smaller values.

    Rotation about the Center-of-Mass Move
    pmrotate (double precision) optional parameter
    • Probability of performing a rotation about the center-of-mass move for a molecule without regard to the box the molecule is currently located in. This move chooses a molecule of the appropriate type at random and then attempts to rotate the entire molecule about the x, y, and z axes that run through the center-of-mass a random number of radians between -rmrot and +rmrot around each of the three axes.
      pmromt (double precision)
      • Probability of performing a rotation move on each type of molecule in the system.
      rmrot (double precision)
      • The initial molecular rotation maximum displacement (radians) for all molecule types in all boxes. As the simulation progresses, these values are updated to yield the desired acceptance rate for each molecule type in each box (see trmaxdispfreq).
      tarot (double precision)
      • The target acceptance rate for the rotation move. Must be a value between 0.0 and 1.0. The rotation displacement (rmrot) is periodically adjusted (see trmaxdispfreq) to yield this acceptance rate. I typically use a value of 0.5, though some researchers prefer smaller values.

    cbmc_analysis (character*30) optional parameter
    • 'normal' : (Current default setting) A relatively small amount of output is output to towhee_cbmc.
    • 'full' : A substantial amount of analysis about the predicted efficiency of configurational-bias algorithms is output to towhee_cbmc.
    cbmc_formulation (character*60) optional parameter
    • 'Martin and Frischknecht 2006' : (Current default setting) Monte Carlo moves that involve the configurational-bias algorithm use the Coupled to pre-nonbond formulation of the coupled-decoupled algorithm that is described in Martin and Frischknecht 2006. This algorithm has a bond length selection, two bending angle selections, and a dihedral selection that are all decoupled from each other; while in turn being coupled to a pre-nonbond loop that is coupled to the nonbonded selection.
    • 'Martin and Siepmann 1999 + Martin and Thompson 2004' : Monte Carlo moves that involve the configurational-bias algorithm use the coupled-decoupled formulation presented in the appendix of Martin and Siepmann 1999 with the addition of a decoupled bond selection first described in Martin and Thompson 2004. This algorithm has decoupled bond length selection, two decoupled angle selections, and a dihedral selection that is coupled to the nonbonded selection.
    cbmc_setting_style (character*30) optional parameter
    • 'Martin and Frischknecht' : (Current default setting) sets up the configurational-bias values according to the strategies described in Martin and Frischknecht 2006. This option sets the following variables for all molecule types in the system. For more information about these variables see the explicit setting for this variable.
        cbmc_fit_strategy = 'equilibrium'
        cbmc_bond_generation = 'autofit gaussian'
        bond_sdev_multiplier = 1.0d0
        cbmc_bend_generation = 'ideal + autofit gaussian'
        bend_a_sdev_multiplier = 1.0d0
        bend_b_sdev_multiplier = 1.0d0
        bend_a_ideal_fraction = 0.01d0
        bend_b_ideal_fraction = 0.01d0
        cbmc_dihedral_generation = 'ideal + autofit gaussian'
        dihedral_peak_weight_style = 'uniform'
        dihedral_sdev_multiplier = 1.0d0
        dihedral_ideal_fraction = 0.01d0
        cbmc_nb_one_generation = 'uniform'
        two_bond_fixed_endpoint_bias_style = 'analytic Boltzmann using angles'
        three_bond_fixed_endpoint_bias_style = 'analytic using max and min 2-4 distance'
        nch_nb_one = 10
        nch_nb = 10
        if cbmc_formulation = 'Martin and Siepmann 1999 + Martin and Thompson 2004' then
          nch_pre_nb = 1
          nch_tor = 100
          nch_tor_connect = 100
          nch_bend_b = 100
        else if cbmc_formulation = 'Martin and Frischknecht 2006' then
          nch_pre_nb = 100
          nch_tor = 1
          nch_tor_connect = 1
          nch_bend_b = 1
        nch_bend_a = 1
        nch_vib = 1
    • 'default ideal' : sets up all of the configurational-bias variables according to a general set of default values using the 'ideal' generation styles. This option sets the following variables for all molecule types in the system. For more information about these variables see the explicit setting for this variable.
        cbmc_fit_strategy = 'equilibrium'
        cbmc_bond_generation = 'r^2 with bounds'
        cbmc_bend_generation = 'ideal'
        cbmc_dihedral_generation = 'ideal'
        cbmc_nb_one_generation = 'uniform'
        two_bond_fixed_endpoint_bias_style = 'analytic Boltzmann using angles and dihedrals'
        three_bond_fixed_endpoint_bias_style = 'analytic using max and min 2-4 distance'
        nch_nb_one = 10
        nch_nb = 10
        nch_pre_nb = 1
        nch_tor = 360
        nch_tor_connect = 360
        nch_bend_a = 1000
        nch_bend_b = 1000
        nch_vib = 1000
        vibrang = 0.85 1.15
    • 'widom ideal' : sets up all of the configurational-bias variables according to a general set of default values using the 'ideal' generation styles a single growth trial for all of the nonbonded selections following the (now discredited) belief that this was the only way to insure the chemical potential was computed correctly using the normal Widom insertion method. This option sets the following variables for all molecule types in the system. For more information about these variables see the explicit setting for this variable.
        cbmc_fit_strategy = 'equilibrium'
        cbmc_bond_generation = 'r^2 with bounds'
        cbmc_bend_generation = 'ideal'
        cbmc_dihedral_generation = 'ideal'
        cbmc_nb_one_generation = 'uniform'
        two_bond_fixed_endpoint_bias_style = 'none'
        three_bond_fixed_endpoint_bias_style = 'none'
        nch_nb_one = 1
        nch_nb = 1
        nch_pre_nb = 1
        nch_tor = 360
        nch_tor_connect = 360
        nch_bend_a = 1000
        nch_bend_b = 1000
        nch_vib = 1000
        vibrang = 0.85 1.15
    • 'Martin and Thompson FPE 2004' : sets up all of the configurational-bias variables to the values used in Martin and Thompson 2004. This option sets the following variables for all molecule types in the system. For more information about these variables see the explicit setting for this variable.
        cbmc_fit_strategy = 'equilibrium'
        cbmc_bond_generation = 'r^2 with bounds'
        cbmc_bend_generation = 'ideal'
        cbmc_dihedral_generation = 'ideal'
        cbmc_nb_one_generation = 'uniform'
        two_bond_fixed_endpoint_bias_style = 'analytic Boltzmann using angles and dihedrals'
        three_bond_fixed_endpoint_bias_style = 'analytic using max and min 2-4 distance'
        nch_nb_one = 10
        nch_nb = 10
        nch_pre_nb = 1
        nch_tor = 360
        nch_tor_connect = 360
        nch_bend_a = 100
        nch_bend_b = 100
        nch_vib = 1000
        vibrang = 0.85 1.15
    • 'default autofit gaussian' : sets up all of the configurational-bias variables to autofit gaussian settings. This option sets the following variables for all molecule types in the system. For more information about these variables see the explicit setting for this variable.
        cbmc_fit_strategy = 'equilibrium'
        cbmc_bond_generation = 'autofit gaussian'
        bond_sdev_multiplier = 1.0d0
        cbmc_bend_generation = 'autofit gaussian'
        bend_a_sdev_multiplier = 1.0d0
        bend_b_sdev_multiplier = 1.0d0
        cbmc_dihedral_generation = 'autofit gaussian'
        dihedral_sdev_multiplier = 1.0d0
        cbmc_nb_one_generation = 'uniform'
        two_bond_fixed_endpoint_bias_style = 'autofit gaussian'
        two_bond_bias_sdev_multiplier = 1.0d0
        two_bond_bias_vibrange = 0.5d0 1.5d0
        three_bond_fixed_endpoint_bias_style = 'autofit gaussian using max and min 2-4 distance'
        three_bond_bias_sdev_multipler = 1.0d0
        nch_nb_one = 10
        nch_nb = 10
        if cbmc_formulation = 'Martin and Siepmann 1999 + Martin and Thompson 2004' then
          nch_pre_nb = 1
          nch_tor = 10
          nch_tor_connect = 10
        else if cbmc_formulation = 'Martin and Frischknecht 2006' then
          nch_pre_nb = 10
          nch_tor = 1
          nch_tor_connect = 1
        nch_bend_a = 1
        nch_bend_b = 1
        nch_vib = 1
    • 'explicit' : All of the configurational-bias options are explicitly required. When using this option you also need to include the following variables.
        cbmc_fit_strategy (character*30)
        • 'equilibrium' : (Current default setting) The equilibrium values are used when fitting the distributions of higher order terms for configurational-bias. For example, equilibrium bond lengths are used when fitting the bending angle distributions.
        cbmc_nb_one_generation (character*30 array) [one line for each simulation box and each line contains one value for each molecule type]
        • 'uniform' : the trial coordinates of first atom inserted into this simulation box are generated uniformly
        • 'energy bias' : the trial coordinates of the first atom inserted into this simulation box are generated by selecting a subvolume according to the Boltzmann weighted energy of a test atom in the center of each subvolume, averaged over all possible insertion atoms in the mapped molecule. Once a subvolume is selected then a position in that subvolume is generated uniformly. This biasing scheme was first used by Snurr et al. 1993. Note that the energy map used in practice is a linear combination of the Boltzmann weighted map described above (0.99 of the fraction) with a random selection of any subcube in the box (0.01 of the fraction) to make sure that the generation probability is strictly positive in all subcubes. The insertion energies are computed using the initial positions of all molecules in simulation box 1. If you are using this to help with insertions in a porous solid then you may want to only compute this for the empty porous solid and use the restart of that map in future simulations. When using this option you need the following variables.
            mapmolty (integer)
            • The molecule type to use as the insertion probe for determining the energy biasing weights.
            lcreatemap (logical)
            • .true. if you wish to create a new towhee_map file. This file contains information about the energy profile of the porous molecule. When using this setting you must also include the following additional variables
                cubex,cubey,cubez (integer)
                  The number of cubelets you want to have in each direction of the box. Each dimension of the box is divided by the corresponding cube* value and for each cube an energy value is computed. This value is then used for biasing insertion/deletion moves. cubex*cubey*cubez mustbe less than or equal to MAXCUBE
            • .false. if you want to use a previously generated map in the towhee_map file. That file must have been generated using version 5.2.4 or later as there was a different, and problematic, format prior to that version.
        nch_nb_one (integer) [one value for each molecule type]
        • The number of trial positions that are sampled for the first atom inserted during a configurational-bias or rotational-bias molecule exchange move (see pm2boxrbswap, pm2boxcbswap, and pm1boxcbswap).
        nch_nb (integer) [one value for each molecule type]
        • The number of trial positions that are sampled for all atoms except for the first atom inserted during a configurational-bias molecule exchange move (see pm2boxcbswap and pm1boxcbswap). This is used for all atoms in a configurational-bias regrowth move.
        nch_pre_nb (integer) [one value for each molecule type]
        • This variable is only required when the cbmc_formulation is 'Martin and Frischknecht 2006'. The number of trials for a selection procedure that takes place after the dihedral selection, but before the nonbond selection.
        cbmc_dihedral_generation (character*30)
        • 'ideal' : dihedral trials are generated according to the ideal distribution. For dihedrals that means trials are generation uniformly on (-Pi, Pi).
        • 'global gaussian' : dihedral trials are generated according a series of gaussian distributions that are a function of the bondpatt on the central two atoms. This series is hard coded into Towhee and details can be found by looking in the getcbdihed.F subroutine. This gaussian bias is then removed in the acceptance rule. When using this option you must also include the following variable.
            sdevtor (double precision)
              The standard deviation (with units of degrees) that is used for each of the gaussian distributions for the dihedral angles. For best results this number should be set to the observed distribution computed in the analyse_movie utility. This number must be positive and a default value of 20.0 is suggested.
        • 'autofit gaussian' : dihedral trials are generated according to a series of gaussian distributions that are individually fit to the Boltzmann factor as a function of dihedral angle for each individual dihedral in the system. This fit is performed automatically at the start of each simulation using the equilibrium bond lengths and bending angles. This bias is then removed in the acceptance rule. When using this option you must also include the following variable.
            dihedral_sdev_multiplier (double precision)
              The factor that is multiplied by the observed gaussian standard deviation for each peak in the automatic dihedral fit in order to create the standard deviations that are used to generate the dihedrals during the simulation. This number must be positive and a default value of 1.0 is suggested.
        • 'ideal + autofit gaussian' : dihedral trials are generated according to a linear combination of the ideal distribution (uniform) and a series of gaussian distributions that are individually fit to the Boltzmann factor as a function of dihedral angle for each individual dihedral in the system. This fit is performed automatically at the start of each simulation using the equilibrium bond lengths and bending angles. This bias is then removed in the acceptance rule. When using this option you must also include the following variables.
            dihedral_peak_weight_style (character*30)
            • 'uniform' : each peak is selected with the same (uniform) probability.
            • 'isolated Boltzmann' : the peaks are selected with a probability proportional to the sum of the Boltzmann weight computed during the automatic fitting process.
            dihedral_sdev_multiplier (double precision)
              The factor that is multiplied by the observed gaussian standard deviation for each peak in the automatic dihedral fit in order to create the standard deviations that are used to generate the dihedrals during the simulation. This number must be positive and a default value of 1.0 is suggested.
            dihedral_ideal_fraction (double precision)
              The fraction of dihedral trials that are generated using the ideal distribution. The remainder of the trials are generated using the autofit gaussians. This number must be in the range [0.0,1.0] inclusive.
        nch_tor (integer) [one value for each molecule type]
        • The number of trial dihedral angles that are sampled during configurational-bias moves. The value must be positive.
        nch_tor_connect (integer) [one value for each molecule type]
        • The number of trial dihedral angles that are sampled during configurational-bias moves when we have grown the molecule such that we need to connect back up with atoms that already exist. This is needed in order to regrow cyclic molecules, and to regrow the interiors of large molecules. The value must be positive.
        cbmc_bend_generation (character*30)
        • 'ideal' : bending trials are generated according to the ideal distribution. This is the Sine distribution for bending angle type A and uniform on (-Pi, Pi) for bending angle type B.
        • 'global gaussian' : bending trials are generated according to gaussian distributions. Bending A trials are generated according to a single gaussian with a mean set to the equilibrium bending angle and a standard deviation set to sdevbena. Bending B trials are generated according to one or more gaussian distributions with means set based upon the bondpatt of the central atom and a standard deviation set to setbenb. For more details on the bending B distribution see the getcbangle.F subroutine. When using this option you must also include the following variables.
            sdevbena (double precision)
              The standard deviation to use when generating the part A bending trials (units of degrees). Must be positive, and it is best to set to the observed distribution of the angles as measured by the analyse_movie utility.
            sdevbenb (double precision)
              The standard deviation to use when generating the part B bending trials (units of degrees). Must be positive, and it is best to set to the observed distribution of the angles as measured by the analyse_movie utility.
        • 'autofit gaussian' : bending trials are generated according to gaussian distributions. Bending A trials are generated according to a single gaussian with a mean and standard deviation fit to Sin(theta)*exp(-beta Ubend). The standard deviation used to generate trials is a product of the observed distribution standard deviation times bend_a_sdev_multiplier. Bending B trials are generated according to one or more gaussian distributions with means fitted to exp(-beta ubend) for rotating the angles about a cone (with everything else set to the equilibrium bond lengths and bending angles). The observed standard deviation from this fit is multiplied by bend_b_sdev_multiplier to get the standard deviation used to generate bending B angles during the simulation. This bias is removed in the acceptance rules. When using this option you must also include the following variables.
            bend_a_sdev_multiplier (double precision)
              This value is multiplied by the observed standard deviation from performing the fit in order to create the standard deviation that is used to generate bending A trials during the simulation. This value must be positive and the currently suggested value is 1.0.
            bend_b_sdev_multiplier (double precision)
              This value is multiplied by the observed standard deviation from performing the fit in order to create the standard deviation that is used to generate bending B trials during the simulation. This value must be positive and the currently suggested value is 1.0.
        • 'ideal + autofit gaussian' : bending trials are generated according to a linear combination of the ideal distributions (sine for bend A, uniform for bend B) and the autofit gaussian distributions. Gaussian bending A trials are generated according to a single gaussian with a mean and standard deviation fit to Sin(theta)*exp(-beta Ubend). The standard deviation used to generate trials is a product of the observed distribution standard deviation times bend_a_sdev_multiplier. Gaussian bending B trials are generated according to one or more gaussian distributions with means fitted to exp(-beta ubend) for rotating the angles about a cone (with everything else set to the equilibrium bond lengths and bending angles). The observed standard deviation from this fit is multiplied by bend_b_sdev_multiplier to get the standard deviation used to generate bending B angles during the simulation. This bias is removed in the acceptance rules. When using this option you must also include the following variables.
            bend_a_sdev_multiplier (double precision)
              This value is multiplied by the observed standard deviation from performing the fit in order to create the standard deviation that is used to generate bending A trials during the simulation. This value must be positive and the currently suggested value is 1.0.
            bend_b_sdev_multiplier (double precision)
              This value is multiplied by the observed standard deviation from performing the fit in order to create the standard deviation that is used to generate bending B trials during the simulation. This value must be positive and the currently suggested value is 1.0.
            bend_a_ideal_fraction (double precision)
              The fraction of bending A trials that are generated using the ideal distribution of Sin(theta). The remainder of the trials are generated using the autofit gaussians. This value must be in the range [0.0,1.0] inclusive.
            bend_b_ideal_fraction (double precision)
              The fraction of bending B trials that are generated using the ideal distribution of uniform on (-Pi,Pi). The remainder of the trials are generated using the autofit gaussians. This value must be in the range [0.0,1.0] inclusive.
        nch_bend_a (integer) [one value for each molecule type]
        • The number of trial angles that are sampled during configurational-bias moves when we are selecting the iugrow-iufrom-iuprev angle. This value must be positive. Currently suggested values are in the range of 100 to 1000 when using cbmc_bend_generation style 'ideal' and in the range from 1 to 10 when using cbmc_bend_generation styles 'global gaussian' or 'autofit gaussian'.
        nch_bend_b (integer) [one value for each molecule type]
        • The number of trial angles that are sampled during configurational-bias moves when we are selecting the rotation about a cone of one of the iugrow angles relative to the others. This value must be positive. Currently suggested values are in the range of 100 to 1000 when using cbmc_bend_generation style 'ideal' and in the range from 1 to 10 when using cbmc_bend_generation styles 'global gaussian' or 'autofit gaussian'.
        cbmc_bond_generation (character*30)
        • 'autofit gaussian' : Generate trial bond lengths according to a gaussian distribution with a mean and standard deviation fitted to r2 exp(-beta Ubond). When using this option you also need to include the following variable.
            bond_sdev_multiplier (double precision)
              This value is multiplied by the observed standard deviation of the r2 exp(-beta Ubond) distribution in order to determine the standard deviation used to generate bond trials. This value must be positive. The currently suggested value is 1.0.
        • 'global gaussian' : Generate trial bond lengths according to a gaussian distribution with a mean set to the equilibrium bond length and a standard deviation specified as sdevvib. When using this option you must also include the following variable.
            sdevvib (double precision)
              The standard deviation of a gaussian distribution that is used to sample bond lengths during a configurational-bias regrowth for a cbmc_bond_generation style of 'global gaussian'. Units are Angstroms. For best results perform a short simulation of single-atom translation moves, analyse that data using the analyse_movie utility, and set this value to the observed standard deviations in the bond length distribution.
        • 'ideal' : Generate trial bond lengths according to the ideal r2 probability distribution within the range (0, max_bond_length).
        • 'r^2 with bounds' : Generate trial bond lengths according to a bounded r2 probability distribution within the ranges set by the vibrang variable. This distribution is proportional to the true distribution, but has a limited sampling range while the true distribution is of infinite extent When using this option you also need to include the following variable.
        nch_vib (integer) [one value for each molecule type]
        • The number of trial bond lengths that are sampled during a configurational-bias move. This value must be positive. Currently suggested values are 1000 trials for cbmc_bond_generation style 'r^2 with bounds' and a value in the range of 1 to 10 for cbmc_bond_generation styles 'global gaussian' and 'autofit gaussian'.
        two_bond_fixed_endpoint_bias_style (character*50)
        • 'none' : No additional biasing is utilized during a configurational-bias move step that involves the growth of a new atom that is separated from an already existing atom by two bonds where the atom between those two bonds does not currently exist.
        • 'analytic Boltzmann using angles' : Additional biasing is utilized during a configurational-bias move step that involves the growth of a new atom that is separated from an already existing atom by two bonds where the atom between those two bonds does not currently exist. The biasing has the following form where i is the atom being grown, k is an atom that already exists, and j is an atom that is bonded to i and j, but has not yet been grown.
          pbias(rik) = Min[minbias,paBua].
          paBua
            = pangle for rik < rij0 + rjk0
            = pbond for rij0 + rjk0 < rik
          pangle = Exp[- &beta uangle( &thetaijk ) ]
            where &thetaijk is computed from the trial rik distance and the equilibrium bond lengths of the two missing bonds (rij0 and rjk0).
          pbond = Exp[- &beta ( uangle( &thetaijk ) + ubond( rij ) + ubond( rjk ) ) ]
            where &thetaijk = &pi, rij = rij0 * rik / ( rij0 + rjk0 ), and rjk = rjk0 * rik / ( rij0 + rjk0 ),
          minbias is a minimum value set in the code in order to avoid division by zero. This is set in the febias.F subroutine and currently has a value of 1.0d-40.
        • 'analytic Boltzmann dihedral energy sum' : Additional biasing is utilized during a configurational-bias move step that involves the growth of a new atom that is separated from an already existing atom by two bonds where the atom between those two bonds does not currently exist. The biasing has the following form where i is the atom being grown, h is an atom that already exists and is the atom from which i is being grown this step, k is an atom that already exists, and j is an atom that is bonded to i and j, but has not yet been grown.
          pbias(rij) = Min[minbias,paBdes].
          paBdes
            = pdihedral for rik < rij0 + rjk0
            = minbias for rij0 + rjk0 < rik
          pdihedral = Exp[- &beta (udihedral(&phihijk(1)) + udihedral(&phihijk(2)) )]
            where &phihijk(1) and &phihijk(2) are the two possible solutions for that dihedral given the following constraints,
              rij = rij0
              rjk = rjk0
              &thetahij = &thetahij0
        • 'autofit gaussian' : Additional biasing is utilized during a configurational-bias move step that involves the growth of a new atom that is separated from an already existing atom by two bonds, where the atom between those two bonds does not currently exist. The biasing is a gaussian distribution that is fit to the following distribution where i is the atom being grown, k is an atom that already exists, and j is an atom that is bonded to i and j, but has not yet been grown.
          pag = rij2 rjk2 &thetaijk Exp[-&beta ( ubond(rij) + ubond(rjk) + uangle(&thetaijk) )]
          where this distribution is computed at the beginning of the simulation and then the mean and standard deviations are stored for use during the configurational-bias growth procedure. When using this option you must also include the following additional variables.
            two_bond_bias_sdev_multiplier (double precision)
            • This value is multiplied by the standard deviation determined from the fit in order to get the standard deviation that is used during the simulation. Must be positive.
            two_bond_bias_vibrange (double precision array)
            • These are the lower and upper bounds of sampling for the rij and rjk bond lengths that are varied at the start of the simulation in order to determine the gaussian fit parameters. They are in units relative to the equilibrium bond lengths. Suggested default values are 0.5 and 1.5.
        • 'self adapting gaussian using 1-3 distance' : Additional biasing is utilized during a configurational-bias move step that involves the growth of a new atom that is separated from an already existing atom by two bonds, where the atom between those two bonds does not currently exist. The biasing is a gaussian distribution based upon the distance between the atom being grown and the target atom that is two bonds away and already exists. This gaussian distribution is self adapted during the course of a simulation so that it represents the observed 1-3 distance distribution. When using this option you must also include the following additional variables.
            two_bond_bias_sdev_multiplier (double precision)
            • This value is multiplied by the standard deviation determined from the fit in order to get the standard deviation that is used during the simulation. Must be positive.
            two_bond_bias_initial_value (character*50)
            • 'file' : the initial distribution is read from the 'towhee_safe_initial' file in the local directory. This file is generally copied from the 'towhee_safe_final' file that is produced at the end of a simulation that employed this two_bond_fixed_endpoint_bias_style.
            • 'autofit gaussian' : the initial distribution is generated by automatically fitting a gaussian to a sampling of the rik distance. This option is recommended when starting a new simulation where there is no appropriate 'towhee_safe_initial' file available.
            two_bond_bias_compute_frequency (integer)
            • Statistics on the two bond bias distributions are taken from the simulation with a step frequency equal to this value. These statistics are then periodically used to update the distributions, as described in the two_bond_bias_update_frequency section. Set this value to 0 if you wish to disable the periodic computation of these distributions.
            two_bond_bias_update_frequency (integer)
            • The distributions used to perform the two bond biasing are updated with this step frequency. The updates combine the distributions computed with a frequency controlled by the two_bond_bias_compute_frequency variable with the previous version of the two bond biasing potentials to create the new two bond biasing potential. Set to 0 to disable this update.
            two_bond_bias_old_fraction (double precision)
            • This factor determines the linear combination of the old distribution, and the observed distribution, to determine the new two bond bias distribution. This number must be in the range [0.0,1.0] inclusive. Setting this value to 0.0 would completely replace the old distribution with the new distribution, while a setting of 0.5 would combine the old and observed distributions equally in order to compute the new distribution.
        three_bond_fixed_endpoint_bias_style (character*50)
        • 'none' : No additional biasing is utilized during a configurational-bias move step that involves the growth of a new atom that is separated from an already existing atom by three bonds where the atoms between those three bonds do not currently exist.
        • 'analytic using max and min 2-4 distance' : Additional biasing is utilized during a configurational-bias move step that involves the growth of a new atom that is separated from an already existing atom by three bonds where the atoms between those three bonds do not currently exist. The biasing has the following form where i is the atom being grown, h is an atom that already exists and is the atom from which i is being grown this step, l is an atom that already exists, and j and k are atoms that have not yet been regrown and bridge the gap between atoms i and l. The biasing with this option depends upon the minimum (rjlmin) and maximum (rjlmax) projected distances between the j and l atoms given the constraints that rij and &thetahij are set to their equilibrium values (rij0 and &thetahij0). These minimum and maximum distances are then compared with the equilibrium distance between the j and l atoms (rjleq). This equilibrium distance is computed by setting rjk = rjk0, rkl = rkl0, and &thetajkl = &thetajkl0). The biasing value is computed as follows.
          paumam24d
            = pstretch for rjleq < rjlmin
            = 1.0 for rjlmin < rjleq < rjlmax
            = pcompress for rjlmax < rjleq
        pstretch
          = Exp[- &beta uangle(&thetajkl)] where &thetajkl is computed given the constraints of rjk0, rkl0, and rjlmin for rjlmin & lt; rjk0 + rkl0.
          = Exp[- &beta ( uangle(&thetajkl) + ubond(rjk) + ubond(rkl) )] where &thetajkl = &pi, rjk = rjk0 * rjlmin / ( rjk0 + rkl0 ), and rkl = rkl0 * rjlmin / ( rjk0 + rkl0 ) for rjk0 + rkl0 < rjlmin.
        pcompress = Exp[- &beta uangle(&thetajkl) ]
          where &thetajkl is computed using the constraints rjk0, rkl0, and rjlmax.
      • 'autofit gaussian using max and min 2-4 distance' : Additional biasing is utilized during a configurational-bias move step that involves the growth of a new atom that is separated from an already existing atom by three bonds where the atoms between those three bonds do not currently exist. The biasing has the following form where i is the atom being grown, h is an atom that already exists and is the atom from which i is being grown this step, l is an atom that already exists, and j and k are atoms that have not yet been regrown and bridge the gap between atoms i and l. The biasing with this option depends upon the minimum (rjlmin) and maximum (rjlmax) projected distances between the j and l atoms given the constraints that rij and &thetahij are set to their equilibrium values (rij0 and &thetahij0). These extrema are then combined with the 'autofit gaussian' biasing described for the two_bond_fixed_endpoint_bias_style option. The bias probability is set to the integrated probability of that autofit gaussian on the limits between rjlmin and rjlmax. When using this option you also need to include the following variable.
          three_bond_bias_sdev_multiplier
          • This value is multiplied by the standard deviation determined from the fit in order to get the standard deviation that is used during the simulation. Must be positive.
      • 'self adapting gaussian using 1-4 distance' : Additional biasing is utilized during a configurational-bias move step that involves the growth of a new atom that is separated from an already existing atom by three bonds where the atoms between those three bonds do not currently exist. The biasing has the following form where i is the atom being grown, h is an atom that already exists and is the atom from which i is being grown this step, l is an atom that already exists, and j and k are atoms that have not yet been regrown and bridge the gap between atoms i and l. The biasing is a gaussian that depends only upon the ril distance. This gaussian distribution can be set to self-adapt during the course of the simulation so that the biasing probability reflects the observed probability for the ril distances. When using this option you also need to include the following variables.
          three_bond_bias_sdev_multiplier
          • This value is multiplied by the standard deviation determined from the fit in order to get the standard deviation that is used during the simulation. Must be positive.
          three_bond_bias_initial_value (character*50)
          • 'file' : the initial distribution is read from the 'towhee_safe_initial' file in the local directory. This file is copied from the 'towhee_safe_final' file that is generated at the end of a simulation run that employed the appropriate three_bond_fixed_endpoint_bias_style.
          • 'autofit gaussian' : the initial distribution is automatically fit to the Boltzmann weight as a function of ril distances using the equilibrium rij, rjk, and rkl bond lengths and the equilibrium &thetaijk and &thetajkl angles. This option is recommended for the initial simulation when no appropriate 'towhee_safe_initial' file exists.
          three_bond_bias_compute_frequency (integer)
          • Statistics on the three bond bias distributions are taken from the simulation with a step frequency equal to this value. These statistics are then periodically used to update the distributions, as described in the three_bond_bias_update_frequency section. Set this value to 0 if you wish to disable the periodic computation of these distributions.
          three_bond_bias_update_frequency (integer)
          • The distributions used to perform the three bond biasing are updated with this step frequency. The updates combined the distributions computed with a frequency controlled by the three_bond_bias_compute_frequency variable with the previous version of the three bond biasing potentials to create the new three bond biasing potential. Set to 0 to disable this update.
          three_bond_bias_old_fraction (double precision)
          • This factor determines the linear combination of the old distribution, and the observed distribution, to determine the new three bond bias distribution. This number must be in the range [0.0,1.0] inclusive. Setting this value to 0.0 would completely replace the old distribution with the new distribution, while a setting of 0.5 would combine the old and observed distributions equally in order to compute the new distribution.

 

The final section of towhee_input contains the information that is used to construct the forcefield for the molecule types in the system. The choice of input_style determines which other variables are required to describe the molecule. Click on the appropriate link for each input_style to learn about the remaining variables that are required for each case.
input_style (character*50)
Return to the main towhee web page

 


Send comments to: Marcus G. Martin

Last updated: August 25, 2016