fix ave/spatial command¶
Syntax¶
fix ID group-ID ave/spatial Nevery Nrepeat Nfreq dim origin delta ... value1 value2 ... keyword args ...
- ID, group-ID are documented in fix command
- ave/spatial = style name of this fix command
- Nevery = use input values every this many timesteps
- Nrepeat = # of times to use input values for calculating averages
- Nfreq = calculate averages every this many timesteps
- dim, origin, delta can be repeated 1, 2, or 3 times for 1d, 2d, or 3d bins
dim = x or y or z origin = lower or center or upper or coordinate value (distance units) delta = thickness of spatial bins in dim (distance units)
- one or more input values can be listed
- value = vx, vy, vz, fx, fy, fz, density/mass, density/number, c_ID, c_ID[I], f_ID, f_ID[I], v_name
vx,vy,vz,fx,fy,fz = atom attribute (velocity, force component)
density/number, density/mass = number or mass density
c_ID = per-atom vector calculated by a compute with ID
c_ID[I] = Ith column of per-atom array calculated by a compute with ID
f_ID = per-atom vector calculated by a fix with ID
f_ID[I] = Ith column of per-atom array calculated by a fix with ID
v_name = per-atom vector calculated by an atom-style variable with name
- zero or more keyword/arg pairs may be appended
- keyword = region or bound or discard or norm or ave or units or file or overwrite or title1 or title2 or title3
region arg = region-ID bound args = x/y/z lo hi x/y/z = x or y or z to bound bins in this dimension lo = lower or coordinate value (distance units) hi = upper or coordinate value (distance units) discard arg = mixed or no or yes mixed = discard atoms outside bins only if bin bounds are explicitly set no = always keep out-of-bounds atoms yes = always discard out-of-bounds atoms norm arg = all or sample region-ID = ID of region atoms must be in to contribute to spatial averaging ave args = one or running or window M one = output new average value every Nfreq steps running = output cumulative average of all previous Nfreq steps window M = output average of M most recent Nfreq steps units arg = box or lattice or reduced file arg = filename filename = file to write results to overwrite arg = none = overwrite output file with only latest output title1 arg = string string = text to print as 1st line of output file title2 arg = string string = text to print as 2nd line of output file title3 arg = string string = text to print as 3rd line of output file
Examples¶
fix 1 all ave/spatial 10000 1 10000 z lower 0.02 c_myCentro units reduced &
title1 "My output values"
fix 1 flow ave/spatial 100 10 1000 y 0.0 1.0 vx vz norm sample file vel.profile
fix 1 flow ave/spatial 100 5 1000 z lower 1.0 y 0.0 2.5 density/mass ave running
fix 1 flow ave/spatial 100 5 1000 z lower 1.0 y 0.0 2.5 density/mass bound y 5.0 20.0 discard yes ave running
NOTE:
The fix ave/spatial command has been replaced by the more flexible fix ave/chunk and compute chunk/atom commands. The fix ave/spatial command will be removed from LAMMPS sometime in the summer of 2015.
Any fix ave/spatial command can be replaced by the two new commands. You simply need to split the fix ave/spatial arguments across the two new commands. For example, this command:
fix 1 flow ave/spatial 100 10 1000 y 0.0 1.0 vx vz norm sample file vel.profile
could be replaced by:
compute cc1 flow chunk/atom bin/1d y 0.0 1.0
fix 1 flow ave/chunk 100 10 1000 cc1 vx vz norm sample file vel.profile
Description¶
Use one or more per-atom vectors as inputs every few timesteps, bin their values spatially into 1d, 2d, or 3d bins based on current atom coordinates, and average the bin values over longer timescales. The resulting bin averages can be used by other output commands such as thermo_style custom, and can also be written to a file.
The group specified with the command means only atoms within the group contribute to bin averages. If the region keyword is used, the atom must be in both the specified group and the specified geometric region in order to contribute to bin averages.
Each listed value can be an atom attribute (position, velocity, force component), a mass or number density, or the result of a compute or fix or the evaluation of an atom-style variable. In the latter cases, the compute, fix, or variable must produce a per-atom quantity, not a global quantity. If you wish to time-average global quantities from a compute, fix, or variable, then see the fix ave/time command.
Computes that produce per-atom quantities are those which have the word atom in their style name. See the doc pages for individual fixes to determine which ones produce per-atom quantities. Variables of style atom are the only ones that can be used with this fix since all other styles of variable produce global quantities.
The per-atom values of each input vector are binned and averaged independently of the per-atom values in other input vectors.
The size and dimensionality of the bins (1d = layers or slabs, 2d = pencils, 3d = boxes) are determined by the dim, origin, and delta settings and how many times they are specified (1, 2, or 3). See details below.
Note
This fix works by creating an array of size Nbins by Nvalues on each processor. Nbins is the total number of bins; Nvalues is the number of input values specified. Each processor loops over its atoms, tallying its values to the appropriate bin. Then the entire array is summed across all processors. This means that using a large number of bins (easy to do for 2d or 3d bins) will incur an overhead in memory and computational cost (summing across processors), so be careful to use reasonable numbers of bins.
The Nevery, Nrepeat, and Nfreq arguments specify on what timesteps the input values will be used to bin them and contribute to the average. The final averaged quantities are generated on timesteps that are a multiples of Nfreq. The average is over Nrepeat quantities, computed in the preceding portion of the simulation every Nevery timesteps. Nfreq must be a multiple of Nevery and Nevery must be non-zero even if Nrepeat is 1. Also, the timesteps contributing to the average value cannot overlap, i.e. Nrepeat*Nevery can not exceed Nfreq.
For example, if Nevery=2, Nrepeat=6, and Nfreq=100, then values on timesteps 90,92,94,96,98,100 will be used to compute the final average on timestep 100. Similarly for timesteps 190,192,194,196,198,200 on timestep 200, etc. If Nrepeat=1 and Nfreq = 100, then no time averaging is done; values are simply generated on timesteps 100,200,etc.
Each per-atom property is also averaged over atoms in each bin. The way the averaging is one across the Nrepeat timesteps to produce output on the Nfreq timesteps, and across multiple Nfreq outputs, is determined by the norm and av keyword settings, as discussed below.
Bins can be 1d layers or slabs, 2d pencils, or 3d boxes. This depends on how many times (1, 2, or 3) the dim, origin, and delta settings are specified in the fix ave/spatial command. For 2d or 3d bins, there is no restriction on specifying dim = x before dim = y, or dim = y before dim = z. Bins in a particular dim have a bin size in that dimension given by delta. Every Nfreq steps, when averaging is being performed and the per-atom property is calculated for the first time, the number of bins and the bin sizes and boundaries are computed. Thus if the simulation box changes size during a simulation, the number of bins and their boundaries may also change. In each dimension, bins are defined relative to a specified origin, which may be the lower/upper edge of the simulation box in that dimension, or its center point, or a specified coordinate value. Starting at the origin, sufficient bins are created in both directions to completely span the bin extent in that dimension. By default the bin extent is the entire simulation box.
The bound keyword can be used one or more times to limit the extent of bin coverage in specified dimensions, i.e. to only bin a portion of the box. If the lo setting is lower or the hi setting is upper, the bin extent in that direction extends to the box boundary. If a numeric value is used for lo and/or hi, then the bin extent in the lo or hi direction extends only to that value, which is assumed to be inside (or at least near) the simulation box boundaries, though LAMMPS does not check for this.
On each sampling timestep, each atom is mapped to the bin it currently belongs to, based on its current position. Note that the group-ID and region keyword can exclude specific atoms from this operation, as discussed above. Note that between reneighboring timesteps, atoms can move outside the current simulation box. If the box is periodic (in that dimension) the atom is remapping into the periodic box for purposes of binning. If the box in not periodic, the atom may have moved outside the bounds of any bin.
The discard keyword determines what is done with any atom which is outside the bounds of any bin. If discard is set to yes, the atom will be ignored and not contribute to any bin averages. If discard is set to no, the atom will be counted as if it were in the first or last bin in that dimension. If (discard* is set to mixed, which is the default, it will only be counted in the first or last bin if bins extend to the box boundary in that dimension. This is the case if the bound keyword settings are lower and upper, which is the default. If the bound keyword settings are numeric values, then the atom will be ignored if it is outside the bounds of any bin. Note that in this case, it is possible that the first or last bin extends beyond the numeric bounds settings, depending on the specified origin. If this is the case, the atom is only ignored if it is outside the first or last bin, not if it is simply outside the numeric bounds setting.
For orthogonal simulation boxes, the bins are also layers, pencils, or boxes aligned with the xyz coordinate axes. For triclinic (non-orthogonal) simulation boxes, the bins are so that they are parallel to the tilted faces of the simulation box. See this section of the manual for a discussion of the geometry of triclinic boxes in LAMMPS. As described there, a tilted simulation box has edge vectors a,b,c. In that nomenclature, bins in the x dimension have faces with normals in the “b” cross “c” direction. Bins in y have faces normal to the “a” cross “c” direction. And bins in z have faces normal to the “a” cross “b” direction. Note that in order to define the size and position of these bins in an unambiguous fashion, the units option must be set to reduced when using a triclinic simulation box, as noted below.
The atom attribute values (vx,vy,vz,fx,fy,fz) are self-explanatory. Note that other atom attributes (including atom postitions x,y,z) can be used as inputs to this fix by using the compute property/atom command and then specifying an input value from that compute.
The density/number value means the number density is computed in each bin, i.e. a weighting of 1 for each atom. The density/mass value means the mass density is computed in each bind, i.e. each atom is weighted by its mass. The resulting density is normalized by the volume of the bin so that units of number/volume or density are output. See the units command doc page for the definition of density for each choice of units, e.g. gram/cm^3.
If a value begins with “c_”, a compute ID must follow which has been previously defined in the input script. If no bracketed integer is appended, the per-atom vector calculated by the compute is used. If a bracketed integer is appended, the Ith column of the per-atom array calculated by the compute is used. Users can also write code for their own compute styles and add them to LAMMPS.
If a value begins with “f_”, a fix ID must follow which has been previously defined in the input script. If no bracketed integer is appended, the per-atom vector calculated by the fix is used. If a bracketed integer is appended, the Ith column of the per-atom array calculated by the fix is used. Note that some fixes only produce their values on certain timesteps, which must be compatible with Nevery, else an error results. Users can also write code for their own fix styles and add them to LAMMPS.
If a value begins with “v_”, a variable name must follow which has been previously defined in the input script. Variables of style atom can reference thermodynamic keywords and various per-atom attributes, or invoke other computes, fixes, or variables when they are evaluated, so this is a very general means of generating per-atom quantities to spatially average.
Additional optional keywords also affect the operation of this fix. The region, bound, and discard keywords were discussed above.
The norm keyword affects how averaging is done for the output produced every Nfreq timesteps. For an all setting, a bin quantity is summed over all atoms in all Nrepeat samples, as is the count of atoms in the bin. The printed value for the bin is Total-quantity / Total-count. In other words it is an average over the entire Nfreq timescale.
For a sample setting, the bin quantity is summed over atoms for only a single sample, as is the count, and a “average sample value” is computed, i.e. Sample-quantity / Sample-count. The printed value for the bin is the average of the Nrepeat “average sample values”, In other words it is an average of an average.
The ave keyword determines how the bin values produced every Nfreq steps are averaged with bin values produced on previous steps that were multiples of Nfreq, before they are accessed by another output command or written to a file.
If the ave setting is one, then the bin values produced on timesteps that are multiples of Nfreq are independent of each other; they are output as-is without further averaging.
If the ave setting is running, then the bin values produced on timesteps that are multiples of Nfreq are summed and averaged in a cumulative sense before being output. Each output bin value is thus the average of the bin value produced on that timestep with all preceding values for the same bin. This running average begins when the fix is defined; it can only be restarted by deleting the fix via the unfix command, or re-defining the fix by re-specifying it.
If the ave setting is window, then the bin values produced on timesteps that are multiples of Nfreq are summed and averaged within a moving “window” of time, so that the last M values for the same bin are used to produce the output. E.g. if M = 3 and Nfreq = 1000, then the output on step 10000 will be the average of the individual bin values on steps 8000,9000,10000. Outputs on early steps will average over less than M values if they are not available.
The units keyword determines the meaning of the distance units used for the bin size delta and for origin and bounds values if they are coordinate value. For orthogonal simulation boxes, any of the 3 options may be used. For non-orthogonal (triclinic) simulation boxes, only the reduced option may be used.
A box value selects standard distance units as defined by the units command, e.g. Angstroms for units = real or metal. A lattice value means the distance units are in lattice spacings. The lattice command must have been previously used to define the lattice spacing. A reduced value means normalized unitless values between 0 and 1, which represent the lower and upper faces of the simulation box respectively. Thus an origin value of 0.5 means the center of the box in any dimension. A delta value of 0.1 means 10 bins span the box in that dimension.
Consider a non-orthogonal box, with bins that are 1d layers or slabs in the x dimension. No matter how the box is tilted, an origin of 0.0 means start layers at the lower “b” cross “c” plane of the simulation box and an origin of 1.0 means to start layers at the upper “b” cross “c” face of the box. A delta value of 0.1 means there will be 10 layers from 0.0 to 1.0, regardless of the current size or shape of the simulation box.
The file keyword allows a filename to be specified. Every Nfreq timesteps, a section of bin info will be written to a text file in the following format. A line with the timestep and number of bin is written. Then one line per bin is written, containing the bin ID (1-N), the coordinate of the center of the bin, the number of atoms in the bin, and one or more calculated values. The number of values in each line corresponds to the number of values specified in the fix ave/spatial command. The number of atoms and the value(s) are average quantities. If the value of the units keyword is box or lattice, the “coord” is printed in box units. If the value of the units keyword is reduced, the “coord” is printed in reduced units (0-1).
The overwrite keyword will continuously overwrite the output file with the latest output, so that it only contains one timestep worth of output. This option can only be used with the ave running setting.
The title1 and title2 and title3 keywords allow specification of the strings that will be printed as the first 3 lines of the output file, assuming the file keyword was used. LAMMPS uses default values for each of these, so they do not need to be specified.
By default, these header lines are as follows:
# Spatial-averaged data for fix ID and group name
# Timestep Number-of-bins
# Bin Coord1 Coord2 Coord3 Count value1 value2 ...
In the first line, ID and name are replaced with the fix-ID and group name. The second line describes the two values that are printed at the first of each section of output. In the third line the values are replaced with the appropriate fields from the fix ave/spatial command. The Coord2 and Coord3 entries in the third line only appear for 2d and 3d bins respectively. For 1d bins, the word Coord1 is replaced by just Coord.
Restart, fix_modify, output, run start/stop, minimize info¶
No information about this fix is written to binary restart files. None of the fix_modify options are relevant to this fix.
This fix computes a global array of values which can be accessed by various output commands. The values can only be accessed on timesteps that are multiples of Nfreq since that is when averaging is performed. The global array has # of rows = Nbins and # of columns = Ndim+1+Nvalues, where Ndim = 1,2,3 for 1d,2d,3d bins. The first 1 or 2 or 3 columns have the bin coordinates (center of the bin) in the appropriate dimensions, the next column has the count of atoms in that bin, and the remaining columns are the Nvalue quantities. When the array is accessed with an I that exceeds the current number of bins, than a 0.0 is returned by the fix instead of an error, since the number of bins can vary as a simulation runs, depending on the simulation box size. 2d or 3d bins are ordered so that the last dimension(s) vary fastest. The array values calculated by this fix are “intensive”, since they are already normalized by the count of atoms in each bin.
No parameter of this fix can be used with the start/stop keywords of the run command. This fix is not invoked during energy minimization.
Restrictions¶
When the ave keyword is set to running or window then the number of bins must remain the same during the simulation, so that the appropriate averaging can be done. This will be the case if the simulation box size doesn’t change or if the units keyword is set to reduced.
Default¶
The option defaults are bound = lower and upper in all dimensions, discard = mixed, norm = all, ave = one, units = lattice, no file output, and title 1,2,3 = strings as described above.