\(\renewcommand{\AA}{\text{Å}}\)

1.1.5. Compute, fixes, variables

This section documents accessing or modifying data stored by computes, fixes, or variables in LAMMPS using the following functions:


void *lammps_extract_compute(void *handle, const char*, int, int)

Get pointer to data from a LAMMPS compute.

This function returns a pointer to the location of data provided by a compute command instance identified by the compute-ID. Computes may provide global, per-atom, or local data, and those may be a scalar, a vector, or an array or they may provide the information about the dimensions of the respective data. Since computes may provide multiple kinds of data, it is required to set style and type flags representing what specific data is desired. This also determines to what kind of pointer the returned pointer needs to be cast to access the data correctly. The function returns NULL if the compute ID is not found or the requested data is not available or current. The following table lists the available options.

Style (see _LMP_STYLE_CONST)

Type (see _LMP_TYPE_CONST)

Returned type

Returned data

LMP_STYLE_GLOBAL

LMP_TYPE_SCALAR

double *

Global scalar

LMP_STYLE_GLOBAL

LMP_TYPE_VECTOR

double *

Global vector

LMP_STYLE_GLOBAL

LMP_TYPE_ARRAY

double **

Global array

LMP_STYLE_GLOBAL

LMP_SIZE_VECTOR

int *

Length of global vector

LMP_STYLE_GLOBAL

LMP_SIZE_ROWS

int *

Rows of global array

LMP_STYLE_GLOBAL

LMP_SIZE_COLS

int *

Columns of global array

LMP_STYLE_ATOM

LMP_TYPE_VECTOR

double *

Per-atom value

LMP_STYLE_ATOM

LMP_TYPE_ARRAY

double **

Per-atom vector

LMP_STYLE_ATOM

LMP_SIZE_COLS

int *

Columns in per-atom array, 0 if vector

LMP_STYLE_LOCAL

LMP_TYPE_VECTOR

double *

Local data vector

LMP_STYLE_LOCAL

LMP_TYPE_ARRAY

double **

Local data array

LMP_STYLE_LOCAL

LMP_SIZE_VECTOR

int *

Alias for using LMP_SIZE_ROWS

LMP_STYLE_LOCAL

LMP_SIZE_ROWS

int *

Number of local array rows or length of vector

LMP_STYLE_LOCAL

LMP_SIZE_COLS

int *

Number of local array columns, 0 if vector

Note

If the compute’s data is not computed for the current step, the compute will be invoked. LAMMPS cannot easily check at that time, if it is valid to invoke a compute, so it may fail with an error. The caller has to check to avoid such an error.

Warning

The pointers returned by this function are generally not persistent since the computed data may be re-distributed, re-allocated, and re-ordered at every invocation. It is advisable to re-invoke this function before the data is accessed, or make a copy if the data shall be used after other LAMMPS commands have been issued.

Parameters:
  • handle – pointer to a previously created LAMMPS instance

  • id – string with ID of the compute

  • style – constant indicating the style of data requested (global, per-atom, or local)

  • type – constant indicating type of data (scalar, vector, or array) or size of rows or columns

Returns:

pointer (cast to void *) to the location of the requested data or NULL if not found.


void *lammps_extract_fix(void *handle, const char*, int, int, int, int)

Get pointer to data from a LAMMPS fix.

This function returns a pointer to data provided by a fix command instance identified by its fix-ID. Fixes may provide global, per-atom, or local data, and those may be a scalar, a vector, or an array, or they may provide the information about the dimensions of the respective data. Since individual fixes may provide multiple kinds of data, it is required to set style and type flags representing what specific data is desired. This also determines to what kind of pointer the returned pointer needs to be cast to access the data correctly. The function returns NULL if the fix ID is not found or the requested data is not available.

The following table lists the available options.

Style (see _LMP_STYLE_CONST)

Type (see _LMP_TYPE_CONST)

Returned type

Returned data

LMP_STYLE_GLOBAL

LMP_TYPE_SCALAR

double *

Copy of global scalar

LMP_STYLE_GLOBAL

LMP_TYPE_VECTOR

double *

Copy of global vector element at index nrow

LMP_STYLE_GLOBAL

LMP_TYPE_ARRAY

double *

Copy of global array element at nrow, ncol

LMP_STYLE_GLOBAL

LMP_SIZE_VECTOR

int *

Length of global vector

LMP_STYLE_GLOBAL

LMP_SIZE_ROWS

int *

Rows in global array

LMP_STYLE_GLOBAL

LMP_SIZE_COLS

int *

Columns in global array

LMP_STYLE_ATOM

LMP_TYPE_VECTOR

double *

Per-atom value

LMP_STYLE_ATOM

LMP_TYPE_ARRAY

double **

Per-atom vector

LMP_STYLE_ATOM

LMP_SIZE_COLS

int *

Columns of per-atom array, 0 if vector

LMP_STYLE_LOCAL

LMP_TYPE_VECTOR

double *

Local data vector

LMP_STYLE_LOCAL

LMP_TYPE_ARRAY

double **

Local data array

LMP_STYLE_LOCAL

LMP_SIZE_ROWS

int *

Number of local data rows

LMP_STYLE_LOCAL

LMP_SIZE_COLS

int *

Number of local data columns

Note

When requesting global data, the fix data can only be accessed one item at a time without access to the pointer itself. Thus this function will allocate storage for a single double value, copy the returned value to it, and returns a pointer to the location of the copy. Therefore the allocated storage needs to be freed after its use to avoid a memory leak. Example:

double *dptr = (double *) lammps_extract_fix(handle,name,0,1,0,0);
double value = *dptr;
lammps_free((void *)dptr);

Note

LAMMPS cannot easily check if it is valid to access the data, so it may fail with an error. The caller has to avoid such an error.

Warning

The pointers returned by this function for per-atom or local data are generally not persistent, since the computed data may be re-distributed, re-allocated, and re-ordered at every invocation of the fix. It is thus advisable to re-invoke this function before the data is accessed, or make a copy, if the data shall be used after other LAMMPS commands have been issued.

Parameters:
  • handle – pointer to a previously created LAMMPS instance

  • id – string with ID of the fix

  • style – constant indicating the style of data requested (global, per-atom, or local)

  • type – constant indicating type of data (scalar, vector, or array) or size of rows or columns

  • nrow – row index (only used for global vectors and arrays)

  • ncol – column index (only used for global arrays)

Returns:

pointer (cast to void *) to the location of the requested data or NULL if not found.


int lammps_extract_variable_datatype(void *handle, const char *name)

Get data type of a LAMMPS variable.

New in version 3Nov2022.

This function returns an integer that encodes the data type of the variable with the specified name. See _LMP_VAR_CONST for valid values. Callers of lammps_extract_variable() can use this information to decide how to cast the void * pointer and access the data.

Parameters:
  • handle – pointer to a previously created LAMMPS instance

  • name – string with the name of the extracted variable

Returns:

integer constant encoding the data type of the property or -1 if not found.


void *lammps_extract_variable(void *handle, const char*, const char*)

Get pointer to data from a LAMMPS variable.

This function returns a pointer to data from a LAMMPS variable command identified by its name. When the variable is either an equal-style compatible variable, a vector-style variable, or an atom-style variable, the variable is evaluated and the corresponding value(s) returned. Variables of style internal are compatible with equal-style variables and so are python-style variables, if they return a numeric value. For other variable styles, their string value is returned. The function returns NULL when a variable of the provided name is not found or of an incompatible style. The group argument is only used for atom-style variables and ignored otherwise, with one exception: for style vector, if group is “GET_VECTOR_SIZE”, the returned pointer will yield the length of the vector to be returned when dereferenced. This pointer must be deallocated after the value is read to avoid a memory leak. If group is set to NULL when extracting data from an atom-style variable, the group is assumed to be “all”.

When requesting data from an equal-style or compatible variable this function allocates storage for a single double value, copies the returned value to it, and returns a pointer to the location of the copy. Therefore the allocated storage needs to be freed after its use to avoid a memory leak. Example:

double *dptr = (double *) lammps_extract_variable(handle,name,NULL);
double value = *dptr;
lammps_free((void *)dptr);

For atom-style variables, the return value is a pointer to an allocated block of storage of double of the length atom->nlocal. Since the data returned are a copy, the location will persist, but its content will not be updated in case the variable is re-evaluated. To avoid a memory leak, this pointer needs to be freed after use in the calling program.

For vector-style variables, the returned pointer is to actual LAMMPS data. The pointer should not be deallocated. Its length depends on the variable, compute, or fix data used to construct the vector-style variable. This length can be fetched by calling this function with group set to the constant “LMP_SIZE_VECTOR”, which returns a void * pointer that can be dereferenced to an integer that is the length of the vector. This pointer needs to be deallocated when finished with it to avoid memory leaks.

For other variable styles the returned pointer needs to be cast to a char pointer. It should not be deallocated.

const char *cptr = (const char *) lammps_extract_variable(handle,name,NULL);
printf("The value of variable %s is %s\n", name, cptr);

Note

LAMMPS cannot easily check if it is valid to access the data referenced by the variables (e.g., computes, fixes, or thermodynamic info), so it may fail with an error. The caller has to make certain that the data are extracted only when it safe to evaluate the variable and thus an error or crash are avoided.

Parameters:
  • handle – pointer to a previously created LAMMPS instance

  • name – name of the variable

  • group – group-ID for atom style variable or NULL

Returns:

pointer (cast to void *) to the location of the requested data or NULL if not found.


int lammps_set_variable(void *handle, char *name, char *str)

Set the value of a string-style variable.

This function assigns a new value from the string str to the string-style variable name. Returns -1 if a variable of that name does not exist or is not a string-style variable, otherwise 0.

Parameters:
  • handle – pointer to a previously created LAMMPS instance

  • name – name of the variable

  • str – new value of the variable

Returns:

0 on success or -1 on failure


int lammps_variable_info(void *handle, int idx, char *buf, int bufsize)

Retrieve informational string for a variable.

New in version TBD.

This function copies a string with human readable information about a defined variable: name, style, current value(s) into the provided C-style string buffer. That is the same info as produced by the info variables command. The length of the buffer must be provided as buf_size argument. If the info exceeds the length of the buffer, it will be truncated accordingly. If the index is out of range, the function returns 0 and buffer is set to an empty string, otherwise 1.

Parameters:
  • handle – pointer to a previously created LAMMPS instance cast to void *.

  • idx – index of the variable (0 <= idx < nvar)

  • buffer – string buffer to copy the info to

  • buf_size – size of the provided string buffer

Returns:

1 if successful, otherwise 0


enum LAMMPS_NS::multitype::_LMP_DATATYPE_CONST

Data type constants for extracting data from atoms, computes and fixes

This enum must be kept in sync with the corresponding enum or constants in python/lammps/constants.py, fortran/lammps.f90, tools/swig/lammps.i, src/library.h, and examples/COUPLE/plugin/liblammpsplugin.h

Values:

enumerator LAMMPS_NONE

no data type assigned (yet)

enumerator LAMMPS_INT

32-bit integer (array)

enumerator LAMMPS_INT_2D

two-dimensional 32-bit integer array

enumerator LAMMPS_DOUBLE

64-bit double (array)

enumerator LAMMPS_DOUBLE_2D

two-dimensional 64-bit double array

enumerator LAMMPS_INT64

64-bit integer (array)

enumerator LAMMPS_INT64_2D

two-dimensional 64-bit integer array

enumerator LAMMPS_STRING

C-String

enum _LMP_STYLE_CONST

Style constants for extracting data from computes and fixes.

Must be kept in sync with the equivalent constants in python/lammps/constants.py, fortran/lammps.f90, tools/swig/lammps.i, and examples/COUPLE/plugin/liblammpsplugin.h

Values:

enumerator LMP_STYLE_GLOBAL

return global data

enumerator LMP_STYLE_ATOM

return per-atom data

enumerator LMP_STYLE_LOCAL

return local data

enum _LMP_TYPE_CONST

Type and size constants for extracting data from computes and fixes.

Must be kept in sync with the equivalent constants in python/lammps/constants.py, fortran/lammps.f90, tools/swig/lammps.i, and examples/COUPLE/plugin/liblammpsplugin.h

Values:

enumerator LMP_TYPE_SCALAR

return scalar

enumerator LMP_TYPE_VECTOR

return vector

enumerator LMP_TYPE_ARRAY

return array

enumerator LMP_SIZE_VECTOR

return length of vector

enumerator LMP_SIZE_ROWS

return number of rows

enumerator LMP_SIZE_COLS

return number of columns

enum _LMP_VAR_CONST

Variable style constants for extracting data from variables.

Must be kept in sync with the equivalent constants in python/lammps/constants.py, fortran/lammps.f90, tools/swig/lammps.i, and examples/COUPLE/plugin/liblammpsplugin.h

Values:

enumerator LMP_VAR_EQUAL

compatible with equal-style variables

enumerator LMP_VAR_ATOM

compatible with atom-style variables

enumerator LMP_VAR_VECTOR

compatible with vector-style variables

enumerator LMP_VAR_STRING

return value will be a string (catch-all)