Commit 30177c4e authored by sjplimp's avatar sjplimp Committed by GitHub
Browse files

Merge pull request #521 from pastewka/17_dump_nc 

Updated NetCDF dump style (dump netcdf)
parents 178eff23 bdf03757

doc/src/dump_modify.txt

+17 −7
Original line number Diff line number Diff line
@@ -16,7 +16,8 @@ dump-ID = ID of dump to modify :ulb,l 
one or more keyword/value pairs may be appended :l

these keywords apply to various dump styles :l

keyword = {append} or {buffer} or {element} or {every} or {fileper} or {first} or {flush} or {format} or {image} or {label} or {nfile} or {pad} or {precision} or {region} or {scale} or {sort} or {thresh} or {unwrap} :l

  {append} arg = {yes} or {no}

  {append} arg = {yes} or {no} or {at} N

    N = index of frame written upon first dump

  {buffer} arg = {yes} or {no}

  {element} args = E1 E2 ... EN, where N = # of atom types

    E1,...,EN = element name, e.g. C or Fe or Ga
@@ -41,6 +42,7 @@ keyword = {append} or {buffer} or {element} or {every} or {fileper} or {first} o 
  {region} arg = region-ID or "none"

  {scale} arg = {yes} or {no}

  {sfactor} arg = coordinate scaling factor (> 0.0)

  {thermo} arg = {yes} or {no}

  {tfactor} arg = time scaling factor (> 0.0)

  {sort} arg = {off} or {id} or N or -N

     off = no sorting of per-atom lines within a snapshot
@@ -139,12 +141,13 @@ and {dcd}. It also applies only to text output files, not to binary 
or gzipped or image/movie files.  If specified as {yes}, then dump

snapshots are appended to the end of an existing dump file.  If

specified as {no}, then a new dump file will be created which will

overwrite an existing file with the same name.  This keyword can only

take effect if the dump_modify command is used after the

"dump"_dump.html command, but before the first command that causes

dump snapshots to be output, e.g. a "run"_run.html or

"minimize"_minimize.html command.  Once the dump file has been opened,

this keyword has no further effect.

overwrite an existing file with the same name.  If the {at} option is present

({netcdf} only), then the frame to append to can be specified.  Negative values

are counted from the end of the file.  This keyword can only take effect if the

dump_modify command is used after the "dump"_dump.html command, but before the

first command that causes dump snapshots to be output, e.g. a "run"_run.html or

"minimize"_minimize.html command.  Once the dump file has been opened, this

keyword has no further effect.



:line
@@ -413,6 +416,13 @@ most effective when the typical magnitude of position data is between 


:line



The {thermo} keyword ({netcdf} only) triggers writing of "thermo"_thermo.html

information to the dump file alongside per-atom data. The data included in the

dump file is identical to the data specified by

"thermo_style"_thermo_style.html.



:line



The {region} keyword only applies to the dump {custom}, {cfg},

{image}, and {movie} styles.  If specified, only atoms in the region

will be written to the dump file or included in the image/movie.  Only

doc/src/dump_netcdf.txt

+5 −11
Original line number Diff line number Diff line
@@ -24,7 +24,7 @@ args = list of atom attributes, same as for "dump_style custom"_dump.html :l,ule 
[Examples:]



dump 1 all netcdf 100 traj.nc type x y z vx vy vz

dump_modify 1 append yes at -1 global c_thermo_pe c_thermo_temp c_thermo_press

dump_modify 1 append yes at -1 thermo yes

dump 1 all netcdf/mpiio 1000 traj.nc id type x y z :pre



[Description:]
@@ -44,7 +44,7 @@ rank. 
NetCDF files can be directly visualized via the following tools:



Ovito (http://www.ovito.org/). Ovito supports the AMBER convention and

all of the above extensions. :ule,b

all extensions of this dump style. :ule,b



VMD (http://www.ks.uiuc.edu/Research/vmd/). :l
@@ -52,15 +52,9 @@ AtomEye (http://www.libatoms.org/). The libAtoms version of AtomEye 
contains a NetCDF reader that is not present in the standard

distribution of AtomEye. :l,ule



In addition to per-atom data, global data can be included in the dump

file, which are the kinds of values output by the

"thermo_style"_thermo_style.html command .  See "Section howto

6.15"_Section_howto.html#howto_15 for an explanation of per-atom

versus global data.  The global output written into the dump file can

be from computes, fixes, or variables, by prefixing the compute/fix ID

or variable name with "c_" or "f_" or "v_" respectively, as in the

example above.  These global values are specified via the "dump_modify

global"_dump_modify.html command.

In addition to per-atom data, "thermo"_thermo.html data can be included in the

dump file. The data included in the dump file is identical to the data specified

by "thermo_style"_thermo_style.html.



:link(netcdf-home,http://www.unidata.ucar.edu/software/netcdf/)

:link(pnetcdf-home,http://trac.mcs.anl.gov/projects/parallel-netcdf/)

src/USER-NETCDF/README

+8 −8
Original line number Diff line number Diff line 
USER-NETCDF

============



This package provides the netcf and netcdf/mpiio dump styles.

See the doc page for dump nc or dump nc/mpiio command for how to use them.

Compiling these dump styles requires having the netCDF library installed

This package provides the netcdf and netcdf/mpiio dump styles.

See the doc page for dump netcdf or dump netcdf/mpiio command for how to use

them. Compiling these dump styles requires having the netCDF library installed

on your system. See lib/netcdf/README for additional details.



PACKAGE DESCRIPTION
@@ -29,11 +29,11 @@ NetCDF files can be directly visualized with the following tools: 
  a NetCDF reader that is not present in the standard distribution of AtomEye.



The person who created these files is Lars Pastewka at

Karlsruhe Institute of Technology (lars.pastewka@kit.edu).

the University of Freiburg (lars.pastewka@imtek.uni-freiburg.de).

Contact him directly if you have questions.



Lars Pastewka

Institute for Applied Materials (IAM)

Karlsruhe Institute of Technology (KIT)

Kaiserstrasse 12, 76131 Karlsruhe

e-mail: lars.pastewka@kit.edu
 
University of Freiburg

Department of Microsystems Engineering

Georges-Köhler-Allee 103, 79110 Freiburg, Germany

e-mail: lars.pastewka@imtek.uni-freiburg.de

src/USER-NETCDF/dump_netcdf.cpp

+100 −252
Original line number Diff line number Diff line 
/* ======================================================================

   LAMMPS NetCDF dump style

   https://github.com/pastewka/lammps-netcdf

   Lars Pastewka, lars.pastewka@kit.edu



   Copyright (2011-2013) Fraunhofer IWM

   Copyright (2014) Karlsruhe Institute of Technology



   This program is free software: you can redistribute it and/or modify

   it under the terms of the GNU General Public License as published by

   the Free Software Foundation, either version 2 of the License, or

   (at your option) any later version.



   This program is distributed in the hope that it will be useful,

   but WITHOUT ANY WARRANTY; without even the implied warranty of

   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

   GNU General Public License for more details.



   You should have received a copy of the GNU General Public License

   along with this program.  If not, see <http://www.gnu.org/licenses/>.

   ====================================================================== */



/* ----------------------------------------------------------------------

   LAMMPS - Large-scale Atomic/Molecular Massively Parallel Simulator

   http://lammps.sandia.gov, Sandia National Laboratories
@@ -33,6 +11,10 @@ 
   See the README file in the top-level LAMMPS directory.

------------------------------------------------------------------------- */



/* ----------------------------------------------------------------------

   Contributing author: Lars Pastewka (University of Freiburg)

------------------------------------------------------------------------- */



#if defined(LMP_HAS_NETCDF)



#include <unistd.h>
@@ -55,11 +37,13 @@ 
#include "universe.h"

#include "variable.h"

#include "force.h"

#include "output.h"

#include "thermo.h"



using namespace LAMMPS_NS;

using namespace MathConst;



enum{INT,DOUBLE};  // same as in dump_custom.cpp

enum{INT,FLOAT,BIGINT}; // same as in thermo.cpp



const char NC_FRAME_STR[]         = "frame";

const char NC_SPATIAL_STR[]       = "spatial";
@@ -208,15 +192,15 @@ DumpNetCDF::DumpNetCDF(LAMMPS *lmp, int narg, char **arg) : 
    perat[inc].field[idim] = i;

  }



  n_perframe = 0;

  perframe = NULL;



  n_buffer = 0;

  int_buffer = NULL;

  double_buffer = NULL;



  double_precision = false;



  thermo = false;

  thermovar = NULL;



  framei = 0;

}
@@ -227,8 +211,7 @@ DumpNetCDF::~DumpNetCDF() 
  closefile();



  delete [] perat;

  if (n_perframe > 0)

    delete [] perframe;

  if (thermovar)  delete [] thermovar;



  if (int_buffer) memory->sfree(int_buffer);

  if (double_buffer) memory->sfree(double_buffer);
@@ -238,6 +221,11 @@ DumpNetCDF::~DumpNetCDF() 


void DumpNetCDF::openfile()

{

  if (thermo && !singlefile_opened) {

    if (thermovar)  delete [] thermovar;

    thermovar = new int[output->thermo->nfield];

  }



  // now the computes and fixes have been initialized, so we can query

  // for the size of vector quantities

  for (int i = 0; i < n_perat; i++) {
@@ -334,9 +322,12 @@ void DumpNetCDF::openfile() 
      }



      // perframe variables

      for (int i = 0; i < n_perframe; i++) {

        NCERRX( nc_inq_varid(ncid, perframe[i].name, &perframe[i].var),

                perframe[i].name );

      if (thermo) {

        Thermo *th = output->thermo;

        for (int i = 0; i < th->nfield; i++) {

          NCERRX( nc_inq_varid(ncid, th->keyword[i], &thermovar[i]),

                  th->keyword[i] );

        }

      }



      size_t nframes;
@@ -354,7 +345,7 @@ void DumpNetCDF::openfile() 
      if (singlefile_opened) return;

      singlefile_opened = 1;



      NCERRX( nc_create(filename, NC_64BIT_OFFSET, &ncid),

      NCERRX( nc_create(filename, NC_64BIT_DATA, &ncid),

          filename );



      // dimensions
@@ -471,14 +462,21 @@ void DumpNetCDF::openfile() 
      }



      // perframe variables

      for (int i = 0; i < n_perframe; i++) {

        if (perframe[i].type == THIS_IS_A_BIGINT) {

          NCERRX( nc_def_var(ncid, perframe[i].name, NC_LONG, 1, dims,

			     &perframe[i].var), perframe[i].name );

      if (thermo) {

        Thermo *th = output->thermo;

        for (int i = 0; i < th->nfield; i++) {

          if (th->vtype[i] == FLOAT) {

            NCERRX( nc_def_var(ncid, th->keyword[i], NC_DOUBLE, 1, dims,

                               &thermovar[i]), th->keyword[i] );

          }

          else if (th->vtype[i] == INT) {

            NCERRX( nc_def_var(ncid, th->keyword[i], NC_INT, 1, dims,

                               &thermovar[i]), th->keyword[i] );

          }

          else if (th->vtype[i] == BIGINT) {

            NCERRX( nc_def_var(ncid, th->keyword[i], NC_LONG, 1, dims,

                               &thermovar[i]), th->keyword[i] );

          }

        else {

          NCERRX( nc_def_var(ncid, perframe[i].name, NC_DOUBLE, 1, dims,

			     &perframe[i].var), perframe[i].name );

        }

      }
@@ -622,46 +620,30 @@ void DumpNetCDF::write() 
  start[0] = framei-1;

  start[1] = 0;



  for (int i = 0; i < n_perframe; i++) {



    if (perframe[i].type == THIS_IS_A_BIGINT) {

      bigint data;

      (this->*perframe[i].compute)((void*) &data);



      if (filewriter)

  if (thermo) {

    Thermo *th = output->thermo;

    for (int i = 0; i < th->nfield; i++) {

      th->call_vfunc(i);

      if (filewriter) {

        if (th->vtype[i] == FLOAT) {

          NCERRX( nc_put_var1_double(ncid, thermovar[i], start,

                                     &th->dvalue),

                  th->keyword[i] );

        }

        else if (th->vtype[i] == INT) {

          NCERRX( nc_put_var1_int(ncid, thermovar[i], start, &th->ivalue),

                  th->keyword[i] );

        }

        else if (th->vtype[i] == BIGINT) {

#if defined(LAMMPS_SMALLBIG) || defined(LAMMPS_BIGBIG)

        NCERR( nc_put_var1_long(ncid, perframe[i].var, start, &data) );

          NCERRX( nc_put_var1_long(ncid, thermovar[i], start, &th->bivalue),

                  th->keyword[i] );

#else

        NCERR( nc_put_var1_int(ncid, perframe[i].var, start, &data) );

          NCERRX( nc_put_var1_int(ncid, thermovar[i], start, &th->bivalue),

                  th->keyword[i] );

#endif

        }

    else {

      double data;

      int j = perframe[i].index;

      int idim = perframe[i].dim;



      if (perframe[i].type == THIS_IS_A_COMPUTE) {

        if (idim >= 0) {

          modify->compute[j]->compute_vector();

          data = modify->compute[j]->vector[idim];

      }

        else

          data = modify->compute[j]->compute_scalar();

      }

      else if (perframe[i].type == THIS_IS_A_FIX) {

        if (idim >= 0) {

          data = modify->fix[j]->compute_vector(idim);

        }

        else

          data = modify->fix[j]->compute_scalar();

      }

      else if (perframe[i].type == THIS_IS_A_VARIABLE) {

        j = input->variable->find(perframe[i].id);

        data = input->variable->compute_equal(j);

      }



      if (filewriter)

        NCERR( nc_put_var1_double(ncid, perframe[i].var, start, &data) );

    }

  }
@@ -908,126 +890,19 @@ int DumpNetCDF::modify_param(int narg, char **arg) 
    iarg++;

    return 2;

  }

  else if (strcmp(arg[iarg],"global") == 0) {

    // "perframe" quantities, i.e. not per-atom stuff



  else if (strcmp(arg[iarg],"thermo") == 0) {

    iarg++;



    n_perframe = narg-iarg;

    perframe = new nc_perframe_t[n_perframe];



    for (int i = 0; iarg < narg; iarg++, i++) {

      int n;

      char *suffix=NULL;



      if (!strcmp(arg[iarg],"step")) {

        perframe[i].type = THIS_IS_A_BIGINT;

        perframe[i].compute = &DumpNetCDF::compute_step;

        strcpy(perframe[i].name, arg[iarg]);

      }

      else if (!strcmp(arg[iarg],"elapsed")) {

        perframe[i].type = THIS_IS_A_BIGINT;

        perframe[i].compute = &DumpNetCDF::compute_elapsed;

        strcpy(perframe[i].name, arg[iarg]);

      }

      else if (!strcmp(arg[iarg],"elaplong")) {

        perframe[i].type = THIS_IS_A_BIGINT;

        perframe[i].compute = &DumpNetCDF::compute_elapsed_long;

        strcpy(perframe[i].name, arg[iarg]);

      }

      else {



        n = strlen(arg[iarg]);



        if (n > 2) {

          suffix = new char[n-1];

          strcpy(suffix, arg[iarg]+2);

        }

        else {

          char errstr[1024];

          sprintf(errstr, "perframe quantity '%s' must thermo quantity or "

                  "compute, fix or variable", arg[iarg]);

          error->all(FLERR,errstr);

        }



        if (!strncmp(arg[iarg], "c_", 2)) {

          int idim = -1;

          char *ptr = strchr(suffix, '[');



          if (ptr) {

            if (suffix[strlen(suffix)-1] != ']')

              error->all(FLERR,"Missing ']' in dump modify command");

            *ptr = '\0';

            idim = ptr[1] - '1';

          }



          n = modify->find_compute(suffix);

          if (n < 0)

            error->all(FLERR,"Could not find dump modify compute ID");

          if (modify->compute[n]->peratom_flag != 0)

            error->all(FLERR,"Dump modify compute ID computes per-atom info");

          if (idim >= 0 && modify->compute[n]->vector_flag == 0)

            error->all(FLERR,"Dump modify compute ID does not compute vector");

          if (idim < 0 && modify->compute[n]->scalar_flag == 0)

            error->all(FLERR,"Dump modify compute ID does not compute scalar");



          perframe[i].type = THIS_IS_A_COMPUTE;

          perframe[i].dim = idim;

          perframe[i].index = n;

          strcpy(perframe[i].name, arg[iarg]);

        }

        else if (!strncmp(arg[iarg], "f_", 2)) {

          int idim = -1;

          char *ptr = strchr(suffix, '[');



          if (ptr) {

            if (suffix[strlen(suffix)-1] != ']')

              error->all(FLERR,"Missing ']' in dump modify command");

            *ptr = '\0';

            idim = ptr[1] - '1';

          }



          n = modify->find_fix(suffix);

          if (n < 0)

            error->all(FLERR,"Could not find dump modify fix ID");

          if (modify->fix[n]->peratom_flag != 0)

            error->all(FLERR,"Dump modify fix ID computes per-atom info");

          if (idim >= 0 && modify->fix[n]->vector_flag == 0)

            error->all(FLERR,"Dump modify fix ID does not compute vector");

          if (idim < 0 && modify->fix[n]->scalar_flag == 0)

            error->all(FLERR,"Dump modify fix ID does not compute vector");



          perframe[i].type = THIS_IS_A_FIX;

          perframe[i].dim = idim;

          perframe[i].index = n;

          strcpy(perframe[i].name, arg[iarg]);

        }

        else if (!strncmp(arg[iarg], "v_", 2)) {

          n = input->variable->find(suffix);

          if (n < 0)

            error->all(FLERR,"Could not find dump modify variable ID");

          if (!input->variable->equalstyle(n))

            error->all(FLERR,"Dump modify variable must be of style equal");



          perframe[i].type = THIS_IS_A_VARIABLE;

          perframe[i].dim = 1;

          perframe[i].index = n;

          strcpy(perframe[i].name, arg[iarg]);

          strcpy(perframe[i].id, suffix);

        }

        else {

          char errstr[1024];

          sprintf(errstr, "perframe quantity '%s' must be compute, fix or "

                  "variable", arg[iarg]);

          error->all(FLERR,errstr);

        }



        delete [] suffix;



    if (iarg >= narg)

      error->all(FLERR,"expected 'yes' or 'no' after 'thermo' keyword.");

    if (strcmp(arg[iarg],"yes") == 0) {

      thermo = true;

    }

    else if (strcmp(arg[iarg],"no") == 0) {

      thermo = false;

    }



    return narg;

    else error->all(FLERR,"expected 'yes' or 'no' after 'thermo' keyword.");

    iarg++;

    return 2;

  } else return 0;

}
@@ -1111,31 +986,4 @@ void DumpNetCDF::ncerr(int err, const char *descr, int line) 
  }

}



/* ----------------------------------------------------------------------

   one method for every keyword thermo can output

   called by compute() or evaluate_keyword()

   compute will have already been called

   set ivalue/dvalue/bivalue if value is int/double/bigint

   customize a new keyword by adding a method

------------------------------------------------------------------------- */



void DumpNetCDF::compute_step(void *r)

{

  *((bigint *) r) = update->ntimestep;

}



/* ---------------------------------------------------------------------- */



void DumpNetCDF::compute_elapsed(void *r)

{

  *((bigint *) r) = update->ntimestep - update->firststep;

}



/* ---------------------------------------------------------------------- */



void DumpNetCDF::compute_elapsed_long(void *r)

{

  *((bigint *) r) = update->ntimestep - update->beginstep;

}



#endif /* defined(LMP_HAS_NETCDF) */

src/USER-NETCDF/dump_netcdf.h

+6 −44
Original line number Diff line number Diff line 
/* ======================================================================

   LAMMPS NetCDF dump style

   https://github.com/pastewka/lammps-netcdf

   Lars Pastewka, lars.pastewka@kit.edu



   Copyright (2011-2013) Fraunhofer IWM

   Copyright (2014) Karlsruhe Institute of Technology



   This program is free software: you can redistribute it and/or modify

   it under the terms of the GNU General Public License as published by

   the Free Software Foundation, either version 2 of the License, or

   (at your option) any later version.



   This program is distributed in the hope that it will be useful,

   but WITHOUT ANY WARRANTY; without even the implied warranty of

   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

   GNU General Public License for more details.



   You should have received a copy of the GNU General Public License

   along with this program.  If not, see <http://www.gnu.org/licenses/>.

   ====================================================================== */



/* ----------------------------------------------------------------------

   LAMMPS - Large-scale Atomic/Molecular Massively Parallel Simulator

   http://lammps.sandia.gov, Sandia National Laboratories
@@ -33,6 +11,10 @@ 
   See the README file in the top-level LAMMPS directory.

------------------------------------------------------------------------- */



/* ----------------------------------------------------------------------

   Contributing author: Lars Pastewka (University of Freiburg)

------------------------------------------------------------------------- */



#if defined(LMP_HAS_NETCDF)



#ifdef DUMP_CLASS
@@ -69,22 +51,6 @@ class DumpNetCDF : public DumpCustom { 
    int ndumped;                  // number of enties written for this prop.

  };



  typedef void (DumpNetCDF::*funcptr_t)(void *);



  // per-frame quantities (variables, fixes or computes)

  struct nc_perframe_t {

    char name[NC_FIELD_NAME_MAX]; // field name

    int var;                      // NetCDF variable

    int type;                     // variable, fix, compute or callback

    int index;                    // index in fix/compute list

    funcptr_t compute;            // compute function

    int dim;                      // dimension

    char id[NC_FIELD_NAME_MAX];   // variable id



    bigint bigint_data;           // actual data

    double double_data;           // actual data

  };



  int framei;                  // current frame index

  int blocki;                  // current block index

  int ndata;                   // number of data blocks to expect
@@ -94,10 +60,10 @@ class DumpNetCDF : public DumpCustom { 
  int n_perat;                 // # of netcdf per-atom properties

  nc_perat_t *perat;           // per-atom properties



  int n_perframe;              // # of global netcdf (not per-atom) fix props

  nc_perframe_t *perframe;     // global properties

  int *thermovar;              // NetCDF variables for thermo output



  bool double_precision;       // write everything as double precision

  bool thermo;                 // write thermo output to netcdf file



  bigint n_buffer;             // size of buffer

  int *int_buffer;             // buffer for passing data to netcdf
@@ -131,10 +97,6 @@ class DumpNetCDF : public DumpCustom { 
  virtual int modify_param(int, char **);



  void ncerr(int, const char *, int);



  void compute_step(void *);

  void compute_elapsed(void *);

  void compute_elapsed_long(void *);

};



}

CRA Git | Maintained and supported by SUSTech CRA and CCSE