This page is a continuous listing of new features and bug fixes for the SPARTA simulator.
In some cases, new features change the syntax or operation of an existing input script command, which can cause SPARTA output to change, or even "break" an earlier working script. These cases are highlighted below with a "BACKWARD COMPATIBILITY" note.
To stay up-to-date with the current SPARTA, you can download the current tarball from the download page at any time, and re-build the code. It will include all the features/fixes listed below.
A SPARTA "version" is the date when it was released, such as 10 Aug 2014. SPARTA is updated continuously. Every time we fix a bug or add a feature, we release it immediately, as listed below. Each dated copy of SPARTA contains all the features and bug-fixes up to and including that version date. The version date is printed to the screen and logfile every time you run SPARTA. It is also in the file src/version.h and in the SPARTA directory name created when you unpack a tarball.
4 Apr 2018
Stan Moore (Sandia) added an option to use per-processor CPU time as a balancing metric to the balance_grid and fix_balance commands.
He also enabled the compute boundary and surf_collide commands to work with Kokkos.
This version also includes a new version of the Kokkos library included with SPARTA.
This is the list of changed files versus the 23Dec17 version.
23 Dec 2017
One more time ... must have drunk too much Christmas eggnog.
This is the list of changed files versus the 22Dec17 version.
22 Dec 2017
We forgot to include the Kokkos library itself in the SPARTA disribution in the the 21 Dec 2017 version! This version has it. See lib/README and lib/kokkos once you unpack the tarball.
This is the list of changed files versus the 21Dec17 version.
21 Dec 2017
Stan Moore, Dan Ibanez, and Tim Fuller (Sandia) have done heroic work to enable SPARTA to work with the open-source Kokkos library. This is implemented as a new KOKKOS package in the code. For an application code like SPARTA, Kokkos is designed to enable one set of source-code modifications to run reasonably efficient on a variety of hardware, e.g. multi-core CPU, GPU, and KNL nodes. See these sections of the manual for more details:
Note that only some styles and options within the code are Kokkos-enabled at this point. This includes the basic move and collide operations, including for geometries with hierarchical grids and embedded surfaces. This Section 3.5 has (k) flags listed for fixes and computes that have been adapated for Kokkos.
We also plan to soon post some benchmark numbers on the SPARTA web page to indicate what kind of performance is possible on different node architectures (CPU, GPU, KNL) via Kokkos.
The timing output produced at the end of a run was enhanced to include info about fixes that operate during the timestepping. Depending on the input script, some of these can invoke expensive operations including compute commmands. This category of operation is now listed as a "Modify" timing in the breakdown of the timestep.
This is the list of changed files versus the 27Jul17 version.
27 Jul 2017
Eric Emdee (PPPL) called attention to an incorrect use of uniform versus Gaussian random numbers in the particle emission models from the simulation box face or surfaces. This makes a small (~3%) difference in the distribution of emitted particle velocities. The following commands are affected:
This is the list of changed files versus the 4 May 2017 version.
4 May 2017
Fixed a bug with the new surf_collide piston command which was not tallying properties correctly on the simulation box boundary when particles were deleted. Also added a couple new keywords to the compute boundary command and made it's tallying similar to the compute surf command, by using weighted particle counts for many of the keywords.
Added grid groups via the group grid command, similar to surface element groups. This allows grid cells to be assigned to one or more groups. The groups are now used as required or optional arguments to several compute and fix commands as well as other commands in this list, to optionally limit which grid cells computations are performed for:
A new grid group-ID argument is required for the computes in the list, is now implemented by the dump grid command (was already an argument), and is required by the fix ave/grid, fix adapt/grid, and adapt_grid commands. A group-ID setting is a new optional keyword for the fix ave/histo and fix ave/histo/weight commands for grid cell selection in the histogramming, as are new optional keywords region and mixture for particle selection in the histogramming.
BACKWARD COMPATIBILITY: This means the syntax for all of the commands in the above list have changed (except dump grid and the fix ave/histo commands) because they now require an additional grid group-ID argument. The format of restart files also changes with this release, because the grid group and surface element group information is now stored differently.
The grid cell group assignments persist when doing grid adaptation, via the adapt_grid or fix adapt/grid commands, which means new refined or coarsened grid cells inherit the same group assignment from their predecessor cells.
Fixed a bug when using adaptive gridding with cell weighting via the global weight command, where the weights were not reset correctly for adapated grid cells.
Added a temperature option to the create_particles command to allow creation of particles with a thermal temperature gradient.
This is the list of changed files versus the 4 Apr 2017 version.
4 Apr 2017
Added a new surf_collide piston which can be used on a simulation box boundary or a set of surface elements that act like a boundary, as a subsonic pressure boundary condition.
Added a new compute fft/grid command to calculated 2d or 3d FFTs of per-grid quantities, so long as the simulation grid is regular (not hierarchical). This can be used for computing energy spectra in turbulent flows.
Added nwt and mflux keywords to the compute surf command, for tallying of weighted particle collisions with surface elements, and also the mass flux onto each surface element.
This is the list of changed files versus the 16 Mar 2017 version.
16 Mar 2017
Fixed two bugs. The first was with grid adaptation sometimes creating a new grid whose grid cells and corner points could not be succesfully marked as inside versus outside a surface.
The second was with surface reactions sometimes creating a new particle that was not in the correct grid cell, which could lead to the particle moving inside the surface, or other errors.
This is the list of changed files versus the 10 Mar 2017 version.
10 Mar 2017
Fixed a bug with the fix ave/surf command when used with a single input value.
Thanks to Dan Barnes (Coronado Consulting) for flagging this issue.
This is the list of changed files versus the 8 Sep 2016 version.
8 Sep 2016
Fixed a couple of logic bugs for multi-group recombination reactions.
This is the list of changed files versus the 23 Aug 2016 version.
23 Aug 2016
Fixed a couple of bookkeeping bugs with recombination reactions, one for single group collisions, one for multi-group collisions.
This is the list of changed files versus the 8 Aug 2016 version.
8 Aug 2016
Enhanced the syntax for many commands to allow multiple selected values from a compute or fix vector, or multiple columns from a compute or fix array, to be specified with a single argument, using a wildcard syntax. E.g. specifiying c_ID[*] as a single argument gets expanded into multiple arguments c_ID, c_ID, ..., using these wildcard rules:
where N is the length of the vector or number of columns in the array.
These commands now have the enhanced syntax:
Also removed a restriction on the header line printed before stats output during a run. The per-column fields were limited to 8 characters each which could mean long IDs for computes, fixes, or variable names were truncated. Now the full column name, as specified in the stats_style command, is output. This includes the "c_", "f_", or "v_" prefix for compute, fix, variable quantities, so there will be no ambiguity if a compute and fix have the same ID.
Added options to the stats_modify format and dump_modify format commands, and made their syntax consistent with each other.
Each now takes a line option to specify the format of an entire line of output, int and float options to specify the format of each integer or floating point value output, an M string option to change the format of a single value in column M of the output, or a none option to restore the default format for all output values.
BACKWARD COMPATIBILITY: Several of the commands listed above previously allowed an entire array of input values (from a compute or fix) to be specified using just the ID of the compute or fix. You now have to use the wildcard syntax above to have the same effect.
BACKWARD COMPATIBILITY: This patch can change the format of one line in the screen or logfile output, namely the header line that describes the columns of stats output. If your output includes quantities from a compute or fix or variable, then that line will have a slightly different syntax, as described above. If you have a post-processing parser that reads that line or uses it to identify columns of data, you may to need to adjust your parse semantics.
BACKWARD COMPATIBILITY: The dump_modify format command previously only allowed the format for an entire line to be specified. Using that option now requires the "line" setting be used after "format", to distinguish it from the other options that were added.
This is the list of changed files versus the 5 Aug 2016 version.
5 Aug 2016
Added fix ave/histo and fix ave/histo/weight commands, to enable histogramming of global, per-particle, and per-grid cell quantities. The histograms can also be time averaged if desired.
Also added an internal-style variable to the "variable" command for use as input to other SPARTA commands which use a variable they can reset. The create_particles command is an example. This is more efficient way to enable variable values to be reset internally by the code.
BACKWARD COMPATIBILITY: The create_particles command now requires an internal-style variable, not an equal-style variable as input for some its species, density, and velocity options.
This is the list of changed files versus the 11 Jul 2016 version.
11 Jul 2016
Fixed a normalization error with the compute thermal/grid press command which was not including a cell weighting factor. This typically gave incorrect answers for axisymmetric problems where radially dependent cell weighting is often enabled.
This is the list of changed files versus the 6 Jun 2016 version.
Thanks to Dmitrii Ivchencko (University of Limoges, France) for pointing out the problem.
6 Jun 2016
Tried to make the cutting and splitting of grid cells by surfaces a more robust operation and give users less to worry about.
This is the list of changed files from the 2 Jun 2016 version.
2 Jun 2016
Disallow Tsurf = 0.0 for the surf_collide diffuse model. This can cause problems with particles effectively sticking to surfaces w/out bouncing off.
This is the list of changed files from the 1 Jun 2016 version.
1 Jun 2016
Fixed a bug with the read_restart command not resetting the global surfmax setting correctly, if it had been set to a large value in the input script that wrote the restart file.
This is the list of changed files from the 31 May 2016 version.
31 May 2016
Added fx, fy, fz options to the compute surf command to allow drag and lift on a body in a flow to be more easily computed.
This is the list of changed files from the 24 May 2016 version.
24 May 2016
Fixed a bug with the write grid parent command, when dumping grids with 64-bit IDs (and parent cells that actually exceeded 32 bits).
This is the list of changed files from the 10 May 2016 version.
10 May 2016
Added 2 new compute commands, compute eflux/grid and compute pflux/grid, for tallying the per-grid cell energy and momentum flux densities. These were implemented with help from Alejandro Garcia (San Jose State University).
Also fixed a bug in the read_restart command when reading very large restart files (with > 2B grid cells), when the code is compiled with the -DSPARTA_BIGBIG option.
This is the list of changed files from the 31 Mar 2016 version.
31 Mar 2016
Fixed a bug with how the compute surf command was interacting with the fix ave/surf and dump surf commands. It had to do with when the normalization factors were applied to surface tallying, which caused problems if the results of the compute were used multiple times (on the same timestep) by different commands.
Thanks to Flavia Bonelli (Karlsruhe, Germany) for sending a script that triggered the issue.
This is the list of changed files from the 21 Mar 2016 version.
21 Mar 2016
Alan Stagg (ORNL) flagged an occasional bug with the species command reading species files with long lists of species.
This is the list of changed files from the 14 Mar 2016 version.
14 Mar 2016
Fixed a typo bug with the group surf union and intersect options.
Thanks to Alan Stagg (ORNL) for pointing out the problem.
This is the list of changed files from the 13 Mar 2016 version.
13 Mar 2016
Fixed some bugs in grid adaptation, surface groups, and the compute distsurf command.
Thanks to Ram Prakash (UNSW-ADFA, Canberra) for sending a script that triggered these problems.
This is the list of changed files from the 12 Mar 2016 version.
12 Mar 2016
Changed the formula used for the compute grid massrho value to be for all particles in the grid cell (i.e. total mass), not the per-particle average.
BACKWARD COMPATIBILITY: This will change the value output by this compute, or fix ave/grid or dump grid, for the massrho value.
Thanks to Israel Sebastiao (Purdue) for pointing this out.
This is the list of changed files from the 11 Mar 2016 version.
11 Mar 2016
A couple of files were mistakenly not checked into the repository for the 9/10 Mar versions.
This is the list of changed files from the 10 Mar 2016 version.
10 Mar 2016
Added a velocity option to the create_particles command. It allows the streaming velocity of created particles to be set by spatially-dependent variables which depend on the position of the created particle. The formulas for both are defined by equal-style variables. Example input script commands and a small animation of a movie with an initial velocity distribution that causes particles to collapse to a point is included in the create_particles command doc page.
This is the list of changed files from the 9 Mar 2016 version.
9 Mar 2016
Fixed a bug in the group surf command with how it was using surface IDs vs surface types.
Thanks to Ram Prakash (UNSW-ADFA, Canberra) for sending a script that triggered this problem.
This is the list of changed files from the 8 Mar 2016 version.
8 Mar 2016
Added two options to the create_particles command, with keywords species and density. The first allows the species of created particles to be set by a spatially-dependent variable which depends on the position of the created particle. The second allows the density of particles created in each grid cell to be set by a spatially-dependent variable which depends on the center point of the grid cell. The formulas for both are defined by equal-style variables. Example input script commands and snapshots of the initial particle distributions are included in the create_particles command doc page.
This is the list of changed files from the 29 Feb 2016 version.
29 Feb 2016
Fixed some bookkeeping issues for grid cells volumes when cells are adjacent to a surface but wholly inside or wholly outside, i.e. the surface elements just touch the face, edge, or corner point of a grid cells.
This is the list of changed files from the 22 Feb 2016 version.
22 Feb 2016
Implemented a near-neighbor algorithm for choosing collision partners. The new option is enabled by the collide_modify nearcp command. This is a means of choosing a collision partner J for a particle I that is geometrically closer (within its grid cell) than the standard algorithm which simply chooses any random particle J within the grid cell. It is more expensive than the standard algorithm but can enable the use of larger grid cells due to the sub-grid-cell resolution if offers.
This is the list of changed files from the 17 Feb 2016 version.
17 Feb 2016
Fixed a logic bug with the fix emit/surf command when running a problem geometry where surfaces split grid cells into sub-cells.
Thanks to Alan Stagg (ORNL) for sending a script that triggered this problem.
This is the list of changed files from the 16 Feb 2016 version.
16 Feb 2016
Changed the inner workings of the global surfpush yes command, which is used when there are round-off issues when mapping surface elements to grid cells which in rare cases can generate an error. The command now allows setting of 3 values (slo,shi,svalue) which determine the range of surface point positions that are altered, and the new value they are set to. Using the yes option sets these 3 settings to generically good values which seem to work for most of the round-off issues we have seen.
Added some error checking and warning messages to the various fix emit command variants when using the subsonic keyword to invoke subsonic boundary conditions. There are unusual cases (empty grid cells, small temperatures) which can lead to extreme flow conditions that we try to protect against more carefully.
This is the list of changed files from the 5 Jan 2016 version.
5 Jan 2016
Added grid-style variables to SPARTA as explained on the variable doc page. Similar to equal-style and particle-style variables, grid-style variables define a formula that is calculated for each grid cell in the simulation. The formulas can contain a variety of scalar quantities and math operations as well as per-grid quantities calculated by a per-grid compute (e.g. the compute grid command), a per-grid fix (e.g. the fix ave/grid command), or another grid-style variable. Grid-style variables can be used as input to the dump grid, compute reduce, and fix ave/grid commands.
This is the list of changed files from the 8 Dec 2015 version.
8 Dec 2015
Added a pressure option to the VALUES listing in the input data file read by the fix emit/face/file command. This allow a subsonic boundary condition to be set on a face of the simulation box where the pressure and temperature or just a pressure are spatially varying. This allows for consistent boundary conditions when the flow is subsonic.
Also fixed some addtitional issues with using the subsonic boundary condition with all 3 of the fix emit commands, when starting with an empty simulation box, i.e. no particles initially exist in the cells where particles are to be inserted.
This is the list of changed files from the 25 Nov 2015 version.
25 Nov 2015
Added a subsonic option to the fix emit/surf command to allow a subsonic boundary condition to be set for surfaces in the simulation box by specifying a pressure and temperature or just a pressure. This allows for consistent boundary conditions when the flow is subsonic.
Also fixed some issues we discovered with the recently released fix emit/face command.
This is the list of changed files from the 6 Nov 2015 version.
6 Nov 2015
Added a subsonic option to the fix emit/face command to allow a subsonic boundary condition to be set for one or more faces of the simulation box by specifying a pressure and temperature or just a pressure. This allows for consistent boundary conditions when the flow is subsonic. At some point, we plan to add a similar capability to the fix emit/face/file and fix emit/surf commands.
Added temp and press values to the compute thermal/grid command, so that either or both a temperature and pressure can be calculated for each grid cell based on the thermal (non-streaming) kinetic energy of particles in the cell.
BACKWARD COMPATIBILITY: Previously, the "compute thermal/grid"-doc/compute_thermal_grid.html command did not take any values, because it only computed a thermal temperature. It now requires at least one value to be specified.
This is the list of changed files from the 3 Nov 2015 version.
3 Nov 2015
Added pxrho, pyrho, pzrho, and kerho values to the compute grid command to tally momentum and kinetic energy density on a per-grid cell basis.
This is the list of changed files from the 23 Oct 2015 version.
23 Oct 2015
Added two mixture keywords for rotational and vibrational temperature, in addition to thermal temperature, which can be used to set the initial rotational and vibrational energy of created particles, e.g. via the create_particles command or the various fixes that emit particles from simulation box faces and surfaces.
Added a nearcp keyword to the collide_modify command to enable selection of collision partners via a near neighbor algorithm, as opposed to randomly selecting collision partners from all the particles in a grid cell. The details of the algorithm and a paper citation are given on the collide_modify doc page.
BACKWARD COMPATIBILITY: The internal syntax of restart files changed to add the two new temperature options to mixtures. This means the current version of the code cannot read old restart files and vice versa.
This is the list of changed files from the 4 Oct 2015 version.
4 Oct 2015
Added a read_particles command which reads a snapshot of particles into the current simulation from a file, in the same format the dump particle command creates. This allows a simulation to start with a previous simulation's set of particles, or to use particles output from another simulation code as input to SPARTA, assuming they have been massaged into the needed format.
This is the list of changed files from the 3 Oct 2015 version.
3 Oct 2015
Fixed a couple issues with how restart files are used to re-initialize a system. It should now be easier to read a restart file early in an input script and not have to re-define other simulation properties at different points of the restart script as compared to the original input script.
Also, if the restart file is read on a different number of processors, it may be necessary to re-balance the partitioning of grid cells to processors so that the simulation can be setup with needed ghost cells. There are new additional gridcut and balance options on the read_restart command to enable this.
This is the list of changed files from the 2 Oct 2015 version.
2 Oct 2015
Added nfrac and massfrac keywords to the compute grid command. Added nparent, nchild, and nsplit keywords to the stats_style command.
Added a compute thermal/grid command that does time averaging for thermal temperature calculation over all timesteps, when it is used as input to the fix ave/grid command. The thermal keyword of the compute sonine/grid command has been removed, since the new command replaces its functionality.
Previously, computing a thermal temperature via the compute sonine/grid thermal command used center-of-mass (COM) velocities that were only calculated for the current timestep. In the new command, when doing time averaging via the fix ave/grid command, the COM velocity is calculated for all particles across multiple timesteps. We think this is a more accurate and less noisy way to do it.
This is the list of changed files from the 22 Sep 2015 version.
22 Sep 2015
Added a massrho keyword to the compute grid command.
Added a vanish style model to the surf_collide commmand, which simply deletes particles when they hit surface elements assigned to that model. This is a way to have a group of surface elements act as an outflow boundary for particles, as well as an inflow boundary via the fix emit/surf command.
Surface collision and surface reaction models also now tally the number of particles that hit and react on the surface elements they are assigned to. Both for the current timestep and cummulative counts during a run. These values can be accessed by the stats_style command for output with statistics, and by variables that define formulas. The latter means they can be used by any command that uses a variable as input, e.g. "the fix ave/time command.
This is the list of changed files from the 12 Sep 2015 version.
12 Sep 2015
Fixed some compiler warnings that were showing up on a Mac compile.
This is the list of changed files from the 11 Sep 2015 version.
11 Sep 2015
Edited/added some src/MAKE/Makefile.machine files to make building SPARTA more straightforward. Also changed some of the example input scripts.
This is the list of changed files from the 21 Aug 2015 version.
21 Aug 2015
Options for adapting the hierarchical grid used by SPARTA have been added. This can either be done before or between run commands via the adapt_grid. Or adaptation can be performed on-the-fly during a run, via the fix adapt command. Adapation can involve refinement and/or coarsening of grid cells, based on a variety of criteria, including values calculated by computes or fixes. The compute distsurf/grid command was added to calculate one such criterion: distance of a grid cell from any surface element.
Commands to modify surfaces have also been added. The move_surf and fix move/surf commands allow a group of surface elements to be moved in various manners either between runs or during a run. The write_surf command write surface elements to a file, which can be useful after clipping has taken place, surface elements have been moved or removed. The remove_surf command can remove a subset of surface elements from the simulation.
The list of new commands is here:
See new example problems that exercise these commmands in the examples/adapt and surf directories, and on the Pictures/Movies page of the web site (soon).
The ParaView tools have also been upgraded to work with grid files that adapt over time, for visualization using ParaView.
This is the list of changed files from the 17 Jun 2015 version.
17 Jun 2015
One more forgotten file!
This is the list of changed files from the 16 Jun 2015 version.
16 Jun 2015
Forgot a new file needed in yesterday's update.
This is the list of changed files from the 15 Jun 2015 version.
15 Jun 2015
Fixed a bug when one of the fix emit face commands emitted particles on multiple faces of the simulation box and the fix balance command was also enabled.
This is the list of changed files from the 15 May 2015 version.
15 May 2015
Small bug in yesterday's post. This should fix it.
This is the list of changed files from the 14 May 2015 version.
14 May 2015
Added a region keyword to the create_grid command which be used in place of the level keyword at any level of a hierarchical grid definition. This allows the grid to be refined based on whether a grid cell is inside a geometric region defined by the region command. Regions can be simple geometric objects or unions or intersections of simple objects or other unions or intersections. This is thus a more flexible way of defining a hierarchical grid.
This is the list of changed files from the 13 May 2015 version.
13 May 2015
Fixed a bug with surface clipping in the read_surf command, when multiple surfaces are read in, and the later ones are clipped.
This is the list of changed files from the 12 May 2015 version.
12 May 2015
Several new features are added to the code in this version.
The ambipolar approximation for plasmas (charged particles) is now supported. See Section howto 4.11 for details of how the ambipolar approximation is implemented in SPARTA and how to use it from an input script. Relevant commands inlcude:
Gas reaction models for ionization were added to the dissociation and exchange options, all of which can now be specified via the react command. These can now include ambipolar reactions involvoing ambipolar ions and their electrons.
Surface reaction models for dissociation, exchange, and recombination can now be specified via the surf_react command. These can include ambipolar reactions, e.g. recombination of ions and their electrons.
Surface groups were added so that properties like collision and reaction models can be assigned to subsets of surface elements. This is implemented via a new group commmand. Several commands now take surface group IDs as arguments, changing their syntax. The following commands are affected.
Also the bound_modify command syntax is changed, since surface reaction models can be assigned to simulation box boundaries as well.
The fix inflow and fix inflow/file commands were renamed to fix emit/face and fix emit/face/file.
A new fix emit/surf command was added to enable outflux of particles from surface elements.
A region option was added to the create_particles, fix emit/face, and fix emit/face/file commands. This allows creation of particles within limited geometric regions of the simualation domain.
All the dump commmands now take a "group" argument, changing their syntax, to select which particles, grid cells, or surface elements are output. I.e. the dump particle and dump image commands take a mixture-ID as an argument, the dump grid command takes a grid group ID as an argument (only "all" is currently an option), and the dump surface command takes a surface group ID as an argument. See the dump command doc page for details.
A bug was fixed for the fix emit/face/file cmomand when using it for 2d axisymmetric models. The appropriate axisummetric volume weightings were not being applied to grid cells to adjust their number of created particles.
BACKWARD COMPATIBILITY: The internal syntax of restart files changed to add group information. This means the current version of the code cannot read old restart files and vice versa. The syntax of these commands changed: read_surf, compute surf, fix ave/surf, surf_modify, bound_modify. Likewise the dump particle, dump grid, and dump surface command syntax changed to take an extra argument. The fix inflow and fix inflow/face commands were renamed to fix emit/face and fix emit/face/file.
This is the list of changed files from the 13 Mar 2015 version.
13 Mar 2015
Added some visualization examples using ParaView and the paraview tools provided with SPARTA to convert SPARTA output to ParaView input. The example pics are on the Pictures & Movies page of the web site. The surface data and STL files used for several of these space craft were added to the data directory of the distribution.
This is the list of changed files from the 6 Mar 2015 version.
6 Mar 2015
Tom Otahal (Sandia) has enhanced his Python scripts which can convert SPARTA grid data (input and output) to ParaView format. It can now read a SPARTA parent grid which defines a hierarchical grid with multiple levels of refinement. It also now allows for specification of 2d slice planes, each of which will output ParaView files with grid values on the 2d plane.
This is the list of changed files from the 5 Mar 2015 version.
5 Mar 2015
Fixed a memory issue when performing reactions on a problem where collisions are performed for multiple species groups within each grid cell.
Also fixed a logic bug with clipping multiple sucessive surfaces, read in by the read_surf command.
This is the list of changed files from the 4 Mar 2015 version.
Thanks to Arnaud Borner (NASA Ames) and Aggelos Klothakis for sending test problem that triggered the issues.
4 Mar 2015
Fixed a small issue with a couple of Makefiles that were causing issues on Macs.
This is the list of changed files from the 26 Feb 2015 version.
Thanks to David Wolfe (NPS) for calling attention to these issues.
26 Feb 2015
Added a fix inflow/file command which allows input of particles on a simulation box face, with properties of the input specified in a file. The file contains a mesh of points that overlay part or all of the box face. Quantities can be defined at the mesh points that affect the created particles, e.g. a streaming velocity, thermal temperature, number fractions of species in the mixture, overall density, etc. The quantities can vary spatially on the mesh, so that the input flux of particles is also spatially varying.
There is an examples/flowfile directory with an example input script, as well as a movie on the SPARTA movies web page that illustrates use of the command.
This is the list of changed files from the 21 Feb 2015 version.
21 Feb 2015
One more change that was mistakenly introduced by the 17Feb15 patch.
This is the list of changed files from the 17 Feb 2015 version.
17 Feb 2015
Fixed bug in restart files not including the global weight cell command setting if it was enabled, which was added for axisymmetric geometries.
This is the list of changed files from the 16 Feb 2015 version.
Thanks to Angelos Klothakis for sending a script that triggered this issue.
16 Feb 2015
Two small bug fixes in the VSS collision model. One could generate unrealistic temperatures as the outcome of certain reactions. The second was a bookkeeping error for the code vs the format of the input data file of reactions, with regard to dissociation reactions producing only monoatomic products (molecule + atom -> 3 atoms).
Also some changes to a few coefficients in the input data file for TCE reactions in air.
This is the list of changed files from the 21 Jan 2015 version.
Thanks to Arnaud Borner (NASA Ames) for calling attention to these issues and flagging the specific TCE reactions.
21 Jan 2015
Fixed an rare bug with particle moves hitting exactly on the edge or corner point of a surface element that is in common with multiple surfaces.
Also found a case where the global surfpush yes command needs to move surface points that are slightly outside a grid cell to be on the surface, not just from inside to the surface (just for the intersection operation). So the surfpush option now does both.
This is the list of changed files from the 13 Jan 2015 version.
13 Jan 2015
Fixed an occasional bug with axisymmetic moves that could cause a particle to end up in the wrong grid cell when being communicated to a new processor. This is with the global gridcut command set to a positive finite value.
This is the list of changed files from the 7 Jan 2015 version.
Thanks to Angelos Klothakis for sending a test geometry that had the problem.
7 Jan 2015
Added a relax variable option to the collide command to allow for per-collision computation of rotational and vibrational relaxation, instead of just the constant parameters used previously. The additional 4 parameters per species needed for the rotational and vibrational relaxation equations can be specified in the VSS or VHS collision data file. See data/vss.air for examples.
This is the list of changed files from the 19 Dec 2014 version.
19 Dec 2014
Added two new reaction styles, acessible via the react command. A Quantum-Kinetic (QK) model due to Bird, and a hybrid Total Collision Energy / Quantum Kinetic (TCE/QK) model (both parts due to Bird). The same input file of reactions and reaction coefficients can be used for either. To use the hybrid style, flags in the reaction file can be set to "A" or "Q" for each reaction to specify whether it is performed using the TCE (Arrhenius) or QK model.
This is the list of changed files from the 10 Dec 2014 version.
10 Dec 2014
Fixed a bug with reading restart files via the read_restart command when a hierarchical multi-level grid is defined.
Also added some logic to support STL unordered maps used by the Clang++ compiler, via the compile-time switch -DSPARTA_UNORDERED_MAP, as discussed in Section 2.2 of the manual.
This is the list of changed files from the 19 Nov 2014 version.
Thanks to Arnaud Borner (NASA Ames) for sending an input script that triggered the problem.
19 Nov 2014
Some testing and debug lines were left in the recent update of the fix inflow command, which cause incorrect input fluxes. This patch removes the lines.
This is the list of changed files from the 15 Nov 2014 version.
15 Nov 2014
The tools/grid_refine.py tool had some incorrect logic for refining 3d grids.
Also added an axisymmetric example problem in examples/axi.
This is the list of changed files from the 13 Nov 2014 version.
Thanks to Baofeng Ma for pointing out the problem with the grid_refine tool.
13 Nov 2014
Fixed another bug with particle moves and surface bounces in axisymmetric models to allow for the rare case when a particle bounces multiple times off the same surface in a single timestep, due to its curvature.
Also, three bug fixes for the fix inflow and surf_collide diffuse commands.
This is the list of changed files from the 3 Nov 2014 version.
3 Nov 2014
Fixed another bug with particle moves and surface bounces in axisymmetric models due to how to interpret the solutions to a quadratic equation.
This is the list of changed files from the 31 Oct 2014 version.
31 Oct 2014
Fixed a numerical round-off issue with the intersection of surface elements and grid cells. This could cause an error when the read_surf command was used if the surface had points that were "nearly" on grid cell boundaries. The work-around is to shift any such points from slightly inside a grid cell to the cell boundary (just for the intersection operation). This operation can now be invoked by the global surfpush yes command. At some point we will likely automatically invoke it if the error occurs consistently.
This is the list of changed files from the 28 Oct 2014 version.
Thanks to Tim Deschenes for sending a test case with a simple geometry that triggered the problem.
28 Oct 2014
Roman Maltsev fixed some energy conservation bugs in the VSS collision model for the collide vss command.
This is the list of changed files from the 27 Oct 2014 version.
27 Oct 2014
Particle tracking across cell boundaries and for particle/surface collisions was too simplistic for some models, leading to lost particles or incorrect surface bounces. This version should fix the problems.
This is the list of changed files from the 25 Oct 2014 version.
25 Oct 2014
Added an option to the read_surf clip command to allow surface points very close to the simulation box boundaries to be pushed to the boundary before a clipping operation is performed. This can help prevent very small line or triangles from being created due to the clip.
This is the list of changed files from the 24 Oct 2014 version.
24 Oct 2014
The surf_collide diffuse translate and rotate options were missing a check to subtract out any component of surface velocity that was normal to the surface, so that only the proper post-bounce tangential velocity was added to colliding particles.
This is the list of changed files from the 20 Oct 2014 version.
20 Oct 2014
The tools/stl2surf.py file was inadvertantly left out of the 12Aug14 patch.
Tom Otahal (Sandia) has written 2 Python scripts which can convert SPARTA grid and surface data (input and output) to ParaView format. ParaView is a popular, powerful, freely-available visualization package. You must have ParaView installed to use the Python scripts. See tools/paraview/README for more details.
Fixed a memory bug with the dump grid command that could cause problems when there are grid cells split by surfaces into two or more sub-cells.
Fixed a logic error when using a global gridcut value > 0.0 that showed up in flow regimes with long timesteps where particle moves across processor boundaries were not correctly completed.
This is the list of changed files from the 9 Oct 2014 version.
9 Oct 2014
Added a tvib option to the compute grid command to enable vibrational temperature to be calculated within the classical approximation.
The tallying of surface properties that depend on surface element area was incorrect for axisymmetric models.
Added an error check to insure grid cell neighbors and ghost cells have been created by the time any surfaces are read in via the read_surf command.
NOTE: We realized particle/surface collisions in axisymmetric models are missing some axisymmetric corrections that can cause problems for big timesteps and collisions near the axisymmetry axis (y = 0). An upcoming patch should include some updated code for this.
This is the list of changed files from the 19 Aug 2014 version.
19 Aug 2014
Added a grid_refine.py tool to the tools directory to create a grid file, readable by the read_grid command, which has been adapted to a surface file, readable by the read_surf command.
The tool has several options for how the adaptation is done. At this point, the adaptivity is strictly based on geometric considerations. At some point we will add flow characteristics to the options, so the new computational grid can reflect attributes of a previously computed flow field.
This is the list of changed files from the 15 Aug 2014 version.
15 Aug 2014
One change for applying cell weights to collisions was missing from the 11Aug14 patch for enabling axi-symmetry.
This is the list of changed files from the 14 Aug 2014 version.
14 Aug 2014
New compute lambda/grid command, which can calcluate the mean free path and Knudsen number for each grid cell. This can be dumped to a file via the dump command and used to adapt the computational grid, e.g. between two runs.
Soon we will also release an external script in the tools directory to perform this adpativity operation.
This is the list of changed files from the 13 Aug 2014 version.
13 Aug 2014
Added a scale_particles command to allow the total number of particles in the simulation to be scaled up or down via particle cloning or deletion. This is useful for changing the resolution of a model between runs.
This is the list of changed files from the 12 Aug 2014 version.
12 Aug 2014
Added a stl2surf.py tool to the tools directory to convert stereolithography (STL) files to SPARTA-format surface files, which can be read by the read_surf command.
The STL files must be in ASCII (text) format, which can be produced by various meshing programs. Note that surface object(s) in the STL file must be "watertight" in order to be used by SPARTA (unless they are being clipped to the simulation box), as described on the read_surf doc page.
This is the list of changed files from the 11 Aug 2014 version.
11 Aug 2014
Added an axi-symmetry capability to allow 2d simulations with a y=0 axis of symmetry to be run. See the boundary and global weight commands for details, and also Section 4.2 for an overview discussion.
This is the list of changed files from the 29 Jul 2014 version.
29 Jul 2014
Typo bug in the create_grid command that affected how it processed some input script options.
This is the list of changed files from the 23 Jul 2014 version.
23 Jul 2014
Removed a couple auxiliary files that slipped into tarball distribution accidentally.
This is the list of changed files from the 21 Jul 2014 version.
21 Jul 2014
Added some output of chemical reaction statistics to the screen and log file.
This is the list of changed files from the 7 Jul 2014 version.
7 Jul 2014
Initial public release of SPARTA. This version supercedes earlier distributed beta versions.