[Espressomd-maintainer] [espresso] Mine and Dominics work on ESPResSo of the last month (#11)

[Georg Rempfer]

New features:

- TCL command to print the whole LB velocity and boundary field in one go 
(lbfluid print ...)
- Visualization of constraints in VMD via an additional parameter in 
prepare_vmd_connection.
- New constraint type "rhomboid".
- New lbboundary type "rhomboid".

In that context, the interface for the CPU and GPU LB has been completely 
unified on the TCL level as well as on the C level. This makes usage, 
maintenance and extension of the LB module a lot easier in the future.

Also everything is documented.

You can merge this Pull Request by running:

  git pull https://github.com/rempferg/espresso lbgpu

Or you can view, comment on it, or merge it online at:

  https://github.com/espressomd/espresso/pull/11

-- Commit Summary --

* Started implementing the lbnode set boundary command for the CPU LB.
* Finished implementing the lbnode set boundary command for the CPU LB. Also 
changed the naming of all functions on the C level involving LB boundaries from 
*border* to *boundary*. Unfortunately integrate segfaults now - see 
ttt/lbnode_set_boundary.tcl.
* Integrate failed because some lbboundary initialisation functions were not 
executed prior to usage of 'lbnode set boundary'. Instead of fixing this, 
removed the whole implementation and started implementing a boundary type 
'rhomboid'. This is more general and convenient. Begun by implementing 
'constraint rhomboid'.
* Missing for 'constraint rhomboid' are only the proper initialization of the 
struct and the distance function.
* ka
* Finished implementation of everything needed for the 'constraint rhomboid' 
except for the distance function.
* Added rhomboid constraint to force and energy calculations and excluded it in 
the polymer functions like most constraints. Also implemented a dummy distance 
function which represents a sphere of radius 10 at (20, 20, 20) for testing 
purposes. TCL test script shows that everything works nicely.
* Completed implementation of 'lbboundary rhomboid'. Also extensively tested 
constraint and lbboundary in combination. Everything works. Still missing: 
distance function.
* 1/3 of distance function for 'constraint rhomboid' implemented. It detects if 
the given position lies within one of the cones defined by a corner and the 
adjacent edges. It calculates the distance to the respective corner as a vector 
and scalar correctly. Missing: distance to an edge if the position lies within 
one of the prisms attached to the edged and distance to the faces.
* 2/3 of distance function for 'constraint rhomboid' implemented. It now also 
detects if a particle is located within one of the prisms adjacent to the 
respective edge of the rhomboid.
* Finished implementing distance function but face distance calculation is not 
working properly.
* Fixed face distance calculation. Missing is distance calculation for the 
volume inside the rhomboid.
* Implementation complete. Resulting shape seems okay as an lbboundary but this 
only guarantees that the sign of the scalar distance is correct. Whether the 
vector valued distance is correnct has to be tested with particles before 
merging into the master.
* Wrote test script for particle interaction. Seems fine. Consulted Axel about 
the implementation - he considers it okay. Will conduct one more test with a 
rhomboid rotated relative to the grid axes and then send the pull request.
* Added small and temporary debug backend in 'tclprint_to_result_Constraint' in 
'constraint.c' and wrote corresponding tcl script 
'constraint_rhomboid_olaf.tcl'. The implementation of 'constraint rhomboid' and 
'lbboundary rhomboid' has now been proven to be correct.
* Removed debugging code and test scripts. Commit is ready to be pulled into 
master.
* doxygen stuff corrected
* small changes in the doxygen code
* unified lb-boundaries and lb-boundaries-gpu rhomboid added
* Added check for left handed trihedron in 'constraint rhomboid' and 
'lbboundary rhomboid'. Users do not have to take care of the handedness any 
more. Added documentation for 'lbfluid print', 'constraint rhomboid' and 
'lbboundary rhomboid'. Wrote a demo script 'ttt/angeberscript.tcl' and Fixed 
the indentation scheme in lb-boundaries.c - somehow it was completely messed up.
* tcl interface of lb gpu and cpu features equalized
* forgotten files added
* temp commit 4 merge with updatet master
* get lb_boundary_falg_gpu added
* small commit hopefully now merge with master
* Merge branch 'master' into lbboundaries
* several tcl fkt changed/added for gpu code (print boundaries dont work!!!)
* lbfluid print boundary 4 gpu finished and lbnode set veloctiy added
* changes in LB tcl interface and independet compilation of GPU and CPU code 
but still not all features tested!
* changes for tcl interface hopefully complete
* Merge branch 'master' into lbgpu
* Merge branch 'lbboundaries' of github.com:roehm/espresso into lbboundaries
* small fix in lb.c to get the Lb cpu code set by default
* Prepared pull request. Removed test scripts and made sure all my and Dominic 
Roehms changes are included.

-- File Changes --

M doc/ug/lb.tex (74)
M doc/ug/part.tex (53)
M scripts/auxiliary.tcl (122)
M src/Makefile.am (1)
M src/communication.c (22)
M src/communication.h (6)
M src/constraint.c (215)
M src/constraint.h (695)
M src/initialize.c (16)
M src/interaction_data.h (18)
M src/lb-boundaries.c (940)
M src/lb-boundaries.h (13)
M src/lb.c (814)
M src/lb.h (19)
D src/lb_boundaries_gpu.c (453)
D src/lb_boundaries_gpu.h (84)
M src/lbgpu.cu (284)
M src/lbgpu.h (80)
M src/lbgpu_cfile.c (597)
M src/polymer.c (3)
M src/statistics_fluid.c (40)
M testsuite/lb_gpu.tcl (1)
A ttt/check_rhomboid_distancefunction.tcl (6)

-- Patch Links --

  https://github.com/espressomd/espresso/pull/11.patch
  https://github.com/espressomd/espresso/pull/11.diff

--- 
Reply to this email directly or view it on GitHub:
https://github.com/espressomd/espresso/pull/11