Carsten Kutzner [Tue, 16 Oct 2012 13:32:17 +0000 (15:32 +0200)]
Made g_tune_pme read (ns/day) from 4.6 md.log output files
Change-Id: I8a3f804ac71cd13b3e060aa7b42032fbbbd93dcd
David van der Spoel [Sat, 13 Oct 2012 19:31:04 +0000 (21:31 +0200)]
Merge "Fixing issues with replica exchange" into release-4-6
Mark Abraham [Sat, 13 Oct 2012 00:00:48 +0000 (11:00 +1100)]
Fixed install-xxx functionality broken in
9628e27d
Only the CPack component definitions needed fixing, inasmuch as they
are what is used for the "make install-xxx" functionality. I separated
"libraries" component into "libraries", "libraries-gmxpreprocess" and
"libraries-gmxana" for, respectively, the core gmx/md/etc. libraries,
and those particular to src/kernel programs and src/tools programs.
Now "make install-xxx" correctly compiles, links and installs only the
required dependencies, and the wrong dependencies introduced in the
above commit are removed.
Fixes #1015
Change-Id: I3e848e69d126305943370d1f294e44d5389e1934
Berk Hess [Fri, 12 Oct 2012 12:46:18 +0000 (14:46 +0200)]
Merge "Fix illegal barrier in worksharing" into release-4-6
Roland Schulz [Fri, 12 Oct 2012 04:54:02 +0000 (00:54 -0400)]
Fix illegal barrier in worksharing
According to section 2.10 of the OpenMP standard:
A barrier region may not be closely nested inside a worksharing
... region.
Replaced parallel-for with parallel plus gmx_omp_get_thread_num.
Fixes #937
Change-Id: Ief31f2097fb084d405e5981913cce2c2bfee868c
Michael Shirts [Tue, 11 Sep 2012 03:11:49 +0000 (23:11 -0400)]
Fixing issues with replica exchange
Avoids unnecessary communication, clarifies variable names, restores
exchange statistics removed in
c7a82654, fixes memory leaks, and some
other minor house-keeping noticed by Mark Abraham.
Fixes #1001
Change-Id: I8a9032e5946117d87f672e16cba525c5ed9720f9
Christoph Junghans [Wed, 10 Oct 2012 15:41:58 +0000 (17:41 +0200)]
fixed header for 3rd party apps
Change-Id: Iaeb900134bfd5cdba2584c256ba04e0208f9ee84
Berk Hess [Mon, 8 Oct 2012 20:38:43 +0000 (22:38 +0200)]
made non-bonded potential shifts mdp options
Change-Id: I03e500c28ebd677917571cca7a7551523f9eb3af
Roland Schulz [Sat, 6 Oct 2012 22:59:36 +0000 (18:59 -0400)]
Fix incorrect free
introduced by
23ca07b5 /
7b6508e8
Change-Id: I713827221923270f8ba94b3ed33ee126f3ce084c
Berk Hess [Mon, 1 Oct 2012 13:31:13 +0000 (15:31 +0200)]
fixed backward compatibility of tpr files with dihres
Change-Id: Ic3104db7fd401d7bf7022da1aeeea49ca8955320
Sander Pronk [Tue, 2 Oct 2012 13:27:46 +0000 (15:27 +0200)]
Added two #defines that state the gromacs API + program version.
Added two #defines in a new CMake-generated header file version.h:
GMX_API_VERSION, and GMX_VERSION so third-party tools can provide some
backward compatibility through
#if !defined(GMX_API_VERSION) || (GMX_API_VERSION < 40600)
.... <do 4.5 specific stuff>
#elif (GMX_API_VERSION < 40700)
.... <do 4.6 specific stuff>
#endif
For now (pre-5.0), version.h is included in typedefs.h, which should
already be included in most gromacs-compatible code.
The API version is a separate CMake variable that currently gets its
value from the major+minor release version.
Change-Id: I9898cf378216d63c1456ce6566a2e42774fc1a0c
Berk Hess [Mon, 24 Sep 2012 16:51:54 +0000 (18:51 +0200)]
improved the energy conservation with SETTLE
Change-Id: Ib1c7dce6a353d80b396b86597ef842a6f4dc6d30
Szilard Pall [Tue, 2 Oct 2012 10:27:00 +0000 (12:27 +0200)]
added Verlet scheme and NxN non-bonded functionality
This commit implements a new "Verlet" cutoff scheme which uses
a exact cut-offs and standard Verlet lists with an automatically
calculated buffer.
The Verlet code-path supports full multi-level heterogeneous
parallelization using MPI/thread-MPI, OpenMP multi-threading,
and GPU acceleration for the non-bonded calculations.
The non-bonded calculations with the Verlet scheme support highly
optimized CPU SIMD acceleration using SSE/AVX and GPU acceleration using
NVIDIA CUDA. The CPU kernels have been tested on and optimized for most
x86 architectures including recent ones like Intel Sandy/Ivy-Bridge and
AMD Bulldozer. The CUDA GPU kernels support hardware of compute
capability 2.0 and above and are optimized for both Fermi and Kepler
architectures.
The new search code has been added in nbnxn_search.c, new non-bonded
kernels in nbnxn_kernels and nbnxn_cuda:
- plain-C kernels: reference CPU implementation and reference GPU
(emulation)
- x86 128- and 256-bit SIMD kernels (SSE2, SSE4.1, AVX_128, AVX_256
intrinsics)
- CUDA (two versions: for recent and legacy toolkit/drivers)
This commit also implements some additional optimizations targeting
performance:
- SSE acceleration for dihedrals;
- automated PP/PME load balancing called "PME tuning" which optimizes
the electrostatics cut-off to improve load balance between CPU and GPU
or separate PP and PME processes;
- hardware detection and automated run-configuration selection.
Change-Id: I3e1a15331c174265ec086565b978ffd079df2aaa
Roland Schulz [Mon, 1 Oct 2012 13:20:00 +0000 (09:20 -0400)]
Merge release-4-5-patches into release-4-6
Conflicts:
src/gmxlib/confio.c
src/gmxlib/copyrite.c
src/gmxlib/statutil.c
src/kernel/readir.c
src/kernel/repl_ex.c
src/kernel/runner.c
Change-Id: I3828c50f8166a6095ee64672cfd4380d3e4c86f6
Roland Schulz [Sun, 30 Sep 2012 19:57:59 +0000 (15:57 -0400)]
Fix problems found by cppcheck 1.56
Doesn't fix all problems. New suppressions in Jenkins for 4.6
(so that I can upgrade without having to fix all):
--suppress=uninitvar:src/gmxlib/md5.c --suppress=nullPointer:src/mdlib/mdebin_bar.c --suppress=nullPointer:src/gmxlib/oenv.c --suppress=nullPointer:src/gmxlib/selection/selhelp.c \
--suppress=memleak:src/tools/levenmar.c --suppress=selfAssignment:src/gmxlib/nonbonded/nb_kernel_bluegene/* \
--suppress=invalidPointerCast:src/gmxlib/checkpoint.c --suppress=invalidPointerCast:src/gmxlib/nonbonded/nb_kernel_sse2_single/* --suppress=invalidPointerCast:src/mdlib/fftpack.c \
--suppress=obsoleteFunctionsusleep:*.c --suppress=obsoleteFunctionsctime_r:*.c --suppress=invalidPrintfArgType_sint:*.c invalidPrintfArgType_uint:*.c
The remaining nullPointer are false positives as far as I know.
Problems found in 4.6 (haven't run cppcheck on 4.5) but
applied to 4.5 because it is also affected.
Change-Id: I5c0b4bce52a0c9ff6ef4ba51721cdf5baff42814
Christoph Junghans [Thu, 27 Sep 2012 20:13:40 +0000 (22:13 +0200)]
Merge "Fixed an off-by-one error in atom counting in gmx_dist." into release-4-5-patches
Teemu Murtola [Thu, 27 Sep 2012 17:56:52 +0000 (19:56 +0200)]
Merge "Fix insolidangle selections near poles." into release-4-5-patches
Sander Pronk [Wed, 26 Sep 2012 18:12:48 +0000 (20:12 +0200)]
Fixed an off-by-one error in atom counting in gmx_dist.
Replaced atom number in gmx_rmpbc with total number, so it can
accomodate distances between parts of molecules.
Change-Id: I6279c6ae6d61fd761c4d8d37a01c7d856742e6b8
Christoph Junghans [Tue, 25 Sep 2012 01:45:03 +0000 (03:45 +0200)]
Merge "Fix REMD use of gmx_large_int_t for #steps variable" into release-4-5-patches
Teemu Murtola [Mon, 24 Sep 2012 04:48:01 +0000 (07:48 +0300)]
Add pdbname selection keyword.
As a supporting change, remove trailing space from t_pdbinfo.atomnm, as
the trailing whitespace does not seem to be used anywhere.
This makes it possible to use it easily in the selection code.
Fixes #1007; fix backported from Iac36bda8.
Also includes changes from
8bddac3 to make the backport easier.
Change-Id: Iac36bda8a84d0a6c131445e7f47ad91d7209fb10
Kasson [Mon, 24 Sep 2012 04:55:31 +0000 (06:55 +0200)]
Merge "Fixed segfault with FE code for older tpr files." into release-4-6
Mark Abraham [Mon, 24 Sep 2012 02:06:37 +0000 (12:06 +1000)]
Fix REMD use of gmx_large_int_t for #steps variable
Change-Id: I0a72a8b838de4d018c2715f9d620a86134969206
Teemu Murtola [Tue, 18 Sep 2012 05:11:47 +0000 (08:11 +0300)]
Fixes for numeric and position selection variables.
Fixed handling of numeric and position selection variables in cases
where they were referenced multiple times.
Backported from Ifbb585ca.
Fixes #1006.
Change-Id: Ieaa479b4b870e044dc269339be57f6b33533c46e
Mark Abraham [Mon, 17 Sep 2012 13:45:35 +0000 (23:45 +1000)]
Made construction of parallel filenames less noisy
Berk introduced this in
6b4a52fc, but it is not essential to the
function there, is not very useful for the user, and spams the stdout
with every possible filename for every simulation in (say) a
multi-simulation.
Change-Id: Ifd017c3907f6a46a4f86ac9bc06c5983a5dea946
Teemu Murtola [Fri, 14 Sep 2012 17:28:29 +0000 (20:28 +0300)]
Fix insolidangle selections near poles.
- Correctly treat cases where a point is so close the one of the poles
that it completely covers one or more bins in the zenith angle
direction.
- Adjust update_surface_bin() to be more robust to rounding errors and
easier to understand by simplifying the code used for wrapping angles
to [-pi, pi] interval. Problems were triggered here by the first
change.
- Update comments.
Fixes #1004.
Change-Id: I2707f775793fabb64ea197bdae5fbfe68d6a8933
Kasson [Sat, 15 Sep 2012 15:04:38 +0000 (17:04 +0200)]
Merge "Fix some memory leaks." into release-4-5-patches
David van der Spoel [Sat, 15 Sep 2012 12:52:22 +0000 (14:52 +0200)]
Merge "Adapted NMR related data to new pdb format." into release-4-5-patches
Christoph Junghans [Fri, 14 Sep 2012 15:47:05 +0000 (17:47 +0200)]
Merge "fixed recent bug with grompp check for P-R P-coupling" into release-4-5-patches
Berk Hess [Thu, 13 Sep 2012 16:34:12 +0000 (18:34 +0200)]
fixed recent bug with grompp check for P-R P-coupling
Fix for recent commit
57dac834
Change-Id: I5bc631d6db848c8702a65c42ae27f096804abbf7
Roland Schulz [Thu, 13 Sep 2012 16:41:47 +0000 (12:41 -0400)]
Backport of md-vv rerun check from
c7a82654f2eb8992
Rerun in parallel with md-vv is broken. The check was added
in
c7a82654f2eb8992 but also affects 4.5.
Fixes #1003
Change-Id: Ie0e200a8328b2baecd2eba9eea1453c65a668633
Sander Pronk [Thu, 13 Sep 2012 09:26:09 +0000 (11:26 +0200)]
Fixed a deadlock in tMPI_Comm_free()
Fixes a deadlock on overcommitted systems while calling tMPI_Comm_free(),
as first noticed by Roland.
Brings tMPI_Barrier()'s use of atomic variables and memory barriers
in line with the rest of the thread_mpi code.
Change-Id: I86613cdf3c657e907f1168feb544e534889d0557
Teemu Murtola [Thu, 13 Sep 2012 04:20:42 +0000 (07:20 +0300)]
Fix some memory leaks.
Fixes part of #1004.
Change-Id: I22280999c5c3c40e8efae45461bd53bae8e2ff40
Roland Schulz [Wed, 5 Sep 2012 10:57:56 +0000 (06:57 -0400)]
Set valid build information if gmx_detectcpu fails
If gmx_detectcpu failed the error message was written
into the build info which could cause a compiler error.
Change-Id: Id7e367595c14270dc8fa91daa97ce13a4c686057
Mark Abraham [Wed, 12 Sep 2012 00:17:28 +0000 (02:17 +0200)]
Merge "Don't fail linking if man page building fails" into release-4-6
Christoph Junghans [Mon, 10 Sep 2012 23:07:21 +0000 (17:07 -0600)]
updated references
Change-Id: I74a53433b31c0d50524e0edb5865a8ef9fca1956
Roland Schulz [Mon, 10 Sep 2012 13:01:01 +0000 (15:01 +0200)]
Merge "Update use of CPack components and add install targets" into release-4-6
Mark Abraham [Mon, 10 Sep 2012 09:52:22 +0000 (19:52 +1000)]
Corrected REMD citation
The Fukushima paper relates exclusively to MC and spin glasses.
Change-Id: I58c4eafbe41b0cbd1f3e4afac6f34458ed27eea2
Christoph Junghans [Fri, 7 Sep 2012 19:29:06 +0000 (21:29 +0200)]
Merge "Fix VMD plugins on Windows" into release-4-6
Christoph Junghans [Fri, 7 Sep 2012 19:28:31 +0000 (21:28 +0200)]
Merge "Check that gmx_detectcpu runs correctly" into release-4-6
David van der Spoel [Fri, 7 Sep 2012 19:13:22 +0000 (21:13 +0200)]
Adapted NMR related data to new pdb format.
Since gromacs had adopted the pdb output format to
some standard, the script to convert xplor files
with NMR restratints, such as can be downloaded from
the pdb do not work anymore. The script and accompanying
table have been updated to work with the new pdb format
and simultaneously the extra argument for a residue offset
has been removed. Now the script can be run like
xplor2gmx.pl conf.pdb < restraints.md > disre.itp
Change-Id: I7c3c8bd49b002e2bf0760f4226c913aa36d64611
Roland Schulz [Thu, 6 Sep 2012 19:42:04 +0000 (15:42 -0400)]
Fix illegal call to MPI without being initialized
Introduced by
babdd87ca126
Change-Id: I428ef8ce706cc3a23fc980fa54c3e26e7a6d68b2
Roland Schulz [Thu, 6 Sep 2012 17:04:48 +0000 (19:04 +0200)]
Merge "Fixed .g96 trajectory reading." into release-4-5-patches
Roland Schulz [Wed, 5 Sep 2012 14:01:26 +0000 (10:01 -0400)]
Don't fail linking if man page building fails
If building of man page fails because binary cannot be
executed (e.g. undetected cross-compile) then fail
silently. A warning is printed when installing.
Only affects deverloper version (not source from cpack).
Change-Id: I9201d44a103602bb26a29b03d14ce08a0089540a
Roland Schulz [Wed, 5 Sep 2012 10:04:00 +0000 (06:04 -0400)]
Check that gmx_detectcpu runs correctly
If it doesn't run correctly set acceleration to None and
print warning. Same as check whether it compiled, which already
existed.
Change-Id: I59a2d531c41f931d162725540d5483b39aa7a3df
Sander Pronk [Wed, 5 Sep 2012 09:36:23 +0000 (11:36 +0200)]
Fixed an undefined return value in tMPI_Scan()
As caught by Szilárd, the tMPI_Scan() return value was undefined
for rank==0. Also removed an unused variable in tMPI_Thread_create().
Change-Id: I0b15f55089f50e0c237635ad322c7d38ce82c275
Christoph Junghans [Tue, 4 Sep 2012 16:42:00 +0000 (18:42 +0200)]
Merge "Fixes too many backup files for *.nroff files" into release-4-6
David van der Spoel [Tue, 4 Sep 2012 13:27:57 +0000 (15:27 +0200)]
Made tutorial slightly less confusing.
Change-Id: Ief3132a54770e8b3c3b69b24b5582ba321ccc5fd
Sander Pronk [Mon, 3 Sep 2012 11:58:25 +0000 (13:58 +0200)]
Fixed .g96 trajectory reading.
The line variable in .g96 configuration reading can now be made
persistent - and is done so for reading .g96 trajectories.
This commit fixes #781
Change-Id: If2005a00bb925b41802c4b0dc92e66eb5635f22c
Kasson [Mon, 3 Sep 2012 10:53:57 +0000 (12:53 +0200)]
Merge "Document correct comm_mode option" into release-4-5-patches
Kasson [Mon, 3 Sep 2012 10:51:40 +0000 (12:51 +0200)]
Merge "Improved pdb2gmx output" into release-4-5-patches
Szilard Pall [Mon, 3 Sep 2012 10:40:56 +0000 (12:40 +0200)]
Merge branch 'release-4-5-patches' into release-4-6
Conflicts:
src/mdlib/tpi.c
Change-Id: I1b5d92291ece7cf3194b4e1d88cd686a5aa18696
Roland Schulz [Mon, 3 Sep 2012 02:49:27 +0000 (22:49 -0400)]
Fix VMD plugins on Windows
VMD libraries are named *.so even on Windows
Change-Id: I4502af3a716b79820502c70fcfe2fc68749a77fc
Roland Schulz [Sun, 2 Sep 2012 16:26:22 +0000 (12:26 -0400)]
Fix stack error in grompp
vsiteatoms array was one to short
Fixes #997
Change-Id: I0897694dfc431a20bd1ebf769b24d09038d588b2
Mark Abraham [Wed, 22 Aug 2012 08:21:34 +0000 (18:21 +1000)]
Update use of CPack components and add install targets
* Removed component runtime and replaced it with component "scripts",
a component for each tool, and the component group tools. This means
CPack is capable of dealing with a single tool, or with the whole
group.
* Moved gmxpreprocess library to libraries component.
* Created CPack dependencies so that shared libraries accompany
executables.
* Created non-CPack install targets for each GROMACS tool, e.g.
"make install-grompp". Shared libraries are installed with these
tools as appropriate. Using library suffices could be right here,
but that's the user's problem to solve.
Development, regression testing and bug testing is more convenient if
you can easily build and install just a subset of tools. This could
always have been done with CMake black magic, but these targets makes
lives easier. Also users can now easily install MPI or
double-precision flavours of selected tools as suitable.
Change-Id: I1053f7ff91d50169ac930f1ac5b84378d3ab2e67
Mark Abraham [Fri, 31 Aug 2012 03:43:29 +0000 (13:43 +1000)]
Improved pdb2gmx output
Now there is explicit feedback that pdb2gmx is starting generating
hydrogens and termini, which can help the user deduce that problems
lie with those databases, and not necessarily the residue topology
database.
Change-Id: Ie4100b8586d6389a6965ea8cb09eb24cc4e7d3d9
Roland Schulz [Thu, 30 Aug 2012 03:34:21 +0000 (23:34 -0400)]
Fixes too many backup files for *.nroff files
Problem introduced by
26bd5f7ac7af. nroff files are generated
automatically and one could get erorr message when 99 backup files
existed.
Change-Id: Ia0a5c7e04064a0d304266e1dc6af347d3e846864
Berk Hess [Wed, 29 Aug 2012 16:00:08 +0000 (18:00 +0200)]
bug fix for pme.c MPI_COMM_NULL fix
dcf8b67e
Change-Id: Idb233db1337bcb881041caec96572e5df6183e6e
Mark Abraham [Tue, 28 Aug 2012 01:16:19 +0000 (11:16 +1000)]
Document correct comm_mode option
Change-Id: I18d9896c51d486dfcc52e4d15b052d65de91426d
Sander Pronk [Tue, 21 Aug 2012 11:10:45 +0000 (13:10 +0200)]
Newer version of thread_mpi, with various fixes, including tMPI_Scan()
- Implements tMPI_Scan()
- Many doxygen fixes
- removed unused variables from atomic operations
- Added C++11 compatible thread and mutex class
Change-Id: I4cce48627fa846da3c599a7de64d3898c358390b
Sander Pronk [Thu, 23 Aug 2012 09:16:05 +0000 (11:16 +0200)]
Fixed segfault with FE code for older tpr files.
Full all_lambda array wasn't allocated for gmx 4.5 tpr files, even
though the rest of the code depends on it.
Change-Id: I3e4b6273d9e769a430fffc95e3a9c94fb672f6f3
Bogdan Iorga [Wed, 22 Aug 2012 16:30:26 +0000 (18:30 +0200)]
Added PGLU entry in aminoacids.n.tdb for OPLS-AA.
Previously, pdb2gmx would fail during terminus selection for this residue.
Change submitted initially on the master branch, then abandoned.
(https://gerrit.gromacs.org/#/c/1290/)
Fixes #804.
Change-Id: Ic810242dc65019858c0332d352ef1ba96d4662d6
Christoph Junghans [Mon, 20 Aug 2012 17:52:48 +0000 (19:52 +0200)]
Merge "Fixing of things associated with aminoacid.rtp header" into release-4-6
Christoph Junghans [Mon, 20 Aug 2012 17:51:08 +0000 (19:51 +0200)]
Merge "Clean up some variable naming for eigensolving" into release-4-6
Christoph Junghans [Mon, 20 Aug 2012 17:50:53 +0000 (19:50 +0200)]
Merge "Fix coordinate is zero check in gmxcheck" into release-4-5-patches
Christoph Junghans [Mon, 20 Aug 2012 17:50:35 +0000 (19:50 +0200)]
Merge "Explicitly noted an issue with use of genbox -maxsol" into release-4-6
Roland Schulz [Sat, 18 Aug 2012 19:30:09 +0000 (21:30 +0200)]
Merge "fix AdResS and long-range electrostatics warning" into release-4-6
Roland Schulz [Thu, 16 Aug 2012 23:50:15 +0000 (19:50 -0400)]
Fix coordinate is zero check in gmxcheck
Change-Id: I7504d235edc659b586d61243f3cbf72add2f92b2
David van der Spoel [Thu, 16 Aug 2012 07:45:41 +0000 (09:45 +0200)]
Optional computation of fluct. props in g_energy. Fixes #954
Added a command line option -fluct_props to turn on the computation
of heat capacity etc. in g_energy. This prevents a warning message
from appearing on the screen each time the program is run.
Change-Id: I0cac44494ec5a366762d4d0a915146298d2ce1e0
Mark Abraham [Tue, 14 Aug 2012 05:02:34 +0000 (15:02 +1000)]
Explicitly noted an issue with use of genbox -maxsol
Change-Id: I6f16c94296e7e17150331e4d3072d6430fb87bff
Mark Abraham [Tue, 14 Aug 2012 04:59:11 +0000 (14:59 +1000)]
Don't silently accept genbox -nmol without -ci
Change-Id: Ia50da1ee741eec9ea49ec04eb9fd14dd6b3116fc
Christoph Junghans [Mon, 13 Aug 2012 01:57:56 +0000 (19:57 -0600)]
fix AdResS and long-range electrostatics warning
Change-Id: Ibe6615e327a1a386f2fec5746e843fcf8fba5401
Mark Abraham [Fri, 10 Aug 2012 01:41:25 +0000 (11:41 +1000)]
Clean up some variable naming for eigensolving
Change-Id: I42c1751f505e931c7245a4804c1609d0e369e23f
Roland Schulz [Sat, 28 Jul 2012 18:32:18 +0000 (14:32 -0400)]
Make GCC OpenMP libraries static if GMX_PREFER_STATIC_OPENMP
Change-Id: I9ef26070c289f83f9d0f9e76635bb3a3dd493570
Roland Schulz [Sat, 28 Jul 2012 17:25:34 +0000 (13:25 -0400)]
Guarantee OpenMP flags for linker and make them user modifiable
- Linker files were set only C/C++ compiler flags. For Fortran
language it wasn't guranteed that the linker had the correct flags.
- The FindOpenMP module doesn't check for correct openmp linker flags
and we assumed that compiler, library-linker, and executable linker
flags are all the same.
- This also allows to create binaries with OpenMP linked in
statically.
Change-Id: Ifde1d3f875bb779f04aa44a969138fe04620f224
Mark Abraham [Sat, 5 May 2012 03:54:18 +0000 (13:54 +1000)]
Fixing of things associated with aminoacid.rtp header
Code, commenting and documentation changes, including
* expanded self-documention boolean variable names
* clean up of dihedral generation code
* more reuse of output code
* more use of t_restp objects
* documentation in every force field's .rtp files
* clarified why and when generated dihedrals get superseded
Change-Id: I384f498f1124795e33c0ec1707ad52558bf15eb1
Mark Abraham [Tue, 31 Jul 2012 01:05:19 +0000 (11:05 +1000)]
Added TIP4P back into CHARMM27
I think Par erroneously removed this in
73b6d636a
Change-Id: Ie9f64636b1b28030ec86248649acad343b7c9a76
Berk Hess [Thu, 9 Aug 2012 12:25:24 +0000 (14:25 +0200)]
fixed trjconv not writing velocities with cpt input
Fixes #930
Change-Id: I554dd729484a4238620f2773b5c3b3672cab72ee
Christoph Junghans [Tue, 7 Aug 2012 16:10:56 +0000 (18:10 +0200)]
Merge "fixed incorrect TPI results with dynamic box" into release-4-5-patches
Carsten Kutzner [Tue, 7 Aug 2012 09:15:59 +0000 (11:15 +0200)]
Fixed essential dynamics / flooding group PBC serial
In former versions, the PBC representation of essential dynamics /
flooding group atoms could be incorrect in serial runs if the ED group
contained more than a single molecule. In multi-molecule cases, the required
steps to choose the correct PBC image in communicate_group_positions()
therefore need to be performed also in serial runs. Since the PBC representation
can only change in neigborsearching steps, we only need to check the
shifts then. In parallel, NS is signalled by the bUpdateShifts
variable, which is set in dd_make_local_ed_indices(). The latter
function is however not called in serial runs; but still we can pass
the bNS boolean to do_flood() to signal the NS status. For essential
dynamics, unfortunately, since do_edsam() is called from constrain(), there
is no information about the NS status at that point. Until someone
comes up with a better idea, we therefore do the PBC check in every step
in serial essential dynamics - the performance impact will be negligible
anyway.
Change-Id: I86336a5e34131bdeac7e28f35b1ccb633450e54e
Berk Hess [Mon, 6 Aug 2012 18:52:59 +0000 (20:52 +0200)]
fixed incorrect TPI results with dynamic box
TPI with a box in the -rerun file different from the tpr gave systematic
errors. In Gromacs 4.0 there was a minor error, but not systematic.
Now TPI with any box setup is fully correct. Fixes #799
Change-Id: I69279874c2d1192e63efa224a0f035d43e919e88
Christoph Junghans [Mon, 6 Aug 2012 17:18:50 +0000 (19:18 +0200)]
Merge "Added comment in get_structure about which structures are included" into release-4-5-patches
Carsten Kutzner [Mon, 6 Aug 2012 13:45:08 +0000 (15:45 +0200)]
Fixed calculation of projection for STAR and/or SORI
Both for STAR and for SORI, if provided to make_edi, the
fit structure and the average structure can be the same
(in these cases the old code works fine), or one can do
the fit on another (possibly larger) part of the molecule.
In the latter case, STAR (or SORI) contains first the fit and
then the average structure (concatenated). Here, the fit has to
be performed using the fit structure while the projection(s) have
to be calculated using the average structure. This is now
taken care of by the 'avindex' index variable.
Change-Id: Id97235030f3394df2931e30249083ad439dbe791
Carsten Kutzner [Mon, 6 Aug 2012 12:26:53 +0000 (14:26 +0200)]
Added comment in get_structure about which structures are included
This patch just adds a comment and refactors natoms -> nav to better
clarify when we deal with the AVERAGE and when with the REFERENCE (fit)
structure. 'natoms' is not a good choice if we are talking about
the number of atoms in the average structure only.
Change-Id: I7025fb6ea59af0a7bcd7953a25abd09be0dd2979
Roland Schulz [Sun, 5 Aug 2012 23:01:45 +0000 (19:01 -0400)]
Remove share/tutor
Not up-to-date and better tutorials exists.
Fixes #811
Change-Id: I52708a534e641039ccb05538e0a5387079370761
Roland Schulz [Sat, 4 Aug 2012 18:12:18 +0000 (20:12 +0200)]
Merge "Fixed segv in essential dynamics module" into release-4-5-patches
Carsten Kutzner [Thu, 2 Aug 2012 09:52:41 +0000 (11:52 +0200)]
Fixed segv in essential dynamics module
Change-Id: Ia5657a535e7e31d55bf2b310a20dfee2a5e3b7f4
Roland Schulz [Fri, 3 Aug 2012 02:02:09 +0000 (22:02 -0400)]
Fix overlapping memcpy
The memcpy were not neccessary.
Fixes #984
Change-Id: Iea1d34756444cba2493940ab3f2252e7c108323f
Christoph Junghans [Wed, 1 Aug 2012 23:32:17 +0000 (17:32 -0600)]
Add note about SD integrator and t-coupling
Change-Id: I6c2bd1a332abbaf5312c9ad46de24bcd7205a085
Roland Schulz [Wed, 1 Aug 2012 00:38:43 +0000 (20:38 -0400)]
Fix use after free memory error with adress
Change-Id: I481a55704a0a93235df531269bbbe40240dd12dc
Michael Shirts [Mon, 30 Jul 2012 03:19:19 +0000 (21:19 -0600)]
Added a check for efep that was missing.
Fixes #983
Change-Id: Icdc1dd68f782fd6385fd55b616993fadea0c4448
Christoph Junghans [Tue, 31 Jul 2012 00:24:04 +0000 (02:24 +0200)]
Merge "Clarified reasons for termination of EM" into release-4-6
Mark Abraham [Thu, 26 Jul 2012 14:34:11 +0000 (00:34 +1000)]
Clarified reasons for termination of EM
"Step size too small" can be misinterpreted as nsteps is too small,
and so the error message is now more explicit.
Change-Id: I5680b3d13f34f7186ce28c553ac5736c4a74818e
Michael Shirts [Wed, 18 Jul 2012 13:59:53 +0000 (09:59 -0400)]
Fix in when to call the constraints for andersen.
Change-Id: Ifa7eebd3a23312cb5abe66f7bf0d7a3299740f9a
Roland Schulz [Sun, 29 Jul 2012 07:24:26 +0000 (03:24 -0400)]
Fix Segfaults in Genion
Introduced by
babdd87c and
c7a82654f
Change-Id: Id42042966931f3b7dfbd2eebcde29d659f80c0f0
Fixes: 982
Roland Schulz [Tue, 3 Jul 2012 00:05:19 +0000 (20:05 -0400)]
Fix that VMD version check was performed for binary not plugin
Gromacs is not using the VMD binary and thus it is not very
insightful to check the VMD binary version. Also checking the VMD
binary version was potential very slow.
The new code tries to find and load one of the VMD plugins and
check the version of the plugin.
Change-Id: Iba7be9cb742c530ce4edb8963597b8ea467e5520
Roland Schulz [Wed, 25 Jul 2012 23:27:56 +0000 (01:27 +0200)]
Merge "cmake: allow switch between single and double" into release-4-6
Roland Schulz [Wed, 25 Jul 2012 23:24:54 +0000 (01:24 +0200)]
Merge "fix a segfault in sse_mask_init" into release-4-6
Roland Schulz [Tue, 24 Jul 2012 11:44:07 +0000 (07:44 -0400)]
Fix installation rules for man pages: don't install gromacs.7.cmakein
8c87e60 added gromacs.7.cmakein and didn't add an exclusion rule.
Change to CPackInit.cmake: Removing wrong (incorrect path) and unnessary file in message.
Change-Id: I8fe7436b8164d80ad6db64eb54dc81424cd44e4a
Roland Schulz [Tue, 24 Jul 2012 11:26:07 +0000 (07:26 -0400)]
Add default root folder to cpack archives
Fixes that source archives had no root folder. If non-default
root folder for binary archives is wanted, this still can be
achived by:
cmake -DGMX_INSTALL_PREFIX={path}/ -DCPACK_INCLUDE_TOPLEVEL_DIRECTORY=0
Fixes #977
Change-Id: Iaf42afd3617873e5c1e5e7cb7c327986efa20fcd
Mark Abraham [Tue, 24 Jul 2012 00:16:27 +0000 (10:16 +1000)]
Clean up of environment variables
In particular, reduce ambiguous usage of "force".
Refs #974
Change-Id: Ia6178ce3392f66451341dd7a4af98312f9e72c0f