ChangeLog.c30
| | |
|
CHARMM 30 |
| c30a1 |
Developmental |
August 15, 2002 |
| c30a2 |
Developmental |
February 15, 2003 |
| c30b1 |
Release |
August 15, 2003 |
| c30b2 |
Release |
February 15, 2004 |
|
| | |
|
About CHARMM30 Development |
|
|
|
As an on-going project, CHARMM development will be carried out with
CHARMM version 30 series. The base revision is the beta release version
c29b0. c30a1 is the first revision in the c30 developmental versions.
The following describes new features and enhancements incorporated in
the c30 development period.
The following are introduced and modified during the development period
of April 15, 2002 through August 15, 2002.
1.1 Nudged Elastic Band Method
Name : Paul Maragakis
Email Address : maragakis@cmt.harvard.edu
Institution : Harvard University
Date : June 27, 2002
We added the nudged elastic band (NEB) approach to the replica path.
In the NEB, one projects the spring forces along the path and the
potential forces perpendicular to the path. It is one of the methods
of choice for a quick and dirty approximation to the minimum energy
path. Because of the projection of the forces there is no exact
energy functional so that the NEB does not pass TEST1.
We introduced two new keywords to the RPATH command:
The keyword NEBA that activates the projections.
The keyword KNEB that is the spring constant for the NEB method.
It is possible to mix the other RPATH springs (e.g. KANG) with the NEB.
Files Modified:
source/energy/epath.src
/fcm/path.fcm
doc/replica.doc
New Testcase:
test/c30test/nebtest.inp
------------------------------------------------------------------------------
1.2 Structural overlap method by using Gaussian overlap integrals
Names : Milan Hodoscek
Email Address : milan@cmm.ki.si
Institution : National Institutes of Chemstry, Ljubljana, Slovenia
Date : July 9, 2002
This is a maximum overlap method to investigate the structural
similarity of flexible molecules. The atoms are described as Gaussians
and the interaction energy between different molecules are basically
overlap integrals. The Gaussians can represent either volume or charge.
Alternatively, the overlap of the electrostatic potential is provided
yielding exponential form.
The method is enabled by OVERLAP pref.dat keyword
Files Modified:
source/charmm/charmm_main.src, iniall.src
/energy/energy.src, eutil.src, printe.src
/fcm/energy.fcm
/util/prpref.src
New Files:
source/misc/olap.src
/energy/eolap.src
/fcm/olap.src
doc/overlap.doc
test/c30test/overlap.inp
------------------------------------------------------------------------------
1.3 TSRI Developments
Name : Michael S. Lee, Michael Feig, Mike Crowley and
Charles L. Brooks, III
Email Address : crowley@scripps.edu, brooks@scripps.edu
Institution : TSRI
Date : july 10 2002
1.3.1 Two-dihedral cross term based on interpolated grid map
pref keyword: CMAP
Files Modified:
install.com
source/charmm/charmm_main.src, iniall.src
/correl/correl.src
/dynamc/dcntrl.src, dynamln.src, dynio.src, mts.src, trnphi.src
/energy/ecntrl.src, ediff.src, eintern.src, energy.src, eutil.src
/intere.src, printe.src
/fcm/block.fcm, code.fcm, dimens.fcm, energy.fcm, exfunc.fcm
/image.fcm, param.fcm, pert.fcm, psf.fcm, rtf.fcm, tbmtsp.fcm
/tsmh.fcm
/flucq/flucq.src
/gener/genpsf.src, modpsf.src, psfsum.src, replica.src, update.src
/image/eimage.src, images.src, imagio.src, upimag.src
/io/mainio.src, psfres.src, rtfio.src
/manip/cstran.src
/mbond/mbond.src
/mc/mc.src, mcener.src
/minmiz/egrad1.src, nraph.src
/misc/hbuild.src, testch.src, xray.src
/pert/block.src, epert.src, icfcnf.src, icpert.src, pert.src
/tsme.src, tsms.src
/quantum/qmset.src
/util/prpref.src, sort.src
/vibran/vibran.src, vibsub.src
New Testcase:
test/c30test/cmaptest.inp
..............................................................................
1.3.2 Analytic and grid-based Generalized Born Methods
Three methods are available:
(1) Grid-based methodology is a two parameter model which reproduces
Poisson-Boltzmann solvation energies to within 1% accuracy.
(2) Analytical method I is a 5-parameter model which reproduces absolute
PB solvation energies to ~1.5% accuracy for any forcefield.
(Only PARAM19/PARAM22/MSI22 have been tested.) Relative solvation
energies are ok, but not much better than GBORN module for PARAM19.
It is faster of the two analytical approaches.
(3) Analytical method II is ~7 parameter model which reproduces absolute
PB solvation energies to ~1% accuracy. Relative solvation energies are
almost as good as grid-based method.
Note that
(A) All methods are intended to reproduce reentrant surface Poisson-Boltzmann
solvation energies at 0.25 Angstrom grid spacing.
(B) All parameters are largely forcefield independent, however tweaking
them for other forcefields probably couldn't hurt.
(C) Salt effects are available but not sufficiently tested.
(D) See JCP paper Vol. 116, #24, 10606-10614 (2002) for details on Methods (1)
and (2). Details of method 3) will be in upcoming J. Comp. Chem. article.
(E) Also consult gbmv.doc and c29test/gbmv.inp.
pref keyword: GBMV
Files Added:
source/energy/gbmv.src
/fcm/gbmv.fcm
Files Modified:
install.com
source/charmm/charmm_main.src, iniall.src
/energy/energy.src, intere.src
/util/prpref.src
doc/developer.doc
New Documentation:
doc/gbmv.doc
New Testcase:
test/c30test/gbmv.inp
------------------------------------------------------------------------------
1.4 Combined SCC-DFTB/CHARMM code for QM/MM Calculations
Name : Qiang Cui, Marcus Elstner
Email Address : cui@chem.wisc.edu, elstner@phys.upb.de
Institution : Dept. of Chem. Univ. of Wisc. Madison
Dept. of Theo. Phys. Univ. of Paderborn, Germany
Date : July. 2002
An approximate density functional theory, the self-consistent-charge
density functional tight binding (SCC-DFTB) method, has been implemented
into CHARMM in a combined QM/MM framework. The method is as fast as the
standard semi-empirical methods such as AM1 and PM3, with superior
accuracy for many systems tested so far (especially hydrogen bonding
interactions and reaction barriers).
The current implementation has analytical first derivative and thus
allows energy minimizations, reaction path search (e.g., travel) and
molecular dynamics simulations; SCC-DFTB/MM also works with Monte Carlo.
Replica can also be used, which makes it possible to use replica path
and related approaches (such the nudged elastic band) for determining
reaction path with the SCC-DFTB/MM potential; along the same line,
path integral simulations can be carried out as well, although only for
equilibrium properties at this stage.
The parameter files for the SCC-DFTB atom types need to be obtained
separately from Dr. M. Elstner (elstner@phys.upb.de).
Files Added:
source/sccdftbint/sccdftbini.src ---> interface code
/sccdftbint/sccdftbsrc/ ---> scc-dftb source code
/fcm/sccdftb.fcm ---> common block
build/UNX/scctbint.mk ---> interface makefile
/scctbt.mk ---> scc-dftb makefile
Files Modified:
install.com
source/charmm/charmm_main.src, iniall.src
/energy/energy.src
/gener/update.src
/mc/mcener.src
/nbonds/nbexcl.src, nbonda.src, nbonds.src
/quantum/addlnat.src
/fcm/gamess.fcm
New Documentation:
doc/sccdftb.doc
New Testcases:
test/c30test/alanine_scc.inp ---> std. SCCDFTB/MM calculation
/neb_scc.inp ---> Replica path with SCCDFTB
/split_scc.inp
------------------------------------------------------------------------------
1.5 NIH Developments
Name : Xiongwu Wu and Bernard R. Brooks
Email Address : brbrooks@helix.nih.gov
Institution : National Institutes of Health
Date : August 5, 2002
1.5.1 EMAP module for manipulate map objects
The EMAP module is designed to manipulate map objects as well as
interexchange between atomic objects and map objects. This module allow
CHARMM to handle low resolution structural information in map format in
molecular modeling studies. A core-weighted correlation function is
implemented for map comparison. A grid-threading Monte Carlo method is
implemented to search for the best fit of individual maps within a complex
map.
Files added:
source/emap/emapop.src, emapsubs.src, emapio.c, emapdock.src
/fcm/emap.fcm
Files modified:
source/charmm/charmm_main.src, iniall.src
/io/mainio.src
/util/prpref.src
doc/commands.doc
New Documentation:
doc/emap.doc
New Testcases:
test/c30test/emap.inp
/data/a7n.psf, a7n.crd
..............................................................................
1.5.2 Replica Path Code Enhancements
Four working replica/path codes are merged.
Files modified:
source/energy/energy.src, epath.src
/fcm/path.fcm
..............................................................................
1.5.3 Coordinate Manipulation Commands
COOR TWIST command is added. COOR DUPLicate command is extended.
Files modified:
source/manip/corman.src
..............................................................................
1.5.4 Minimizer Enhancements
NOEnergy function is added to STEEPD minimizer.
Files modified:
source/minmiz/steepd.src
..............................................................................
1.5.5 Standard Coding and Minor Enhancements
Tailing black lines are removed as some compilers do not compile source
files with extra lines. All variables must be decleared and fcm/impnon.fcm
should be included at the beginning of each subroutine and function.
Misplaced and/or misused ## keys are fixed.
Minor enhancements are:
(1) UNBUFIO() is called only for UNIX compilation
(2) ROOMT(298.15D0) and ZEROC are moved to fcm/consta.fcm
(3) Ability to move lonepairs added
(4) Some dimensions increased:
WDMAX=20 to 50 in miscom.src
MAXPAR=256 to 512 in cmdpar.fcm
(5) Marked comments in demands
(6) SETMSI calls added
Files fixed:
source/cff/efscalar_cff.src, enbfast_cff.src, escalar_cff.src, parrdr_cff.src
/charmm/charmm_main.src, iniall.src, miscom.src
/correl/mantim.src
/dynamc/dcntrl.src, dynamc.src, lonepair.src, nose.src, prssre.src
/energy/ecnstr.src, genborn.src
/fcm/cmdpar.fcm, consta.fcm, new_time.fcm, pbeq.fcm, reawri.fcm
/flucq/fluqqmmm.src
/gamint/blur.src
/image/nbndgcm.src
/machdep/machutil.src
/manip/corman3.src
/mbond/mbutil.src
/minmiz/tnpack.src
/misc/eef1.src, grid.src, surfac.src
/nbonds/ace.src, ewald.src, nbndcc.src, pmeutil.src
/pert/epert.src, tsmp.src, tsms.src
/prate/charmmrate.src
/quantum/qmleps.src, qmset.src
/rxncor/rxndef.src
/util/array.src, clcg.src, imsl.src
The following are introduced and modified during the development period
of August 15, 2002 through February 15, 2003.
2.1 TSRI Developments
Name : Sandeep Patel, Tom Cleveland, Z. Zhou, Mike Crowley,
Charles L. Brooks, III
Email Address : crowley@sripps.edu, brooks@scripps.edu
Institution : TSRI
Date : January 7, 2003
2.1.1 CHarge EQuilibration methods (CHEQ)
The CHEQ and associated modules implement polarization via the
fluctuating charge method as based on the CHarge EQuilibration methods
outlined in the literature. While the current forcefield parameters are
valid for most small molecules and proteins, the force field is
constantly undergoing refinement and development.
The electrostatic model derives formally from the density
functional theory of atoms in molecules; polarization is effected
as a result of chemical potential equalization everywhere within a
molecule, forcing charge flow from regions of high to low chemical
potential based on atomic properties. These properties are the atomic
hardness and electronegativity. The parameters are treated as such
and are determined from fits to density functional calculations of
charge responses and mono- and dipole moments of small molecules in
vacuum.
The method can be used to perform energy, minimization, and
dynamics calculations for the above-mentioned systems. For dynamics,
the charges are coupled to Nose-Hoover baths to maintain proper
adiabaticity. Several normalization schemes are allowed to maintain
charge constant over desired partitions. Several water models are
supported including the SPC-FQ and TIP4P-FQ models of Rick et al.
pref keyword: CHEQ
Files Added:
source/cheq/fqcom.src, fqener.src, fqminut.src
/fcm/cheqdyn.fcm, derivq.fcm
build/UNX/cheq.mk
doc/cheq.doc
test/c30test/cq1fsvvacu.inp, cq1fsvwat.inp, cqmeohvacu.inp, cqnmawat.inp,
cqtip4pbulk.inp
test/data/1fsv_dimens.stream, 1fsv_solv-octa.crd, 1fsv_vacu.crd, meoh.crd,
newcfq3.prm, newcfq3.rtf, newcfq3spc.prm, nmatip4p.crd, tip4p.crd
Files Modified:
install.com
source/charmm/charmm_main.src, iniall.src
/correl/anacor.src, solana.src
/dynamc/dynamc.src, dcntrl.src, dynio.src, dynsub.src,
trnphi.src, nose.src, dynamvv.src, dynamcv.src
/energy/ecntrl.src, energy.src, eutil.src, printe.src
/fcm/cheqdyn.fcm, derivq.fcm, contrl.fcm, energy.fcm, rtf.fcm, traj.fcm
/gener/genpsf.src, modpsf.src
/image/upimag.src
/io/mamio.src, psfres.src, rtfio.src, trajio.src
/manip/corman2.src, corman3.src, dynanal.src,
hbanal.src, intcor.src, scalar.src
/mbond/mbback.src
/mc/mcio.src, mvhmc.src
/minmiz/steepd.src, conjug.src, egrad1.src, minmiz.src,
abner.src, nraph.src, powell.src, tndriv.src
/misc/nmr.src
/nbonds/enbfast.src,enbond.src, enbonda.src, etable.src,
evdwf.src, ewald.src, ewaldf.src, ewaldf2.src, pme.src
/pert/block.src, icfcnf.src, puic.src
/rxncor/lupopt.src, travel.src
/util/selcta.src
/vibran/quasi.src, rbquas.src, vibio.src
..............................................................................
2.1.2 Makefiles and Code for Recent IBM and ALPHA machines
The modification involves better Makefiles and code changes for
more recent IBM machines, alpha and multiple MPI libraries (MPICH,
LAM, MPI_LITE, vendor).
For ALPHA platforms, modern alpha compile flags are incorporated
and alphamp is removed.
ibmrs, ibmsp, ibmsp3 are replaced with ibmaix and ibmaixmp in
install.com. The pref keyword used is IBMAIX. This should work on
all current ibm machines and allow us to remove the other IBM stuff
from code and from install.com. For now, the only ibm architectures
that install.com will allow are ibmaix for serial and ibmaixmp for
parallel with vendor mpi.
Files Added:
build/UNX/Makefile_ibmaix, Makefile_ibmaixmp
Files Modified:
install.com
test/test.com
build/UNX/Makefile_alpha
source/fcm/impnon.fcm, machio.src, machutil.src, space.src, startup.src
..............................................................................
2.1.3 FASTEW for SGI
Loop rearrangement and temp array usage for speedup in REWALD().
Only applies to sgi. FC3 macro is set to -O3 in Makefile_sgi and
Makefile_sgi64.
Files Modified:
install.com
build/UNX/Makefile_sgi, Makefile_sgi64, nbonds.mk
source/fcm/ewald.fcm
/nbonds/ewald.src
------------------------------------------------------------------------------
2.2 Karolinska Institute Fixes and Enhancements
Name : Lennart Nilsson
Email Address : Lennart.Nilsson@biosci.ki.se
Institution : Karolinska Institute
Date : January 13, 2003
2.2.1 COORdinate Analyses Enhancements
Modified calculation of hydration number, with time dependence and
more extensive output. Warn if site around which g(r) is calculated
is also in the excluded volume set. Fixed a bug in application of
excluded volume correction.
Files involved:
source/correl/solana.src
doc/corman.doc, subst.doc
..............................................................................
2.2.2 Entropy Calculation
Added entropy calculation according to Andricioaei and Karplus or
Schlitter, with atom selections and breakdown per residue. Also
changed internal order in which matrix is accessed (several places,
also for COOR DMAT) to use first index in inner loop, gives substantial
speedup for large systems. The IMSL routine LINV3F is introduced.
Added selection and breakdown per residue for THERMOdynamics
estimations in VIBRAN calculations.
Files involved:
source/manip/corman2.src
/util/imsl.src
/vibran/quasi.src, thermo.src, vibran.src,
doc/corman.doc, vibran.doc, subst.doc
test/c30test/entropyosc.inp, entropylj.inp
..............................................................................
2.2.3 RMSDyn Enhancements
Now reads trajectories using READCV. Selected atoms from
specified part of trajectory/(ies) read into memory before processing.
Added warning message for PQ option when calculation is not formally
converged.
Files involved:
source/manip/rmsdyn.src, helix.src
test/c30test/pqtest.inp
------------------------------------------------------------------------------
2.3 Cornell University Developments
Name : Nilesh Banavali, Toby Allen, Benoit Roux
Email Address : nilesh.banavali@cornell.edu,
toby.allen@cornell.edu, benoit.roux@med.cornell.edu
Institution : Weill Medical College of Cornell University
Date : December 31, 2002
2.3.1 MMFP Enhancements and Bugfixes
The enhancement involves addition of restraint to constrain a
particular set of atoms to the axis formed by another set of selected
atoms. Possible uses could be constraining motion of a ion along the
axis of an ion channel.
Flat bottom dihedral restraints are also added, which can be used
to allow unrestrained sampling of dihedral within a certain range
while restricting drifting of the dihedral beyond this range.
Restraint parameter can be written to specified file at each
step of energy determination.
Fixed problems in distribution of forces for angle/dihedral
constraints using MMFP and with use of distance-planar or
distance-cylinder constraints.
Files Modified:
source/energy/energy.src
/fcm/mmfp.fcm
/misc/mmfp.src
/pert/epert.src
doc/mmfp.doc
New Testcase:
test/c30test/mmfp_test5.inp
..............................................................................
2.3.2 Relative 1-D RMSD Restraints
Added 1-D RMSD relative restraint to constrain RMSD of a structure
with respect to two different endpoint structures with imposition of a
single restraint. Orients the present structure to selected atoms on
both end-point structures and imposes a harmonic energy penalty for
deviation of the difference in RMSD from the specified reference value.
E = force*(RMSD2 - RMSD1)**2
where RMSD1 is RMSD from end-point structure 1, RMSD2 is RMSD from
end-point structure 2.
Files Modified:
source/dynamc/dcntrl.src
/energy/rmsd.src
/fcm/rmsd.fcm
/manip/cstran.src
doc/cons.doc
New Testcase:
test/c30test/test_rmsd2.inp
..............................................................................
2.3.3 GSBP Fix
In GSBP, calculation of the reaction field matrix involves PBEQ
vacuum calculations, which should use a dielectric constant of 1.0 for
solute. Instead, they were using the user-specified dielectric
constant for these calculations, fixed by changing the solute
dielectric constant value passed to the MAYER subroutine. The impact
is small, but this seems more correct formally. Use of eps_protein=1
for reaction field matrix vacuum calculation in GSBP instead of
specified protein dielectric constant.
Files involved:
source/misc/gsbp.src, gsbp2.src
..............................................................................
2.3.4 ELSEIF Fix
There was a little screwup in the nested if-else-endif structure.
The nested if-else-endif structure was messing up the original if
statement of miscom. The fix implements complete backward compatibility.
File involved:
source/charmm/miscom.src
New Testcase:
test/c30test/if.inp
------------------------------------------------------------------------------
2.4 QM/MM (Semiempirical) Enhancements
Name : Jingzhi Pu, Mireia Garcia-Viloca, and Jiali Gao
Email Address : garcia@chem.umn.edu
Institution : University of Minnesota
Date : January 9, 2003
(1) The implementation of UHF method had a bug in the original
distribution, which has been fixed.
A line has been commented in subroutine moldat, which was adding
an extra electron to the total number of electrons for open shell
calculations with the UHF method, making the total number of beta
electrons equal to the total number of alpha electrons.
(2) Implementation of UHF method with QM/MM-GHO calculations.
The Generalized Hybrid Orbital method based on Unrestricted HF
theory (GHO-UHF) is implemented at semiempirical level (AM1, PM3) in
the quantum module. With this extension, GHO boundary treatment can
be used for open shell QM fragments in combined QM/MM calculations.
For a GHO-UHF wavefunction, we have two sets of auxiliary hybrid
orbitals for alpha spin and beta spin electrons respectively. The
charge density assigned to each of these auxiliary hybrid oribtal is
0.5(1.0-qB/3.0d0), while qB denotes the MM partial charge of GHO
boundary atom B. Similar to GHO-HF, the hybridization basis
transformation is carried out between the density matrix and Fock
matrix, both the alpha and the beta sets.
Analytical gradients and Mulliken population analysis are also
implemented for GHO-UHF.
(3) User defined damping SCF converger for UHF calculations with
QM/MM-GHO method.
A simple density damping option is added to the SCF driver for
the quantum module. The motivation of adding this option is to
provide a possibility to overcome SCF convergence difficulties.
Currently, this damping accelerator is only used to improve the SCF
convergence rate for GHO-UHF type calculations.
For an SCF iteration with density damping turned on, the actual
density matrix used for next iteration is computed by a linear
combination of the current density with previous one:
Pi = a * Pi-1 + (1-a)* Pi
The damping factor a is a user defined floating point number between 0
and 1. One can specify this damping factor with the optional keyword
'DAMP a' in the QUANtum command line. The default of this damping
factor is 0.0, i.e., no damping at all.
In the current implementation, several "damped" steps (with a
user defined damping factor "a" ) are carried out until the alpha and
beta density matrices are partially converged (density changes are
smaller than 100 times the density convergence criterion), then
"undamped" steps (a=0.0) follow until the final convergence is reached.
(4) Mulliken analysis for UHF and RHF calculations with QM/MM-GHO method.
(5) A bug has been fixed in the implementation of the GHO method for
multiple GHO atoms case.
(6) Updating AM1 parameters for sulfur.
The AM1 parameters for sulfur in the initial distribution did not
correspond to the MOPAC-version 6.0 sulfur parameters. The MOPAC-version
6.0 parameters have been introduced in the qmdata file and the
original AM1 sulphur parameters in this file have been commented.
Files Modified:
install.com
source/fcm/qmlink.fcm
/quantum/qmdata.src, qmene.src, qmset.src, qmjunc.src
doc/qmmm.doc
------------------------------------------------------------------------------
2.5 Parallelization of a variety of methods in CHARMM
Names : Milan Hodoscek, Robert Petrella
Email Address : milan@cmm.ki.si, petrella@tammy.harvard.edu
Institution : National Institutes of Chemstry, Ljubljana, Slovenia
and Harvard University
Date : January 15, 2003
Several methods within CHARMM were parallelized for more efficient
simulations. These are:
1. TSM Thermodynamic Perturbation
2. TMD Targeted Molecular Dynamics
3. HQBM Bised Molecular Dynamics
4. EEF1 Implicit solvent model
5. BYCC Nonbond list generation
6. PSSP Soft core potential in PERT
7. ADUMB Adaptive Umbrella Sampling
8. GAMESS New version of GAMESS (CC method is new)
9. ZEROM ZERo Order Minimization
10. MTS Multiple Time Step
Files Modified:
source/adumb/cmds.src, eadumb.src, eval.src, umb.src
/charmm/iniall.src
/dynamc/dynamc.src
/fcm/actclus.fcm
/nbonds/nbndcc.src, nbonds.src
/pert/epert.src, icpert.src, tsmp.src
doc/parallel.doc
-------------------------------------------------------------------------------
2.6 GAMESS-US Update and Code Cleanup
Names : Milan Hodoscek
Email Address : milan@cmm.ki.si
Institution : National Institutes of Chemstry, Ljubljana, Slovenia
Date : January 15, 2003
2.6.1 Support for new version of GAMESS
New GAMESS version 20 June 2002(R1) is supported in CHARMM
Files Modified:
source/energy/enst2.src (subroutine MINV renamed to MINVST)
/gamint/blur.src, ddi.src
/nbonds/fma.src (sunbroutine MTRANS renamed to BTRANS)
/util/array.src
..............................................................................
2.6.2 Code cleanups and compiler bugs
Some bugfixes for compiler, and modification for flexibility to
deal with new compilers (xdisp.c)
Files Modified:
source/flucq/fluqqmmm.src
/machdep/xdisp.c
/nbonds/grape.src
------------------------------------------------------------------------------
2.7 GAMESS-UK Interface Updated
Name : Paul Sherwood
Email Address : p.sherwood@daresbury.ac.uk
Institution : CLRC Daresbury Lab.
Date : January 3, 2003
Added tests in guk.mk for failure to configure gamess, previously
it just went on to compile it anyway using old settings. In gukini.src,
corrected tests for atom types for gamess interface and added some replica
path code. Reorganise ## directives in paral2.src, which is needed to get
Global Array version to work. Update the chmguk web address in
gamess-uk.doc.
Files Modified:
build/UNX/guk.mk
source/gukint/gukini.src
/fcm/gamess.fcm
/flucq/fluqqmmm.src
/machdep/paral2.src
doc/gamess-uk.doc
At the CHARMM meeting (The Scripps Research Institute, July 11-13,
2003), we decided to update c30a2 with the following enhancements and
fixes and release it as c30b1. The following have been introduced and
modified during the development period of February 15, 2003 through
August 15, 2003.
3.1 TSRI Developments
Name : Sandeep Patel and Charles L. Brooks, III
Email Address : brooks@scripps.edu
Institution : The Scripps Research Institute
Date : June 10, 2003
3.1.1. Polarizable model fluctuating charges
Code for generalized born model added. Works with proteins.
Calling argument list of WRITCV is fixed.
Files added:
toppar/par_all30_cheq_prot.inp, top_all30_cheq_prot.inp
Files Modified:
doc/cheq.doc
source/cheq/fqener.src
/dynamc/dcntrl.src, dynamc4.src, dynamvv.src, nose.src
/fcm/cheqdyn.fcm
/minmiz/steepd.src
/util/prpref.src
New Testcases:
test/c30test/cq1fsvvacu.inp, cq1fsvwat.inp, cqmeohvacu.inp, cqnmawat.inp
/cqtip4pbulk.inp, cq1ab1gbdyn.inp
test/data/top_all30_cheq_prot.inp, 1ab1gbtest.crd, 1fsv_min.crd, cubic.xtl
..............................................................................
3.1.2 Build fixes
IBM platform makefils are fixed and some compile options are set for
platform dependent compilation. Compromise solution to avoid ifc
optimization problems for svdcmp.f and intcor2.f. FC1 is used in
1adumb.mk and manip.mk for compilation, respectively, where ifc FC1 is
set to -O0. stdlib.h is added in socket.c. IBMAIX key is added in
machutils.rc. test.com is fixed to run IBM parallel testing.
Files modified:
build/UNX/Makefile_alpha, Makefile_ibmaix, Makefile_ibmaixmp,
/Makefile_ibmsp3
/adumb.mk, manip.mk
source/machdep/socket.c, machutil.src
test/test.com
..............................................................................
3.1.3 GBMV Fixes
New parametrization and bugfixes. Added a guard around a
sequence of assignments in Subroutine RunGBMV1. Removed local
variable R2 and from Subroutine RunGBMVGrid1. Completely removed local
variable RF from Subroutine RunGBMVGrid1.
Files modified:
source/energy/energy.src, gbmv.src, intere.src
/fcm/gbmv.fcm
..............................................................................
3.1.3 Standard coding clean-up
Modified SUBROUTINE GAUSSJ to conform to Charmm coding standards
(fqener.src). PARALLEL code fixes (epath.src). quicksort routine is
moved out of ##IF block for CHEQ (pbeq.src). CHEQ and FASTEW conflict
is terminated with an error message (ewald.src). Some print
statements removed (tsms.src) and variables are initialized (wham.src).
Files modified:
source/cheq/fqener.src
/energy/epath.src
/misc/pbeq.src
/nbonds/ewald.src
/pert/tsms.src, wham.src
..............................................................................
3.1.4 Testcase fixes
fixes for missing parameters (pathint) and to multinode I/O
problems. In molvt6, TOLGrad is set to 0.000002 to attain convergence
on SGI platforms.
Files modified:
test/c26test/pathint.inp
/c27test/gb_mmff94_na-test.inp, gb_mmff94_prot-test.inp
/gb_mmff94_test.inp
/c29test/mctest03.inp, mctest04.inp
/c30test/cftiangl.inp, cftidihe.inp, cftidist.inp, cftmala10.inp,
/cftmtst1.inp, cftmtst2.inp, cftmtst3.inp, cftmtst4.inp
/cftmtst5.inp, molvt6.inp
------------------------------------------------------------------------------
3.2 Developments at University of Kansas
Name : Krzysztof Kuczera
Email Address : kkuczera@ku.edu
Institution : University of Kansas
Date : June 11, 2003
3.2.1. MOLVIB module bug fix and upgrade
Eliminated bugs caused by (a) conversion to FORTRAN77 and (b) hidden
fixed dimension in routine MATNV. Changed PED to allow systems with above 99
coordinates. Modified array IO to simplify running test cases.
Files Modified:
source/molvib/molvco.src, molvib.src, gfdiag.src
/molvut.src, molvio.src, molvsb.src
New test cases:
test/c30test/molvt1.inp, molvt2.inp, molvt3.inp, molvt4.inp,
/molvt5.inp, molvt6.inp, molvt7.inp
..............................................................................
3.2.2. Conformational free energy thermodynamic integration
Test cases and description are expanded for conformational free energy
thermodynamic integration (part of TSM module).
Files Modified:
doc/cfti.doc
Files Removed:
test/c26test/cftigas.inp, cftmgas.inp
New test cases:
test/c30test/cftidist.inp, cftiangl.inp, cftidihe.inp, cftmala10.inp,
/cftmtst1.inp, cftmtst2.inp, cftmtst3.inp, cftmtst4.inp,
/cftmtst5.inp
------------------------------------------------------------------------------
3.3 GAMESS-UK Interface Updated
Name : Paul Sherwood
Email Address : p.sherwood@daresbury.ac.uk
Institution : CLRC Daresbury Lab.
Date : July 1, 2003
Fix for the special case of Gaussian blur with wmain=0.0 (should
implement a point charge but the last version did not).
Changes to GAMESS-UK build procedure for use with CHARMM are
added to install.com. Integer*8 build is needed for compatibility
with some GAMESS-UK builds so this is now supported by install.com
(was previously available for GNUALPHA only). This was partially
implemented in a previous checking but is complete now.
Files Modified:
install.com
build/UNX/guk.mk, Makefile_alpha
source/gukint/gukini.src
/machdep/paral3.src
/gamint/ddi.src
/cadint/cadini.src
/flucq/fluqqmmm.src
/machdep/space.src
/util/util.src, datstr.src
------------------------------------------------------------------------------
3.4 SUN64 port for running on 64-bit Solaris
Name : Bogdan Costescu
Email Address : bogdan.costescu@iwr.uni-heidelberg.de
Institution : IWR, University of Heidelberg
Date : June 6, 2003
Add SUN64 port which allows using 64-bit file access and memory
allocations on 64-bit Solaris. A new keyword (SUN64) is introduced to
allow code changes with respect to the SUN port; SUN64 is to be used
in addition to the SUN keyword. A new host type sun64 is added into
install.com. Usage of the "64" install switch is not supported.
not valid:
When reopening the file for direct access, the record size is
measured in bytes on SUN (and derived) platform, the same as for T3E
and GNU, so the correct size is 512. Add SUN keyword to the "##IF T3E
GNU" line which allows setting the correct record size.
The compiler does not accept INTEGER*1 type; LOGICAL*1 or BYTE can
hold integer values between -128 and 127; however such a variable
cannot be used as loop index. Replaced several arrays typed INTEGER*1
with BYTE. The one loop index (IPass) was replaced with a normal
INTEGER as the type size does not matter in this case. All
modifications are protected by "##IF SUN".
To avoid breaking output format for other ports, printing of
heap-related 64-bit values which do not fit in the existing integer
field width is protected by "##IF INTEGER8".
Files Added:
build/UNX/Makefile_sun64
Files Modified:
install.com
source/machdep/machutil.src, space.src
/solvation/fft.src
/util/util.src
/correl/solana.src
/energy/gbmv.src
------------------------------------------------------------------------------
3.5 Developments at NIC-Slovenia
Names : Milan Hodoscek
Email Address : milan@cmm.ki.si
Institution : National Institutes of Chemstry, Ljubljana, Slovenia
Date : July 7, 2003
Code cleanups and compiler bugs
Some bugfixes for compiler, and minor modifications
Files Modified:
source/charmm/charmm_main.src
/fcm/dimens.fcm
/gukint/gukini.src
/machdep/socket.c
/misc/olap.src
/pert/pert.src
/sccdftbint/scfdftbini.src
/util/array.src
------------------------------------------------------------------------------
3.6 Bugfixes
(1) RXNENE/RXNDEF fixes
Aaron R. Dinner
In rxnene.src, in DECEND, incorrect formulas for derivatives of
direction defined as a cross-product of two vectors and of distance
between a point and a line. In umbrella.inp, addition of the umbrella
potential causes the system to move in a direction away from DEL0
along the reaction coordinate. The formulas are correctly
implemented.
In rxndef.src, DELSTT array is now reinitialized when STATistics are
redefined.
Files involved:
source/rxncor/rxnene.src, rnxdef.src
..............................................................................
(2) MC fixes
Aaron R. Dinner
In mcener.src, incorrectly declared array EB in MCENER is
eliminated. The EEF1 keyword surrounding EEF1EN call in MCENER is
removed.
In moveio.src, incorrect dimension of array IROOTG in MOVERD is
corrected from IROOTG(MBONDT) to IROOTG(MMVTYP).
In enbfast.src, in ENBFS8, loop construct changed so handles
IFRSTA .GT. 1 incorrectly, which breaks MC. ITEMP variable
initialized to INBL(I-1) when IFRSTA .GT. 1.
Files involved:
source/mc/mcener.src, moveio.src
/nbonds/enbfast.src
..............................................................................
(3) SOLANA fixes
Aaron R. Dinner
In solana.src, local arrays VVCG (line 339) and VX (line 1571) are
declared with adjustable parameters. VVCG is declared with static
parameter MAXA and VX allocated on stack in the calling routine and passed.
File involved:
source/correl/solana.src
..............................................................................
(4) STACK usage fix in CORREL
Tibor Rudas(tibi@mdy.univie.ac.at)
In correl/correl.src in subroutine correl space for the SERVAL
array is allocated with SERVAL=ALLSTK(INTEG4(MAXSR)), but later on the
passed stack space is used as REAL*8 SERVAL(*). It is correctly
allocated.
File involved:
source/correl/correl.src
..............................................................................
(5) Preprocessing error fix
Tibor Rudas(tibi@mdy.univie.ac.at)
Some routines in manip/corman2.src need parameters which are only
set in fcm/consta.fcm if either the PATHINT or the PBEQ keywords are
set in pref.dat. If neither is present in pref.dat manip/corman2.src
will not compile. The ##IF keys are removed.
File involved:
source/fcm/consta.fcm
..............................................................................
(6) EEF1 temperature variable fix
Themis Lazaridis (themis@rilke.sci.ccny.cuny.edu)
The temperature default value ZEROC is fixed to ROOMT.
File involved:
source/misc/eef1.src
..............................................................................
(7) RMSD constraint fix
Benoit Roux (benoit.roux@med.cornell.edu)
There were 2 problems with the CONS RMSD. The first one was that
the rmsd energy constraint as called on each node with MPI.
The call to ECNST3 in ecnstr.src is now protected with a "IF(MYNOD.EQ.0)"
to be called only once. This is a quick fix for now. A better
parallelization of the CONS RMSD could be done in the near future.
The second problem was that the passing variables in the call to
ECNST3 had been modified since c30a1, but the changes were apparently
not implemented in ecnstr.src, this is done now.
File involved:
source/energy/ecnstr.scr
..............................................................................
(8) Plasma PME bugfix
Bernard R. Brooks (brbrooks@helix.nih.gov)
EWVIRIAL fixed in the PMEPLASM code.
File involved:
source/nbonds/pme.src
..............................................................................
(9) ATOMID call fix
Bernard R. Brooks (brbrooks@helix.nih.gov)
Atom ID is fixed to get the IDs of the atom J.
File involved:
source/manip/cstran.src
The following are introduced and modified during the development period
of August 15, 2003 through February 15, 2004.
4.1 TSRI Developments
Name : Scott Brozell, Mike Crowley and Charles L. Brooks, III
Email Address : crowley@sripps.edu, brooks@scripps.edu
Institution : The Scripps Research Institute
Date : December 29, 2003
4.1.1 Intel 64 bit Machine Ports
Added support for new machine architecture and compilers. The new
install.com keys and switches are
hpitanium for HP-UX Itaniums
altix for SGI Altix series (64 bit)
gnu EFC for efc intel 64 bit compiler
Roberto Gomperts added mpi8 routines. These convert 8 byte integer
arguments for charmm built with I8 into 4 byte integer arguments for
mpi calls and are guarded by the pref keyword ALTIX_MPI which is
activated by the install.com target altix. This mechanism enables
coexistence with other mpi implementations, such as LAM, that do not
support MPI_INTEGER8.
Modified to build 64 bit on Linux Itaniums using install keyword EFC.
Roberto Gomperts identified some mpi_allreduce calls that were not
passing the correct message lengths for 64 bit integers.
A general solution would be to use MPI_INTEGER8, but that is not
supported by all mpi implementations yet. So the message count is
doubled.
Files Added:
build/UNX/Makefile_altix, Makefile_hpitanium
Files Modified:
install.com
build/UNX/Makefile_gnu, quantum.mk, solvation.mk
source/fcm/impnon.fcm
/machdep/machio.src, machutil.src, paral1.src, paral2.src
/space.src, startup.src
/manip/shake.src
/nbonds/pmeutil.src
..............................................................................
4.1.2 McIntosh OSX Port
Port added for for Macintosh OSX. The install.com target is set to OSX
and the compiler switch is either xlf or g77. The pref.dat keywork is OSX.
install.com osx [ xlf | g77 ]
Files Added:
build/UNX/Makefile_osx
Files Modified:
install.com
build/UNX/misc.mk
source/cadint/cadini.src
/correl/solana.src
/emap/emapio.c
/fcm/impnon.fcm
/gukint/gukini.src
/machdep/cstuff.c, machio.src, machutil.src, paral1.src
/paral2.src, parset.src, space.src, startup.src
/quantum/qmset.src
/util/prpref.src
..............................................................................
4.1.3 Testcase Validation
A full set of output for all tests is placed in test/Testoutput
for comparison and validation purposes. It is intended to be used
with test.com as follows:
test.com gnu G20 Testoutput 20
for the c20test tests to be run, where G20 is the test output
directory. The result is compared with the output in Testoutput,
and the pared-down diffs placed in G20.rpt.
A validation awk script is also submitted, comare.awk, which is
to be used to determine pass/failure and to report the relevant diffs
only.
awk -f compare.awk G30.rpt
will process G20.rpt and report PASS of FAIL based on relevant diffs
only. A more verbose output with the relevant diffs reported is
produced with:
awk -f compare.awk verbose=1 G20.rpt
Some tests were modified to get rid of extraneous stuff.
Files Added:
test/compare.awk
Files Modified:
test/c30test/cq1ab1gbdyn.inp, cq1fsvvacu.inp, cq1fsvwat.inp,
/cqmeohvacu.inp, cqnmawat.inp, cqtip4pbulk.inp
..............................................................................
4.1.4 Bugfixes
Fixes to code for compaq alpha mostly centered on emap code.
Some fixes are submitted here and at least the code compiles now, but
emap still broken.
(1) Added a note that GBMV requires GENBORN prefx name for compilation.
(2) Fixed FASTEW known incompatibility for CHEQ.
(3) Fixed timing bug in dynamics
(4) arg list missing vars in epath.src
(5) Fixed problem with integer*4 and integer*1 declarations in GBMV
(6) Bugfix of problem printing dihe constraints.
Files Modified:
source/dynamc/dcntrl.src, dynamc.src
/emap/emapio.c
/energy/energy.src, epath.src, gbmv.src
/machdep/cstuff.c, socket.c
/manip/cstran.src
/mbond/mbback.src
/mc/mvhmc.src
/nbonds/ewald.src
Modified Documentation:
doc/gbmv.doc, cheq.doc
------------------------------------------------------------------------------
4.2 Bugfixes
4.2.1 CMAP bugfix
Name : Micheal Feig/Alex MacKerell
Email Address : feig@msu.edu/alex@outerbanks.umaryland.edu
Institution : Michigan State/U. Maryland
Date : January 7, 2004
With all trans phi, psi geometry CMAP energies are wrong. This is
due to a machine precision problems. offset values to avoid precision
problem.
File fixed:
soruce/energy/eintern.src
