Skip to content

Commit

Permalink
Check into development branch
Browse files Browse the repository at this point in the history
  • Loading branch information
twindus committed May 24, 2004
1 parent dcaa2c2 commit 713c033
Showing 1 changed file with 189 additions and 0 deletions.
189 changes: 189 additions & 0 deletions release.notes.4.6
Original file line number Diff line number Diff line change
@@ -0,0 +1,189 @@
NWChem Version 4.6 Release Notes
(Also available at
http://www.emsl.pnl.gov/docs/nwchem/release-notes/release.notes.4.6.html)

NOTE: These release notes are not a substitute for reading the User Manual!
They are meant to give a thumbnail sketch of the new capabilities and bug
fixes that are available in NWChem Version 4.6. When there is a conflict
between the release notes and the User Manual, the User Manual takes
precedence.

The changes for version 4.6 are listed at the bottom of this file.

Several new major capabilities are available in NWChem, including several new
modules. The new modules are:

o CCSD(T) and CCSD[T] for closed- and open-shell systems with Abelian symmetry
(contributed by Alex Auer, University of Waterloo)
o EOM-CCSD, EOM-CCSDT, EOM-CCSDTQ for excitation energies, transition
moments, and excited-state dipole moments of closed- and open-shell
systems
o CCSD, CCSDT, CCSDTQ for dipole moments of closed- and open-shell
systems
o Constraints module that is capable of freezing atom positions and
restraining bond lengths during optimization.

For each of these capabilities, the user is referred to the User
Manual for more information.

Additional major additions to the capabilities include:
o van Leeuwen-Baerends potential for DFT and TDDFT
o Casida-Salahub asymptotic correction for DFT and TDDFT
o A new self-contained asymptotic correction for B3LYP DFT and TDDFT
o A new density functional TPSS, energy term only
o Added the functional HCTH/407+
o A new keyword in TCE to allow/suppress recomputing the Fock
matrix. This is useful for one-component relativistic CI/CC/MBPT/EOM-CC
o Two new I/O algorithms (DRA and GA+EAF) in TCE
o QHOP proton hopping method in MD
o Charmm force field (Urey-Bradley)

Again, for each of these capabilities, the user is referred to the User Manual
for more information.

Listed below are the other major and many minor changes for each module
with significant changes. These descriptions are somewhat terse and more
information is generally available in the User Manual.


Basis:
o BUG FIX: Iodine Stuttgart Large Core ECP to conform to the one
published by Bergner et. al. (Mol. Phys. 1993)
o Definition of PERMANENT_DEF_DIR, SCRATCH_DEF_DIR
NWCHEM_BASIS_LIBRARY_PATH at compile time is gone. These
environmental variables will be ignored. Compilation automatically
stops if the NWCHEM_TOP name is longer than 65 chars.

BLAS:
o Introduce faster dgemm (with more aggressive blocking)

CCSD(T):
o Improved memory allocation
o BUG FIX: TOL2E convergence parameter is now properly handled

COSMO:
o BUG FIX: Mass was used for radius in ECP calculations.

DFT:
o Better handling of exponential underflow speeds up certain
platforms.
o Implemented GA mirroring to address network latency issues
(activated by set dft:xcreplicated t)
o Speed improvement by introducing more aggressive screening for
basis functions evaluations and grid partitioning weights
o Added pstat profiling
o In-core coulomb algorithm: improved output
o XC grid restructuring for getting fault-tolerance in gradients;
added "grid nodisk" keyword for skipping I/O operation for grid
o Increased limit for grid size
o BUG FIX: The final vectors are now printed when requested
o BUG FIX: Bq atoms with different charges are no longer considered to
be different atom types
o BUG FIX: Fix associated with VWN functional
o BUG FIX: d4d group was causing loss of weights normalization


Geometry:
o BUG FIX: geometries containing bqs were causing the autosym
module to crash
o BUG FIX: autosym was not correctly detecting the degeneracy of the
moments of inertia

Hessian:
o BUG FIX: Fixes associated with linear dependencies.
o CPHF SCF residual error termination is now set to 1d-2 (was 1d-4)


Input:
o Default memory increased from 200 mb to 400 mb

Integrals:
o Performances improvements in the McMurchie-Davidson algorithm
o BUG FIX: increased limpair value to address "getmem memory
overflow" problem

MD:
o Molecule-based periodic boundary conditions
o Revised file formats for frg, sgm and top files
o Radius of gyration and diffusion coefficient are optional properties
o Reduced memory allocation for work arrays
o Prefetching using non-blocking communication
o Option to aid in membrane setup
o Trajectories recorded in multiple files
o Redesign cell-cell list to avoid most common 'lbbl errors'
o Multiple parameter databases
o Mixed parameter definitions
o Improved performance in prepare module for very large systems

MP2:
o Added scratchdisk keyword to limit disk usage on a per process base

NBO:
o BUG FIX: title writing turned off since requires risky character
to integer conversion that breaks many compilers
o BUG FIX: Intel compiler 8.0 was crashing on Hollerith characters

Numerical gradient:
o BUG FIX: atoms list was corrupted when symmetry is on

NWPW:
o Pseudoptential library: NWCHEM_NWPW_LIBRARY_PATH is gone in a
similar fashion to what happened to Gaussian basis
library. Compilation automatically stops if the NWCHEM_TOP name is
longer than 65 chars.
o BUG FIX: ELF generation in pspw_dplot fixed.
o Mulliken projection now uses pseudoatomic orbitals
o Automatic grid conversion
o Preliminary version of HGH pseudopotentials
o PSPW: virtual orbital calculations
o Car-Parrinello: geometry constraints
o BAND: Monkhorst-Pack k-point generation
o BAND: LMBFGS minimizer
o PAW: free space boundary conditions
o Performances improvements in the non-local pseudopotential
and orthogonalility calculations

Properties:
o Ecce support added for NMR properties


QMD:
o Released restriction on theory type that can be used in the computation.

Rtdb:
o Added option to store rtdb in memory (-DDB_INMEM). This option
creates a dummy .db file, therefore restart will not work.

SELCI:
o Now runs on 64 bit platforms

SCF:
o BUG FIX: Linear dependency is accounted for in vector projections
o BUG FIX: vectors "lock" was causing hang-ups in certain parallel
runs.
o Got rid of hard-wired limit of 10,000 basis functions.

TDDFT:
o BUG FIX: Fixed S**2 expectation value when frozen core is used

Util:
o BUG FIX: made the interface to LSF to estimate wall-time
remaining more fault-tolerant
o I/O on system with inhomogeneous disk sizes is made more robust

New Platform added:
o Mac OSX 10.3 with GNU (gcc and g77 3.4) and IBM (xlf) compilers
o Cray X1
o Opteron Linux with GNU, Portland Group and PathScale compilers.
o Linux ppc64 with IBM compilers
o INTERIX platform using the Windows Services For Unix
(http://www.microsoft.com/windows/sfu/default.asp)

New compiler support:
o Intel fortran compiler version 8.0 for IA32 and IA64
o Use of g77 version 3.3 (3.3.1, 3.3.2 and 3.3.3) is supported, but
not recommended. We had to put in place version fix for this
compiler (e.g. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13037)
g77 3.1.x 3.2.x and 3.4 are known to work


0 comments on commit 713c033

Please sign in to comment.