We are happy to announce the release of Siesta 3.1. Users are encouraged to use this version for production runs.
You can find the Siesta 3.1 release in the Siesta web site (please use the http://www.icmab.es/siesta/ entry point if www.uam.es does not work). You can access it as usual, in the downloads page that you find immediately after logging in, using the username and password obtained with your license. To report any remaining issues, see Docs/REPORTING_BUGS Thanks for your patience and continuous support. The Siesta team August 26, 2011 ================================================================ Release Notes for SIESTA-3.1, August 26, 2011 GENERAL NOTES Users are encouraged to use SIESTA-3.1 for production runs. To report any remaining issues, see Docs/REPORTING_BUGS MOST RELEVANT CHANGES SINCE SIESTA-3.0-rc2 This is only a subset of all the changes. A complete list can be found in the Docs/CHANGES file. BUG FIXES AND IMPROVEMENTS * Fix setting of applied electric field in parallel Due to a programming error, the user-specified electric field was not correctly set in non-master nodes. * Fix mixing bug in TranSiesta In TranSiesta mode, the program first might carry out a calculation at zero bias and with periodic boundary conditions, just as a normal Siesta calculation. Then the program goes into "transiesta mode" proper, during which a new mixing cycle takes place. In late June 2010, a change in the structure of the mixing code introduced a bug: the counter holding the number of history records was not properly initialized upon the ending of the preliminary siesta-style calculation. The mixing counter is now properly reset. * Improved grid for re-parametrization of pseudopotentials When the option Reparametrize.Pseudos is in effect, the program resamples the pseudopotential data into a new logarithmic grid with parameters chosen so that the grid-point separation at the typical rc's for orbitals is small enough (of the order of 0.01 bohr). The previous choice of parameters resulted in grid-point separations of the order of 0.005 bohr near the origin, which apparently is not fine enough for s orbitals, leading to oscillations in the gradient near the origin. A new set of default parameters has been chosen to provide a finer grid spacing (1.e-5 bohr) near the origin and 0.01 bohr around r=10 bohr. * Some cleanup and new options in O(N) solver code The O(N) solver code is being cleaned up, and more options are being added, but should still be used with care and for systems for which it is appropriate. COSMETIC CHANGES * Avoid use of generic functions in FoX declaration statements The use of generic functions in declaration statements was confusing the Intel Fortran compiler. They have been replaced by calls to the specific versions. Note that the FoX interface still seems to expose a number of bugs and weaknesses in some compilers. In some cases it might be necessary to deactivate the FoX library in the Siesta building process. To do so, insert the line DUMMY_FOX= --enable-dummy in your arch.make file, and recompile Siesta (i.e, "make clean; make"). * Modify bonds routine to print real locations In routine bonds (called at the beginning and at the end of the program) we now print the real location of the neighbor atom in space, and not, as in earlier versions, the location of the equivalent representative in the unit cell. This information appears in the files SystemLabel.BONDS and SystemLabel.BONDS_FINAL. * Avoid reporting unstable inversion in pulayx Typically, inversion problems in the DIIS procedure for mixing of the density matrix occur when the error estimate for a step is very small compared to the previous ones. This is harmless, and the reporting of a failure in the output file is confusing. In case it is needed, a hook to allow a full diagnosis can be compiled in with the preprocessor flag -DDEBUG_PULAY_INVERSE * New organization of TranSiesta-TBTrans tests The TranSiesta-TBTrans tests have been consolidated in a single directory under Tests, and provided with streamlined execution scripts.
