• Improved memory usage for real space fields.
• Improved portability to Windows.
• Expanded testsuite.
• Numerous bugfixes and cleanups.

• Reimplemented ionic relaxation.
• Implemented first order perturbation theory for local potential.
• Added lattice stresses for planewaves.
• Use BLAS calls for matrix multiplies.
• Added new infrastructure for literate documentation of the commands.
• Moved the development from CVS into darcs.
• Use automake.
• Added a testsuite.
• Various cleanups, bugfixes and optimizations.

• Plane wave pseudopotential and wavelet all-electron calculations in a unified framework.
• Reimplemented gradient corrected functionals (GGA).
• Single compiler (C++).
• ANSI standartization.
• Compiles and runs on MS Windows(TM).
• and more...

• The new command "dump-bands" is an overall action, that dumps the wavefunctions for each band into a separate file (after the electronic minimization if there is one).
• It is possible to specify PAW (Projector Augmented Wave) projectors in the pseudopotential description file. If such projectors are specified, an output file called PROJ will be dumped at the end of the run.
• Various minor bug fixes.
• Created revision tag "cornell" for version 2.1 at the time of moving the repository to Cornell. You can check out the snapshot of the code at this time by specifying the tag "cornell" to the cvs checkout command. The items above are not included in this revision.

• A new command "coords_type" is added to allow for entry of ionic positions in cartersian (real space xyz) coordinates instead of the default lattice coordinates.
• One now has the option of using the in-house supplied jacobi.c matrix diagonalizer. It is slow and non-optimized, but 100% portable and in plain C++. This is useful if you are having problems linking with LAPACK/ESSL (or perhaps the diagonalizations are not a time-sink and you don't care). This allows for complete elimination of the LAPACK/ESSL dependency if needed.
• Some minor bug fixes were performed in various files.

The routines that calculate the symmetries of the bravais lattice had some peculiar errors that showed up when the lattice vector matrix was not symmetric (source file symm.c). The error has been fixed and the source code .tar.gz package has the updated symm.c file.

These web pages are now the official home of DFT++, and contain the documentation for the project. Thanks to all the developers of the software (in alphabetical order): Tomas Arias, Jason Cline, Gabor Csanyi, Ken Esler, Sohrab Ismail-Beigi, Nicolaj Moll, Ickjin Park, Evan Reed, and Tairan Wang. And by a strange coincidence, it is Independence Day in France...