ACES II uses the coupled-cluster approach to handle electron correlations in atoms and molecules. This approach sums entire select classes of Feynman diagrams in many body theory to provide a good approximation of the correlation energy. It has been used successfully on small molecules in computing ground state electronic properties and excited states.

As of Nov. 10, 2009 there are two similar versions of ACES II on the University of Kentucky IBM Supercomputer. On the BCX machine we have ACESS-II MAB (Mainz-Austin-Budapest-Version):

and on the P-series machine we have the Florida version:

This has to do with compatibility issues of the software and the differing architectures of the machines. Currently there are now two relatively independent groups issuing differing versions of ACES. The MAB group has recently released their new CFOUR version and the Florida group has formed a company:

which has recently released ACES III. This version is a completely rewritten ACES II that has been designed to scale up well using the latest technology.



The development of ACES II began in early 1990 in the group of Professor Rodney J. Bartlett at the Quantum Theory Project (QTP) of the University of Florida in Gainesville. There, the need for more efficient codes had been realized and the idea of writing an entirely new program package emerged. During 1990 and 1991 John F. Stanton, Jürgen Gauss, and John D. Watts, all of them at that time postdoctoral researchers in the Bartlett group, supported by a few students, wrote the backbone of what is now known as the ACES II program package. The only parts which have been taken from others were the integral packages, where the MOLECULE package of J. Almlöf, the VPROP package of P.R. Taylor, and the integral derivative package ABACUS of T. Helgaker, P. Jorgensen J. Olsen, and H.J. Aa. Jensen had been adapted for the use within ACES II and had been modified for that purpose.

The first version of ACES II had been written for CRAY supercomputers (under the unix-based operating system UNICOS) and, consequently, a lot of effort had been devoted to the exploitation of matrix-vector operations through optimized Blas-routines. However, more or less simultaneously, versions for so-called "unix-boxes" were created. The original design strategy of keeping machine-dependent code to an absolute minimum facilitated this process considerably.

In the following years, the development of new techniques was for a large part continued in Karlsruhe (1991 - 1995, e.g., development of schemes for calculating NMR shifts), Austin (since 1993, e.g., development of gradient techniques for EOM methods), Mainz (since 1995, development of analytic CC second derivatives) and Budapest (since 1997) besides ongoing work in Gainesville. The code was also ported to unix-based workstations (first IBM AIX workstations and HP, since 1996 also on DEC alpha and SGI), before at the end of the nineties LINUX versions (for x86-based PCs) have been created. As the last merge between the original Florida version of ACES II and the version maintained in Austin and Karlsruhe and later in Mainz dates back to roughly 1995, it has been decided that both versions are now separately maintained and distributed (for the Florida version of ACES II.

History taken from: Visit site for more information.

Running an ACES II Job

The decision as to whether to run ACES II or ACES II - MAB depends on your analytical requirements and system performance. These factors will have to be determined by your own experience with running the programs. The manuals for the two programs available on the two websites are very good. The reader is referred to those in creating the input files. The batch command for the BCX, ACES II-MAB is:

batchaces-mab [-x] [KEY=some_file]

where the -x option will not delete the scratch disk.

The job will automatically create the scratch space needed, copy files and delete the scratch when done.

The batch command for the P-Series ACES II is:

batchaces [KEY=some_file]

The operation is as before except this version does not have the -x option. Typing the name of either script with no arguments will output help information on the script.