From case@scripps.edu Wed Nov 3 10:47:00 1993 Received: from scripps.edu for case@scripps.edu by www.ccl.net (8.6.1/930601.1506) id KAA16823; Wed, 3 Nov 1993 10:42:29 -0500 Received: from lanczos.scripps.edu by scripps.edu (5.61/1.34) id AA13691; Wed, 3 Nov 93 07:42:26 -0800 Received: by lanczos.Scripps.EDU (4.1/SMI-4.1) id AA11472; Wed, 3 Nov 93 07:41:35 PST Date: Wed, 3 Nov 93 07:41:35 PST From: case@scripps.edu (David Case) Message-Id: <9311031541.AA11472@lanczos.Scripps.EDU> To: chemistry@ccl.net Subject: Re: Comparison of current DFT codes? Status: R > > At this moment, I'm trying to make a decision on which code to use to > investigate the effects of conformational energetics on redox and > electronic spectroscopic properties of a series of polynuclear Co(III) > complexes that we have synthesised in our labs. We have been please with the "ADF" (Amsterdam Density Functional) codes in our calculations on polynuclear iron-sulfur, iron-oxo and manganese-oxo clusters. For an overview of this work, see Adv. Inorg. Chem. 38:423-470 (1992). ADF is supplied in source code format, so you can examine the details of the calculations and make modifications if you wish. ADF uses Slater type orbital basis sets, which appear to do a good job, and the overall approach has been widely tested -- see e.g. the review by Tom Ziegler, Chem. Rev. 91:651-667 (1991). The ADF programs can place some severe constraints on your cpu and available disk space. We have also made considerable use of DGauss for transition-metal complexes, and are generally pleased with it, although we have not yet figured out how to carry out broken symmetry calculations on multi-metal systems with it. DGauss is a part of a package called Unichem, which is distributed by Cray Research and by MSI. It uses Gaussian basis sets. There is a consortium which helps to support its development; members of the consortium receive updates about a year or so in advance of other users. As far as I know, it is not possible to a source code license for DGauss. Our experience with DMol on fairly large clusters (three dozen atoms or so with two transition metals) has been mixed, with troubles arising in obtaining self-consistency. But we have never run a side-by-side test to see whether the problems are intrisince to these molecules or reflect somehow on the code. DMol uses partially numerical basis sets. It is difficult to compare the three codes because of the difference in philosophy, accuracy requirements, and basis sets. ...dave case ===================================================================== David A. Case | internet: case@scripps.edu Dept. of Molecular Biology, MB1 | bitnet: case%scripps.edu@sdsc The Scripps Research Institute | fax: 619-554-6717 10666 N. Torrey Pines Rd. | phone: 619-554-9768 La Jolla CA 92037 USA | =====================================================================