| Site Navigation |
| UKCP Home Page | |
| The CASTEP Code | |
| Technical Support | |
| Organisation | |
| UKCP and Accelrys | |
| Useful Links | |
| Conference Calendar | |
| Job Opportunities | |
| Further Information | |
| Publications |
| Related Sites |
 |
Accelrys |
|
|
|||||||||||||||
|
First-principles simulationCalculating the properties of materials fron scratchFirst-principles calculations involve no adjustable parameters, are based on an explicit quantum treatment of the electrons in a model system, which means solving Schroedinger's equation to find the electronic ground state There are still approximations in the solution, but these are of an extremely general kind. The calculations yield the total energy of assemblies of atoms, and forces on the atoms. This knowledge enables us to calculate a vast range of properties with near chemical accuracy. Density Functional Theory (DFT) is a powerful, general theory relating the total energy of a system of interacting electrons in an external potential to the electron density. Almost all our calculations employ DFT to describe the valence electrons. The expansion in the scope of such calculations has been breathtaking in the last fifteen years or so, and they are now routinely used to study fundamental aspects of the physics and chemistry of condensed matter in every conceivable guise, ranging from catalysts to protiens, minerals to liquid metals. The great importance of DFT and its application in simulation was recognised by the award of the 1998 Nobel Prize in chemistry to Walter Kohn and John Pople. Car-Parrinello MethodsThe bedrock of UKCP's science is the DFT plane-wave pseudopotential method, pioneered in the 80's by Roberto Car and Michele Parrinello. This approach has transformed the first-principles landscape, making possible calculations far beyond previous approaches. Throughout this time the UK has played a leading role in the development and application of the methods, an effort founded on Mike Payne's work with the CASTEP code. The methods rely on a plane-wave basis, pseudopotentials and the use of density-functional theory to describe the valence electrons in a model system. Other ingredients include fast-Fourier transforms and minimization of the total energy rather than matrix diagonalization. The main approximationsIn Car-Parrinello calculations there are two main approximations. The first is the way in which the exchange-and-correlation energy of the electrons is handled. DFT provides a very general theory involving functionals of the electron density, but does not say what these functionals are, only that they must exist. In practise it has been found that the Local-Density Approximation (LDA), sometimes extended to take account of density gradients (the Generalised Gradient Approximation or GGA) are remarkably accurate, despite their simplicity. The LDA has been in widespread use for well over 20 years, and is known from vast numbers of calculations to be very reliable for large classes of systems. The second approximation is the pseudopotential approximation. This is basically a way of avoiding the need to include all the electrons in the calculations. Only the valence electrons are treated explicitly. The core electrons are assumed to be in exactly the same state as in the free atom. The effective interaction between the valence electrons and the ion cores is described by a pseudopotential for each kind of atom. This pseudopotential is generated by first-principles calculations on free atoms. In the early days of the pseudopotential method, only a limited range of chemical elements could easily be handled. This ceased to be true a long time ago, and nowadays the method is routinely used for transition metals and first-row elements, which used to be regarded as `difficult'. What you can calculateThe total energy and forces supplied by first-principles calculations can be used to do many things. One of the simplest and most important is the determination of equilibrium structures. The LDA-pseudopotential method was already in routine use for studying crystal structures ten years ago. Now, it is widely used for more complex problems, such as the equilibrium structure of point defects, grain boundaries, surfaces and molecules at surfaces. Closely related to the treatment of equilibrium structures is the calculation of vibrational properties, which is also widely practiced. We also use first-principles calculations for dynamical simulations, where the forces on the atoms are used to generate the time evolution of the atomic positions. This is the idea that has long been used in classical molecular dynamics simulation, but now the forces are calculated from first principles, instead of being represented by empirical models. This means that dynamical processes involving the making and breaking of chemical bonds can be studied. Dynamical simulations allow the study of solids and liquids in thermal equilibrium, through the calculation of quantities such as thermodynamic functions, radial distribution functions, diffusion coefficients and dynamical structure factors. Even the first-principles calculation of phase diagrams is becoming possible. Another use of dynamical simulation is for the study of non-equilibrium processes like the dissociative chemisorption of molecules at surfaces. Technical Description and HistoryBeginningsThe Cambridge Serial Total Energy Package was developed in the mid-'80s by Mike Payne. During that time and subsequently, many people in numerous research groups contributed to the code. CASTEP calculates the total energy, forces and stresses in a 3D-periodic system. The review article Rev. Mod. Phys. 64 (1992) 1045 sets out in detail the theory and technical implementation of CASTEP. By the end of the '80s it was clear the recipe of Density-Functional Theory (DFT), plane-waves, pseudopotentials and Fast-Fourier Transforms (FFT's) was revolutionising first-principles calculations and endowing them with unprecedented power. CETEP: Meeting the Grand ChallengesUKCP was formed in 1990 with the aim of exploiting parallel supercomputers for first-principles simulation. It was recognised that realistic and accurate simulations would inherently be on a grand scale, and that parallel machines were essential. Out of this came CETEP, a highly efficient parallel version of CASTEP in which the real- and reciprocal-space grids were decomposed and distributed among the nodes of the parallel machine. UKCP's ability to exploit parallel computers has consistently enabled us to tackle and solve the most difficult and demanding of problems - real Grand Challenges. This relies heavily on the UK's investment in supercomputing, from the Meiko Computing Surface in the early days through to the latest CSAR service.CASTEP NowSince 1993 Accelrys Ltd (formely MSI) have developed and consolidated CASTEP as a commercial product, while CETEP remained the workhorse of UKCP academics. However, within the new UKCP-Accelys agreement these strengths are combined. CASTEP is now a state-of-the-art plane-wave code, incorporating all the latest and best technology. Once parallel development is complete it will derive the benefits of several years' of CETEP parallel experience. CASTEP 4.2 Key Features
DistributionThe UKCP-Accelrys agreement secures the right to the CASTEP source code for all UK academics. UKCP believes this is a remarkable benefit, offering a high-quality code which is the product of many years' effort by academic researchers and software professionals. Even more important though is the opportunity for UK researchers to contribute to the continued development and enhancement of CASTEP. Welcome to CASTEP and UKCP! Obtaining the codeDaresbury Laboratory distributes CASTEP to UK academics. To obtain
the code you will be required to sign a license agreement and provide some details
about yourself and your institution. After we receive these the
software will be sent to you. Our address is: Academic release 4.2 is now available for distribution. Please write to the above address, clearly marking your letter or e_mail "UKCP/CASTEP", to obtain the requisite forms. Once we receive these completed and in order the software will be posted to you. Please allow reasonable time for delivery! Please note: at present this agreement covers academic researchers in the UK only, and we regret that we cannot distribute CASTEP to researchers in other countries. Electronic versions of the CASTEP paperwork are also available via e-mail.The Accelrys Cerius 2 interfaceAccelrys provide a full graphical user interface to CASTEP as part of the
Cerius 2 package. However, this software is not covered by the UKCP-Accelrys agreement,
and must be bought from Accelrys. UKCP has, though,
agreed special rates with Accelrys
for a Cerius 2 CASTEP package available to UK academics.
Accelrys's contact
details are as follows. |
||||||||||||||||
|
||||||||||||||||