The Materials Computation Center, University of Illinois

The Materials Computation Center, University of IllinoisDavid M. Ceperley, University of Illinois at Urbana Champaign, DMR 0325939

The Travel Award Program offers financial support to junior, U.S.-based scientists travel to European workshops and conferences on computational physics and materials science.

European partners are:

• Centre Européen de Calcul Atomique et Moléculaire (CECAM): an organization devoted to the application of advanced computational methods to problems in frontier areas of science and technology, and

• Psi-k: a charity devoted to building cooperation in the field of electronic structure calculations.

Program chair: David Ceperley (Physics, Illinois)Website: http://www.mcc.illinois.edu/travel/

Travel Award Program for Young ScientistsProviding specialized, forefront training

http://www.mcc.illinois.edu/travel/





Software ArchiveWidely available codes for computational physics and materials science


The Software Archive is a website containing ~40 research and education codes.

To date, the Archive has delivered more than 10,000 downloads to 3,150 users. Summer School codes are popular downloads.

Although most codes have been online for 6+ years (a long time in software development), all have been requested and downloaded within the past eight months. Some are downloaded monthly.

A few software projects, such as SIESTA and QMCPack, have homepages located externally to MCC, but continue to be accessed via the MCC Archive. For example, between 2005-2010 SIESTA (a Density functional Theory code) has been downloaded 230 times from the Software Archive, and QMCPACK (a Quantum Monte Carlo package) has been downloaded 320 times.

The current trend for this type of software is to develop and distribute online, which is a different model from the Software Archive. However, the Software Archive filled a need during its lifetime.

http://www.mcc.illinois.edu/software/

http://www.mcc.illinois.edu/software/


Software Archive: Download statistics

Year Downloads

2004 3122

2005 1604

2006 1277

2007 1472

2008 1231

2009 936

2010 460

Total 10102

Most popular software titles Total downloadsTBPWAn electronic structure code primarily intended for pedagogical purposes

990

Kinetic Monte-Carlo Simulation KMC (Duane Johnson)Kinetic Monte-Carlo Simulation (KMC) with irradiation

730

Blair Tuttle Summer School LabElectronic Structure of atoms, molecules and solids using VASP

617

Material Information and Design LaboratoryWindows-based software for electronic structure calculations

612

TDDFTTime-dependent density functional theory

474

Richard Martin Summer School LabElectron bands in crystals: calculations in a plane wave basis with empirical or model potentials

434

MP-DFTMultiple Parallel Density Functional Theory

415

Duane Johnson Summer School LabSummer School Lab: Kinetic Monte Carlo

407

OHMMSOHMMS is an object-oriented framework for Multiscale Materials Simulation

383

DataSpork 1.1 -- Full versionAnalysis toolkit for performing statistical analysis of computer simulation data

326

QMCPACKQMC simulation code

318


Every year, the selection committee chooses participants to achieve variety and balance of:• Students, postdocs, and faculty• Experience in chemistry, physics, computer science, and materials science• Domestic and international applicants• Small and large schools

Diverse participants

“I also used this as an opportunity to gauge myself and improve based on my peers.”

Schools have covered quantum chemistry, computational geophysics, nano-technology, and biophysics.

The series has created ~150 hours and 3,000 pages of training materials: all quickly and freely available over the Internet.

Networking opportunities

Accessible materials

Cutting-edge topics

“As a smaller university, UMBC simply cannot always provide graduate-level courses in theory, and therefore the 2006 summer school was just needed.” –Dr. Susan K. Gregurick, Assistant Professor University of Maryland, Baltimore County, Chemistry

Materials created specifically for schools and workshopsinclude the Structural Database of properties of materials and an archive of talks from 20 years’ of electronic structure workshops.

Summer Schools on Computational Materials ScienceImmediate and long-term benefits for young scientists

Poster presenter Floyd Fayton, Jr. (Howard U.) discusses his research on “Prediction of Excited States for Carbon, Nitrogen and Oxygen systems using QMC Methods” with associate professor Leonardo Viana (U. Federal de Alagoas).

“The videos look great! People in my group have already started watching them, as well as some from previous years.” – 2007 Summer School Participant


2009-2010 Summer Schools and Workshops

3rd World University Network SPIN International Conference on Spintronics Materials and TechnologyJune 21-25, 2010, Co-chairs: J-P Leburton and D. Johnson• Two-and-half day meeting held at the University of Illinois.• Attended by more than 80 people with invited papers from US, Canada, Europe and Asia, involved in spintronics topics from dilute

magnetic semiconductors to spin torque and single spin confined systems. • Co-sponsors: Materials Computation Center, Beckman Institute, IEEE-NTC and WUN.

2010 Tutorial on Electronic-Structure Calculations for Spintronic-related Materials June 26-27, 2010, Chair: D. Johnson• One-and-half-day of lectures and hands-on computing tutorials on KKR multiple-scattering theory, electronic-structure applications to elements and alloys (disordered, partially ordered, and fully ordered) and critical point and planar defects.

Workshop on Recent Developments in Electronic Structure MethodsJune 2010, UT-Austin, Chair: J. Chelikowsky• Contributed $5,000 towards travel support for lecturers.

2010 Nano-Biophotonics Summer SchoolMay 24–June 4, 2010, NCSA-Illinois, Chair: G. Popescu et al• Contributed $5,000 towards travel support.

Biennial African School Series on Electronic Structure Methods and Applications Jul 19-30, 2010, National Institute For Theoretical Physics, Muizenberg, Cape Town South Africa. Chair: N. Chetty.• School includes pedagogical presentations of the theoretical underpinnings of density functional theory and associated algorithms as well as general solid state physics - and especially current challenges in solid state physics• Includes hands-on computational sessions.• A major objective of this workshop series is to enhance scientific collaboration and networking in Africa. • Contributed travel support for co-chair R. Martin.


This work presents the concept of using nanoscale tubes, or synthetic atoms, and the quantum effects which govern their behavior, to store energy and act as digital quantum batteries.

Preliminary experiments suggest that the amount of energy stored in arrays of synthetic atoms can be even higher than in chemical batteries (such as lithium batteries).

Other advantages are:•Charge / discharge times are in the femto-second range, in contrast to minutes and hours for chemical batteries;•Synthetic atoms have undiminished energy storage efficiency between absolute zero temperature and 500oC, whereas chemical batteries have a 50-degree temperature range;•Nanovacuum tubes can be made from lower-toxicity materials (for example, iron instead of lithium).

The devices can probably be built with standard lithographic techniques, but there is a number of open questions to be investigated, including a theory of vacuum break down in small gaps, vacuum leaks of nano cavities under load, and the efficiency of self-healing processes.

PublicationsA. Huebler and O. Osuagwu, "Digital quantum batteries: Energy and information storage in nanovacuum tube arrays", Complexity (2010).

Digital Quantum BatteriesFaculty: Alfred W. Hübler; Student: Onyeama Osuagwu

Above: A nanovacuum tube, or synthetic atom. Energy would be stored in the blue conical region between the yellow (gold or tungsten) plate (the negative electrode) and tungsten tip or carbon nanotube (the positive electrode ).

Above: Array of synthetic atoms.


When metal is slowly bent it responds with (failure) slips similar to earthquakes. These slips can be detected as acoustic emission. The jumps have a similar size distribution as earthquakes. We use dislocation dynamics simulations to model slips in these systems. The jump size distribution in the model agrees with experiments and our analytical mean field theory. We also calculate predictions for experiments, such as the depinning exponent β. It is the critical exponent which governs the relationship between the average strain rate and the stress in the depinned phase where the material is deforming continually. It is found to agree with our analytic predictions from mean field theory. Since our model accounts for time explicitly, we are able to extract predictions for the avalanche durations and calculate the power spectra of the dislocation activity below the failure stress. The power spectra exhibit power law behavior. Finally, we use a Phase Field Crystal model to study shear rate effects.

Modeling slips during plastic deformation of sheared metalsFaculty: Karin Dahmen and Nigel Goldenfeld. Student: Georgios Tsekenis

Figures: Acoustic emission signals (top) are modeled and analyzed to understand the statistics of slip avalanches in sheared metals. The mean strainrate as a function of distance to the failure stress is shown in the bottom figure.

Significant results: Our simulations and experiments confirm that the analytical mean field theory correctly predicts slip avalanche statistics in sheared crystals. They also show that the statistics of slips in metals in the lab and of earthquakes on tectonic scales are remarkably similar.


MotivationA bottleneck in many electronic structure simulations is the computation of block diagonal elements of the inverse of a given matrix . There are several approaches to solving this problem, such as direct sparse methods, iterative methods, or a stochastic approach, as in the case of our proposed algorithm. Namely, we investigate use of Monte Carlo sampling of the scattering amplitude in an effort to accurately approximate the trace.

A Monte Carlo Approach to Computing the Trace of A-1 Using Scattering Amplitude Theory with Applications to Electronic Structure CalculationsFaculty: Luke Olson and Duane Johnson; Students: Elena Caraba, Suffian Khan

Figure 1: The relative error for different number of samples as the total number of matvecs are predetermined by a fixed tolerance and maximum number of iterations.

ResultsOne approach is to use the scattering amplitude algorithm to compute the trace of the inverse of a matrix, by calculating the individual diagonal entries on the inverse of the matrix and then summing them up. The most obvious choice for b and c would be the column vectors of the identity, since , however this means using the algorithm n times, where n is the number of columns in the matrix.

Instead, we employ a Monte Carlo approach wherein the trace is computed by summing up only a sample of the diagonal elements of the inverse. Although less accurate, this approach is more effective in terms of accuracy per computational cost, as it can be seen from Figure 1. This scattering amplitude algorithm is easily parallelized by assigning several blocks to each processor. Thus, relative to the errors incurred in the BiCG and for small blocks (e.g. m = 16), the trace is computed exactly (via elementary vectors). Publications

•in preparation


We have carried out a quantum Monte Carlo (QMC) study of hydrogen adsorption on Ti-ethylene molecular systems as representation of systems that may be relevant for practical hydrogen storage. The requirements set by the US Department of Energy are at least 6% be weight reversible storage at temperatures ~ 20-50 °C. Previous work using density functional theory has indicated the possibility of meeting the goals; the purpose of this work is to use more exact QMC methods to establish more definitively the relevant energies and numbers of H2 molecules that can be reversibly absorbed. ResultsThe QMC calculations have demonstrated adsorption of at least 3 H2 molecules on one TiH2C2H4 (meeting the weight criterion) at binding energies ~ 0.3 eV, the right magnitude for the need temperature range. In the figures are shown typical molecules with adsorbed H2, and the energies for successive adsorption (in milli-Hartrees mHa, 10 mHa = 0.27 eV). The lowest energy is always a non-magnetic singlet state; the spin 1 triplet state is higher very singlet and triplet This system is proposed to be representative of larger carbon-transition-metal systems that may be more practical for hydrogen storage. This work required many steps that were successfully done: demonstration that the methods are accurate for very weak boning of H2 to benzene.Demonstration that the structures and the transition element d state are well described in the atom and TiH2 molecule; and finally the work on H2-Ti-ethylene systems.

EducationGraduate student Todd Beaudet has completed his thesis in September, 2010. He has mastered many computation methods and has contributed to development of QMCPack and gained much experience with chemical and physical theory and computation.

Studies of systems for hydrogen storageFaculty: Richard M. Martin and Jeongnim Kim; Student: Todd Beaudet

Work in progressThis work has been completed recently and the next steps are to write papers on the results. Possible future work is to consider other molecules and to search for cases where magnetic fields could induce adsorption/desorption by coupling t the transition metal spins.

Publications•T. D. Beaudet, M. Casula, J. Kim, S. Sorella, and R. M. Martin, J. Chem. Phys.129, 164711 (2008)


We developed a spacetime finite-element method for concurrent coupling of continuum and atomistic models at zero temperature that virtually eliminates spurious reflections at coupling interfaces [1]. Extension to finite temperatures requires that unresolved atomistic modes be represented as thermal energy in continuum domains. A non-Fourier (e.g., Maxwell-Cattaneo-Vernotte) conduction model with finite wave speeds is required to properly model conduction at atomistic scales.

We developed a spacetime discontinuous Galerkin (SDG) finite-element method for fully coupled thermoelastodynamics to meet this need, based on 3-field elastodynamic and hyperbolic conduction models [2,3].

Unique Results• New SDG code for thermoelastodynamics with optimal hp+1 convergence.•Simulation of thermoelastodynamic response in a periodic perforated domain.•Coupling with finite temperature atomistic model is next step.

Thermoelastodynamics with Hyperbolic ConductionRobert Haber and Duane D. Johnson, (former) Student: Scott Miller

SDG solution sampled at two times for coupled thermoelastodynamic response in a perforated periodic medium loaded by a sudden, prescribed-temperature boundary condition along the top edge. Color encodes temperature and height encodes magnitude of material velocity. Thermal wave speed is 10 times the mechanical dilatational wave speed. Note reflections and back-scattering by the perforations. Leading thermal wavefront becomes less sharp over time due to diffusive effects.

Publications1. Kraczek et al., "Adaptive spacetime method using Riemann jump conditions for coupled atomistic-continuum dynamics,” J. Comp. Phys. (2010)2. Miller et al., "Multifield spacetime discontinuous Galerkin methods for linearized elastodynamics,” Comp. Meth. Appl. Mechs. Engnrg. (2009)3. Miller et al., “Spacetime discontinuous Galerkin method for hyperbolic heat conduction,” Comp. Meth. Appl. Mechs. Engnrg. (2008)


Using concepts from quantum cluster methods, we derived a cluster mean-field theory (MFT) to predict quickly and accurately solid-solid phase transitions using general cluster-lattice Fourier transforms with clusters of size Nc and a lattice coarse-grained into cells of size Ncell. For Ncell = Nc we reproduce the dynamical cluster approximation (DCA) applied to classical phase transition. However, for Ncell >> Nc we provide a new method that is significantly faster computationally and more accurate for fixed Nc that the DCA.

Significance• Faster estimates of phase diagrams for rapid materials design for alloys.• Offers potential method to extend quantum cluster calculations to large sizes.• Connects field theory methods to coarse-graining cluster methods.

PublicationsTeck L. Tan and D. D. Johnson “Quick, Topologically-Correct Phase Boundaries and Transition Temperatures for Ising Hamiltonians via Cluster Models related to the Dynamical Cluster Approximation,” submitted.

Fast Phase Boundaries and Transition Temperatures via Cluster Models related to Dynamical Cluster Approximation

Duane D. Johnson and Teck Tan (Student)

Figure (left) Phase Diagram: Temperature vs. Magnetic Field for Antiferromagnetic Ising model, for Weiss, DCA-based approach with 1-atom cluster, and (exact) Monte Carlo. New single-site theory gets topology and transition temperatures well.

(right) Normalized pair-correlation for Ferromagnetic Ising model versus lattice cluster momenta (Ncell) for fixed cluster size (Nc). Already for Nc=2 and Ncell large, we obtain the exact Monte Carlo result!

The Materials Computation Center, University of Illinois

Documents