Skip to main content »

Trinity College Dublin

Ongoing Research Projects supported by TCHPC

TCHPC allocates resources to assist research in many different fields. Below is a list of current projects been undertaken with the help of TCHPC.
  • Life Sciences
    Compute
    B
    Dr. Richard Anney
    Trinity College Dublin
    Our work currently examines genetic association across a number of psychiatric traits. We have collected GW-level data on a variety of array-based SNP platforms (400K SNPs to 1.1M SNPs; Affymetrics, Perlegen and Illumina). To examine the overlap of signal we require genotyping to be performed on the same genetic markers. Unfortunately, the main arrays have minimal overlap. Statistical approaches have been developed to impute missing data, taking advantage of information on phased haplotypes from reference human populations (1000 genome project). In this study we intend to impute all studies to a core overlapping dataset of >2M SNPs using the jave-based imputation software Beagle 3.2.
  • Computer Science
    Compute
    C
    Dr. Carl Vogel
    Trinity College Dublin
    Henan University of Technology
    As an implicit inference mechanism, classification has been successfully applied to solve a wide variety of natural language processing (NLP) problems, such as sentiment analysis, named entity recognition, semantic role labelling, speculation detection, and so on. With text classification techniques, different kind of useful information can be obtained from text, and these information can be further used to support complex tasks like machine translation. In this project, we are exploring how to design effective text classification algorithms for solving different NLP problems. Structure information about datasets and text data is being considered in our design. We experiment with different NLP problems and different datasets.
  • Economics
    Compute
    C
    Professor Colm Kearney
    Trinity College Dublin
    The motivation behind this work is to see whether, before, and after announcements of takeovers, there are possible profitable arbitrage opportunities based on the interaction between trading volumes and equity returns. This involves estimating bivariate and multivariate GARCH models, and trying to estimate them in various ways with alternative assumptions about the structure of the data distributions. Since our analysis includes a relatively large number of time series, we are particularly interested in simpler approaches to estimating large covariance matrices, and we have settled on the MacGyver method introduced by Engle (2007) which is designed to solve these high dimensional problems and simplifies the estimation process for large systems.
  • Physics
    Storage
    Compute
    A
    Dr Claude Ederer
    Trinity College Dublin
    Transition metal oxides are notoriously difficult to treat using standard approximations to density functional theory (DFT). This is due to the fact that many of these materials are either Mott insulators or are close to a Mott-insulating phase. In contrast to conventional "band-insulators", where the insulating character is due to either completely empty or completely filled electronic bands, the insulating character of Mott insulators is due to the Coulomb repulsion between electrons in partially filled bands. The physics leading to such behavior is not well accounted for in the standard approximations to DFT. However, very powerful algorithms such as dynamical mean-field theory have been developed in recent years, that allow to study the physics of Mott insulators within simplified models involving only a very small number of electronic states. In this project we combine DFT electronic structure calculations with dynamical mean-field theory. The resulting method allows to accurately describe properties of many technologically important transition metal oxides, including titanates, manganites, and cuprates.
  • Physics
    Storage
    Compute
    A
    Dr Claude Ederer
    Trinity College Dublin
    Materials where magnetization and electric polarization coexist (multiferroics) and/or are coupled (magneto-electrics) offer tremendous potential for future applications as sensors, switches, actuators, or memory devices. Of particular interest is the possibility to control magnetic properties using electric fields and vice versa. However, until very recently only few such materials were known, and the underlying mechanisms were not well understood. This project aims at developing a good physical understanding of magneto-electric and multi-ferroic properties of complex oxides, both in bulk form and in nano-composites, and to identify promising new materials for future device applications. To this effect, we are using first principles electronic structure calculations and combine the results of these calculations with simplified models and symmetry analyis.
  • Physics
    Storage
    Compute
    B
    Prof. Stefano Sanvito
    Trinity College Dublin
    Fraunhofer IWM Freiburg
    Perovskite oxides have attracted increasing interest in science and technology as potentialalternatives to various silicon-based microelectronic components. Strontium titanate (STO) isa representative perovskite oxide that can be used as a dielectric in thin-film capacitors orfield-effect devices. One of the crucial characteristics of these microelectronic devices is theleakage current. From experiments it is evident that leakage currents vary over many orders ofmagnitude depending on the electrode metal and on external conditions such as temperatureand applied voltage.Since the perovskite oxides are insulating materials with a relatively small band gap (typicallyaround 3.5 eV), the thermionic emission is considered as one of the primary sources of leakagecurrent. This conduction mechanism is directly related to the Schottky barrier (SB) that formsat the interface between the dielectric and the electrode.In order to perform a systematic study of the SB-dependence on the structure and composi-tion of the dielectric-electrode interfaces, we have calculated the SB-heights (SBH) for va-rious transition-metal/STO(100) interfaces using a mixed-basis pseudopotential approach ofdensity-functional theory (see [2],[1] and references therein). For a precise analysis of SBformation mechanisms we used a step-by-step procedure, which enables to distinguish thecontributions due to atomic and electronic rearrangements [2].The aim of our future work is to investigate the transmission probability across the hetero-junctions, which is related to the quantum conductance. Calculations will be done using theS MEAGOL package that is based on a non-equilibrium Green’s function formalism for phase-coherent transport [3, 4]. This method also enables to examine the dependence of the SBH andits contributing mechanisms on the applied voltage.In order to increase my proficiency of the S MEAGOL package I would like to visit the groupof Prof. Sanvito at the Trinity College in Dublin. Prof. Sanvito is one of the S MEAGOL devel-opers. His group has studied electronic transport properties of magnetic Fe/MgO(100) tunneljunctions as well as the ferromagnetic perovskite oxide strontium ruthenate. He has broad ex-perience and knowledge in this research field and I believe that I could greatly profit from thisvisit.
  • Chemistry
    Storage
    Compute
    B
    Dr Isabel Rozas
    Trinity College Dublin
    We aim to computationally design, synthesise and pharmacologically test a number of guanidinium derivatives as alpha2-adrenoceptor antagonists with antidepressant application.Based on two recently discovered alpha2-AR antagonists, found in Rozas’ group, showing antidepressant activity, it is proposed to undertake a ligand based drug design strategy to identify pharmacophoric elements for antagonistic activity. In particular, the guanidinium cation is a very aprticular one that can establish not only ionic and hydrogen bond interactions but also cation-pi interactions with aromatic systems. All this type of interactions will eb also research.
  • Physics
    Storage
    Compute
    C
    Dr Stefan Hutzler
    Trinity College Dublin
    Institute of Mathematics and Physics, Aberystwyth University, Wales, UK
    The problem of how to pack objects into a confined space is something that every economy-class flyer is familiar with!!!. Surprisingly, such problems find an extremely broad range of scientific and industrial applications including how a manufacturer of Chinese egg rolls can pack their food products into cylindrical containers in the most efficient way all the way to a scientific understanding of how molecules can self-assemble themselves into nanostructures that form the basis of next-generation technology. For packings of hard spheres in cylindrical channels, a wide range of chiral and achiral close-packed structures below a tube-to-sphere size ratio of 2.713, in which cases there exist no internal balls, have previously been discovered by Pickett et al. [1] and by ourselves [2] through the computational method of simulated annealing. Here, we would like to extend our numerical investigation to cases with a tube-to-sphere ratio larger than 2.713 where there will be the existence of internal balls in the close-packed structures. The numerical results obtained will be vital to an extension of our current theory [2], which is only valid for cases without the presence of internal balls, to a general theoretical description of hard-sphere packing in cylindrical channels.[1] G. T. Pickett, M Gross and H. Okuyama, Spontaneous Chirality in Simple Systems, Physical Review Letters 85, 3652 (2000).[2] A. Mughal, H. K. Chan
  • Life Sciences
    Storage
    Compute
    B
    Dr. Juan Pablo Labrador
    Trinity College Dublin
    Transcriptional factors (TFs) are proteins that bind in a specific DNA sequence. Their DNA binding domains recognize collections of short related DNA 'motifs’.Identification of regulatory binding motifs in DNA is a notoriously difficult task as these binding motifs are hidden amongst a vast amount of genomic noise. A wide range of computational methods to find cis regulatory sites has been proposed. However, at present individual computational techniques for the detection of binding sites have limited predictive value. The aim of this project is to develop and implement a web based platform with a collection of tools that work together to enhance the prediction of cis regulatory sites. The TCHPC clusters will be used to carry out comprehensive test runs of these tools to determine optimal parameter settings.
  • Physics
    Storage
    Compute
    C
    Dr Stefan Hutzler
    Trinity College Dublin
    Institute of Mathematics and Physics, Aberystwyth University, Wales, UK
    It has been found that hard spheres, of diameter d, when packed into a tube of diameter D, will form crystalline structures. The exact nature of the structure depends on the ration D/d. Carrying on from previous work conducted by Pickett et al., we intend to apply simulated annealing techniques to predict the exact crystalline arrangement of particles required at high values of D/d. This is of particular interest as we expect to see unique spiral defects emerge from our simulations as the inner sphere structure attempts to form an fcc structure, a structure which is not commensurate with the curving boundary conditions of the cylinder. The results from these simulations will be compared, using ray-tracing visualisation techniques, with monodisperse microfoam experiments, in order to determine the exact inter-particle potential present within the foam experiments.

Last updated 24 Oct 2014Contact TCHPC: info | support.