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.
  • Mathematics
    Storage
    Compute
    A
    Dr. Sinead Ryan
    Trinity College Dublin
    In recent years, there has been renewed interest in charm physics. Many charmonium-like states, such as X(3872), Y(4260), X(3940), Y(4140) etc., have been observed in experiments. The precise study of charmonium spectrum is an important topic both experimentally and theoretically. Lattice QCD is the only method to study the low-lying hadron spectrum quantitatively from first principles. It requires high precision numerical simulations. In this project, we study the charmonium spectrum on a set of anisotropic lattices. The anisotropic lattices are discretized in the way that the lattice spacing at temporal direction is much smaller than the lattice spacing at spacial directions. By doing this, we have the advantage of having small discretization errors in the temporal direction while keeping the computational cost down.
  • Engineering
    Storage
    Compute
    B
    Prof. Patrick Prendergast
    Trinity College Dublin
    The engineering of medical devices begins with the identification of a clinical need followedby the design, prototyping, and pre-clinical testing phases. The pre-clinical testing phase involves laboratory bench tests (including prototype tests in cadavers), computer simulations,and animal trials. Next randomized clinical trials are performed, followed by multi-centretrials and, eventually, the device is released onto the market. The complexity and expense ofanimal and clinical trials highlights the importance of the best possible pre-clinical testingmethods: this project creates a team of researchers to make advances in computer simulationmethods for pre-clinical testing of medical devices.Present computational techniques for the pre-clinical testing of load-bearing medical implantsanalyse the implant’s performance in an ideal case (ideal patient and surgical procedure).However, implants that perform well in an ideal case may perform badly in population-basedmulti-centre trials[1,2,3]. The aim of this project is to develop a computational scheme forprediction of implant performance in a population-based multi-centre trial.
  • Engineering
    Storage
    C
    Prof. Anil Kokaram
    Trinity College Dublin
    There has been an exponential increase in the creation of digital media. Digital visual media e.g.pictures and movies have seen a particular explosion in the last five years. But as it has become easier to acquire images, it has become more difficult to organise them. Few people bother to create picture albums anymore for instance. The same problem is happening in science. Much more visual imagery is being recorded of important phenomenon, but it is difficult to locate the important features. While many researchers are concentrating on making media easier to search, there is less work on making media easier to use. This project explores the concept of Content Aware visual Media Processing (CAMP). The idea is to use salient visual features to control image processing tools, for example highlighting important features in a medical video, or making picture content clearer for viewing on a small screen.One strand of this project deals with research into tools for post-production for stereo-3D content. As part of the project, we have collected a database of stereo-3D video which we wish to make available to researchers in other academic institutions. Therefore we are looking for storage space so that can be accessible over the internet. We would require an initial storage space of 600 GB but that may need to rise in the future to greater than 1TB.
  • Life Sciences
    Storage
    Compute
    B
    Dr. Arun Bokde
    Trinity College Dublin
    Cyceron, University of Caen and Paris Descartes, France
    The pattern of degenerative changes in the white matter (WM) of the brain over the course of ageing in both healthy older people and those suffering from Alzheimer’s Disease (AD). In this project, we analyze diffusion tensor imaging (DTI) data using Tract-Based Spatial Statistics (TBSS) in healthy controls (CON), and AD patients. DTI allows us to visualize WM tracts in the brain and to assess the effects of neurodegeneration on these tracts. Our data, derived from calculations performed on the Trinity College high performance computer (HPC), seek to provide new methods which can be used for the early diagnosis of AD. Our initial pilot studies have found that particular indices of water diffusion (axial diffusion, radial diffusion and mean diffusivity) are significantly better at pin-pointing the earliest changes that occur in AD, when compared with measures of fractional anisotropy (FA) which are traditionally used in the clinic to assess damage to WM tracts. Our future work aims to develop automated methods of AD diagnosis based on analysis of DTI scans.
  • Chemistry
    Compute
    B
    Prof. Graeme Watson
    Trinity College Dublin
    Performing MD simulations of a lipid / protein system to test the applicability of the parameter set developed for the MOLDY project in the MD program AMBER. Comprises 3 MD simulations of 20 ns duration.
  • Life Sciences
    Dedicated human Support
    C
    Susanne Barth
    Teagasc
    White clover (Trifolium repens L.) is a perennial legume that is important in grasslands as forage and feed for sheep and cattle, and as a natural source of soil nitrogen. White clover is self-incompatible, meaning that a plant producing male and female gametes cannot self-fertilise due to a biochemical incompatibility. The aim of this project is to find out more about the mechanism controlling self-incompatibility in white clover. During this project, we have used genetic linkage mapping to locate the white clover self-incompatibility locus on the distal end of homologue 1, which correlates to a similar result in red clover and demonstrates that only a single locus is involved in this species. Using degenerate primers and next-generation sequencing (454) of the transcribed genes in the female reproductive organs, we are trying to identify candidates for one of the genes involved in self-incompatibility in white clover. Knowledge of the mechanism of self-incompatibility in white clover would contribute to the study of self-incompatibility in general, which is highly prevalent in flowering plants. It could also be of help to clover breeders. I contacted the TCHPC group in relation to 454 sequencing data from thewhite clover pistil (female reproductive tissue) transcriptome, which has not been sequenced before. This kind of data in general is relatively new and I wanted advice on analysis tools. The group were very helpful.
  • Life Sciences
    Storage
    Compute
    C
    Prof. John O'Docherty
    Trinity College Dublin
    Understanding the way in which value is represented in the human brain is a critical question in the study of the humanreward learning system and decision-making in general. Economic decisions between different goods are typicallymade on the basis of associated valuations of the available goods. Previous work using functional magnetic resonanceimaging has implicated areas of the orbital frontal cortex in the encoding of this decision value signal. However it is notknown whether this signal contains information specific to the good in question or rather, homogeneously reflects valueacross all goods. We aim to answer this question by attempting to decode BOLD signals using multivariate patternclassification techniques while participants placed economic valuations on items drawn from three distinct categories.
  • Life Sciences
    Compute
    A
    Prof Hugh Garavan
    Trinity College Dublin
    University of Utrecht
    Voxelwise statistical analysis of DTI Fractional Anisotropy (FA) data will be carried out using TBSS (Tract-Based Spatial Statistics, [Smith 2006]), part of FSL [Smith 2004]. First, FA images will be created by fitting a tensor model to the raw diffusion data using FDT, and then brain-extracted using BET [Smith 2002]. All subjects' FA data will then be aligned into a common space using the nonlinear registration tool FNIRT [Andersson 2007a, 2007b], which uses a b-spline representation of the registration warp field [Rueckert 1999]. Next, the mean FA image will be created and thinned to create a mean FA skeleton which represents the centres of all tracts common to the group. Each subject's aligned FA data will then be projected onto this skeleton and the resulting data fed into voxelwise cross-subject statistics. This analysis will be performed on a large dataset (N=184) and will be performed as a "blind analysis" with information of the group identifier and subject age given but without clinical information on the patient group. [Smith 2006]
  • Computer Science
    Compute
    C
    Dr. Saturnino Luz
    Trinity College Dublin
    Multiparty meeting browsing and retrieval is an active research area. One problem is how to segment multimodal recordings automatically, and offer reference points on topic change. A novel solution based on content-free features is proposed in this study.A solution to general meeting browsing is difficult, because the meeting structures vary, and there is no uniform definition on topic. The research proposed here is confined to multidisciplinary medical team meetings (MDTMs). MDTMs have become an established practice in many hospitals, and they have relatively predictable structure. The meeting participants are clinical specialists working together, the discussion topics are divided by patient cases, and for each patient, the discussionis organized by certain steps. All of these structures can be used to design an algorithm, in order to automatically locate the reference points in audiovisual recordings.The most salient event in MDTM is the patient case discussion (PCD). A PCD at an MDTM is a highly structured event and therefore its vocalization patterns should be amenable to automatic generalization. This research focuses on automatic PCD segmentation. It has the potential that the solutions in MDTMs can be extended to more general meeting segmentation tasks.The initial steps in this study are speaker turn detection, speaker segmentation and speaker clustering. Once vocalization turns are segmented automatically, the research turns into the main question: to perform topic segmentation (PCD segmentation).In further steps, the PCDs can even be segmented into discussion stages. This study leads to good understanding of meetings and finally multimodal meeting data can be incorporated into electrical patient records.
  • Mathematics
    Compute
    C
    Dr. Mike Peardon
    Trinity College Dublin
    Depfa Bank
    Currently involved in developing a mechanism for pricing covered bonds via analytic and Monte Carlo methods. This tool is being implemented on GPU technology within the CUDA framework. The pricing of covered bonds is computationally intensive and requires large amounts of processing time. The CUDA implementation will allow for massively parallel computation which will significantly decrease processing time and bring improved accuracy. Further investigation into error bounds between single and double precision representations will also be undertaken.

Last updated 19 Dec 2014Contact TCHPC: info | support.