May
Mo	Tu	We	Th	Fr	Sa	Su
	1	2	3	4	5	6
7	8	9	10	11	12	13
14	15	16	17	18	19	20
21	22	23	24	25	26	27
28	29	30	31

NGSSC Project Catalog

A list of previous NGSSC projects from 1996-2002

NGSSC graduate student projects

Projects admitted 1996

1. Statistical Mechanics of Protein Folding

Graduate student: Erik Sandelin

Responsible advisor: Carsten Peterson, Dept. of Theoretical Physics, Lund University

Abstract:
We aim at understanding of the thermodynamics and kinetics of protein folding by studying simplified models where the conditions can be controlled. In recent years there has been an increasing interest in such models. A popular choice is to use lattice-based models, which have the advantage that the ground state can be determined by enumeration techniques for small/medium sized problems. However, lattice models have the disadvantage that local features of the energy landscape are not accurately reproduced. In this project we take an off-lattice approach. where the local interactions mimic functional proteins and the non-bonded forces are given by the Lennard Jones potential that emulate hydrophobicity. By developing novel simulation techniques like "dynamical parameter methods" it will be feasible to map out the thermodynamically dominant states and understand their sequence dependence.

Related SSF programs:
Structural Biology - Particularly good contacts with Sture Forsen, who heads a node in this program.

Complex Systems - Proposal from Carsten Peterson.

2. Simulations of Optically Active Materials

Graduate student: Maria Engström

Responsible advisor: Hans Ågren, Dept. of Physics and Measurement Technology, Linköping University

Abstract:
The program aims at development of efficient computer algorithms relating to approximate solutions of the quantum equations, and at the use of these to simulate properties of systems that also can be of technological interest. This development will connect to the work of the computational physics group within self-consistent field theory and analytic response theory. Applications will focus on molecular properties, spectra and reactions. Of special interest will be the modeling of media effects including solid matrices, metallic surfaces, solutions, dielectric media. As a first step we will consider some simple model systems that are relevant for the design of optically active or conducting molecular devices. The understanding of the microscopic origin of strong non-linearity of electric and magnetic properties in these compounds will form a research objective. A goal is to relate the appearance of the strong non-linearity with both electronic and geometric structures of the compounds, with the ultimate goal in mind to be of help in designing optically active materials through simulations. Such modeling will be used to predict specific properties and functions of different chemical groups and substituents.

Materials with strong second and third harmonic generations are of interest because they can double, triple,, laser frequencies, and can also be used for diagnostics of e.g., molecules adsorbed on surfaces. These harmonic generations are directly connected with hyperpolarizability tensors that are to be computed for surface adsorbates, polymeric compounds, metal-polymer interfaces, and solid charge transfer complexes, all with exceptional properties in this area.

The third field of applications of relevance for this Ph.D. program concerns electro-optic and magneto-optic Kerr effects. These effects are described by the electric, respectively magnetic, hyperpolarizabilities and susceptibilities, and are caused by the action of external field (electric or magnetic) on the polarization directions of light. "Kerr" compounds hold great promises as new materials with applications in optical computers.

Related SSF programs:
Forum Scientum - 50 % funding of the student

3. Monte Carlo Simulation of polyelectrolytes and polyampholytes

Graduate student:Malek Khan

Responsible advisor: Bo Jönsson, Dept. of Theoretical Chemistry, Lund University. Assistant advisor: Roland Lindh, HPC-lecturer, Dept. of Chemical Physics and Theoretical Chemistry, Lund University.

Abstract:
The structure and dynamics of polyampholytes are not only affected by their net charge but also by the topological charge distribution. We will use a newly developed algorithm, which permits simulation of "unphysical" variables, to study polyampholytes. The method allows all possible moves of charges including the motion of charges within a polymer. This means that positive and negative charges will obtain optimal positions and the final simulation result will be dominated by the most probable ampholytes.

A particular interesting aspects of polyampholytes is how they can be used to modify the interaction of other charged macromolecules. Both simulations and analytical methods will be used to study this aspect.

DNA is a very highly charged polyelectrolyte and it will have special properties in a solution containing multivalent ions like the naturally occuring spermine and spermidine. Due to the strong ion-ion correlations one could expect to see attractive interactions between the negatively charged chains as well as within a single chain. These attractive interactions will be studied expertimentally and in Monte Carlo simulations. (Future collaboration with Akzo Nobel).

Related SSF programs:
Structural Biology

4. Domain decomposition methods for combustor flow problems

Graduate student: Andreas Kähäri

Responsible advisors: Bertil Gustafsson and Lina Hemmingsson, Dept. of Scientific Computing, Uppsala University.

Abstract:
A project concerning combustor flows has recently been started in cooperation with Volvo Aero Corporation (VAC) in Trollhättan. We will develop new methods for computing low Mach number flow that occurs in combustors. The computational problem is very large, and it is necessary to use high performance computers. The flow is characterized by low Mach-numbers, and if explicit methods are used, the stability restriction forces the time-step to become exceedingly small. There are several ways to treat this problem. One way is to use Helmholtz decomposition, where the acoustic and convection phenomena are separated. With this and other techniques, there is an implicit part that introduces coupling between all the grid-points. For parallel computers, it is natural to use domain decomposition techniques, where each processor is treating a subdomain. The key problem is to find a technique for local decoupling between the subdomains in each iteration, such that good convergence properties are still obtained.

Schur-complement methods will be studied, where the original system of equations is reduced to a smaller system for the unknowns at the interfaces between the subdomains. When this system is solved, each subdomain system can be solved independently of the others. Here GMRES type methods can be used for both types of systems.

The cooperation with VAC means that part of their existing code VOLSOL, based on a finite volume approximation, can be used for validation and experiments. However, the new part of the code has to be parallelized for implementation on a parallel computer, and this implementation is a non-trivial undertaking. The plan is to use the PVM software for this.

Related SSF programs:
National Network in Applied Mathematics - In connection with aerodynamical applications

5. Active Control of Boundary-Layer Transition

Graduate student: Markus Högberg

Responsible advisors: Dan Henningson, Dept. Mechanics, KTH and FFA and Martin Berggren, FFA

Abstract:
Study and design of active control strategies for transition in boundary layer flows is proposed. The control strategies will be designed using the optimal-control approach to control of the Navier-Stokes equations and the adjoint-equation technique for associated gradient computations. The strategies will be designed to control or delay bypass transition. This represents a significant new step compared to previous work almost exclusively devoted to anti-phase modal suppression of two-dimensional TS-waves or wave packets. In particular the aim is to control the growth of streaky structures associated with most bypass transition scenarios.

The HPDR part of the project concerns a graduate student involved in the computational aspects of the problem where the complete transition process will be calculated by Direct Numerical Simulation (DNS) of the Navier-Stokes equations. This is highly computationally intensive and the code will be implemented on parallel computers with distributed memory. Experimental and theoretical aspects of the problem will be funded from other sources.

Related SSF programs:
Integrated Vehicle Structures

6. Molecular Control Theory

Graduate student: Johan Paulsson

Responsible advisor: Måns Ehrenberg, Dept. Molecular Biology, Uppsala University

Abstract:
1. Stochastic, mathematical models for copy number control in the E.coli plasmids R1 and ColE1 are developed. The analytical and computational tools from technical control theory are modified to fit the biological context. The analysis focuses on loss rates of plasmids from the host cells, copy number distributions and energy consumption to drive the control mechanisms. The total costs for control are calculated. and system performance is optimized to maximize population genetic fitness of the host-plasmid population. The behavior of optimally tuned systems is compared with in vivo data.

2. The plasmid models are modified to obtain synchrony in initiation of replication. An attempt is made to model initiation of bacterial chromosome replication, with guidance from the behavior of the simpler plasmid systems.

3. Aspects of bacterial steady state cell physiology are modeled and system behavior is optimized for maximal growth. Predictions from optimal behavior on protein and RNA-levels are compared with in vivo data. Chosen control systems are introduced and analyzed in increasingly complex situations. Attempts are made to model control systems not understood today to create a basis for experimental work.

Related SSF programs:
Cell Factory for Protein Expression

7. Numerical Procedures for Inverse Heat Conduction - Problems and Industrial Applications

Graduate student: Fredrik Berntsson

Responsible advisor: Lars Eldén, Dept. of Mathematics, Linköping University and Dan Loyd, Dept. of Mechanical Engineering, Linköping University

Abstract:
If the heat flux or temperature histories at the surface of a solid are known as functions of time, then the temperature distribution inside the solid can be found by solving a initial-boundary value problem for a parabolic partial differential equation. In many industrial heat transfer situations, the surface is inaccessible for measurements, and the computation must be based on temperature measurements at interior locations. This is the case when it is necessary to measure the temperature of a fluid with a shielded thermocouple. The shielding is required, e.g. because of the high temperature of the fluid. A general characteristic of this situation is that the temperature measurements are erroneous due to the shielding.

This problem is an examples of an inverse heat conduction problem, and it is ill-posed in a mathematical sense: the solution, if it exists, does not depend continuously on the data. Nevertheless, because there are numerous applications in industry and science, where such problems occur, it is of prime importance to develop theory and algorithms for their solution.

In this project we will develop and analyze numerical procedures for the solution of inverse heat conduction, and, in a collaboration between numerical analysts and thermal dynamicists, apply such numerical procedures to actual inverse heat conduction problems from industry.

Related SSF programs:
Industrial Systems Technology
Combustion Science and Technology Research
Energy Systems
National Network in Applied Mathematics

8. Solvent Exchange Mechanisms by MD Simulation

Graduate student: Daniel Spångberg

Responsible advisor: Kersti Hermansson, Dept. of Inorganic Chemistry, Uppsala University

Abstract:
The solvated metal ion is the precursor of all solution phase metal ion coordination chemistry. A clear understanding of the structure and dynamics of solvated metal ions is therefore essential for understanding the of ions in solution. The pure solvent exchange reaction in aqueous ionic solutions, for example, can be formulated like this:

Mn+(H2O)6 + H2O* --> Mn+(H2O)5(H2O)* + H2O

The detailed mechanisms for the solvent exchange process for a certain ion and solvent is often difficult to establish with certainty from experiment.

The overall goal of this Ph.D. project is to develop a microscopic picture of solvent exchange processes around some metal ions in solution using MD methods. Mechanisms, reaction rates and pressure influence will be studied, as well as the effect of the competition of different ligands on the exchange process.

The intermediate transition complexes in these processes occur infrequently and are short-lived. Very long - and thus CPU-demanding - simulations runs will therefore have to be performed to obtain sufficient statistics. Modifications of the normal MD procedures will have to be made to allow also the study of somewhat slower exchange processes which are not accessible by regular MD simulations. The project will involve an integrated training in research strategies, inorganic chemistry, chemical computational methods, numerical methods and computing.

Related SSF programs:
National Network in Applied Mathematics
The Brinell Centre

9. Numerical Simulation of Dendritic Growth Using Phasefield Model

Graduate student: Christer Andersson

Responsible advisor: Gunilla Kreiss, Dept. of Numerical Analysis and Computer Science, KTH

Abstract:
We wish to model growth of a dendrite from a seed in a sub cooled liquid bath including the following effects: Heat flux by conduction and convection, diffusion and advection of alloying elements, anisotropy of the solid, surface tension at the interface, the relation between the liquidus temperature and the concentration of the alloying element, and the varying velocity field. Most simulations have used simpler models and front tracking methods. Such methods are difficult to implement, especially in 3D. We shall instead use the recently introduced phasefield approach, where the sharp front is replaced by a thin interface region. A smoothly varying phasefield variable identifies each point as being in the liquid, in the solid or in the interface. The phasefield variable satisfies an evolution equation in the whole domain. This approach is more amenable to analysis, and methods based on this approach do not require an explicit computation of the front position.

Related SSF programs:
National Network in Applied Mathematics

10. Computer algorithms in environmental monitoring

Graduate student: Fredrik Norström

Responsible advisor: Bo Ranneby, Dept. of Forest Survey, Swedish University of Agricultural Sciences

Abstract:
The objective of this project is to further develop algorithms to be used in environmental monitoring systems. The project will be integrated with the MISTRA-project "Remote sensing for the environment"; especially with the sub-projects "Sampling procedures and stochastic modeling for environmental monitoring" and "Monitoring of forest ecosystems". As there is a wide range of environmental issues with quite different information sources, different methods and models have to be developed.

To meet the needs of society it is necessary to develop a cost-effective, flexible, multisource monitoring system. In a cost-effective monitoring system, information from many different sources has to be integrated. Main sources in such a system may be field data, multitemporal remote sensing data from different sensors, digital maps, and deterministic and stochastic models.

While a good deal of research has been conducted on use of specific image types, little has been done on effectively combining information from multiple images of the same or different types with field data through objective sampling designs. The two sources of data (ground and remotely sensed) on their own most certainly fall short in providing accurate, cost-effective state and change-of-state information. New sampling designs, together with supporting computer algorithms, focusing on the best of both data sources must be developed.

Related SSF programs:

11. Electromagnetic modeling

Graduate student: Torleif Martin (not financed by NGSSC)

Responsible advisor: Karl-Fredrik Berggren, Dept. of Physics and Measurement Technology, Linköping University

Abstract:
Electromagnetic modeling is a vast and dynamic field. Internationally it is expanding rapidly because of its importance in engineering and science. Modern computational techniques make it possible to perform reliable numerical simulations of the electromagnetic environement in realistic situations, e.g., complex engineering problems with awkward geometry and boundary conditions etc.

This project is focused on a numerical method called FDTD (Finite Difference Time Domain). As indicated by the name it is a time domain method where Maxwell's equations are discretisized both in space and time.

There are several advantages with this method compared to other "traditional" methods such as Method of Moment, especially for problems involving lossy dielectrics. Since the space is disctretisized there is a lower limit of the object dimension which can be resolved with this method. Further developement of the FDTD technique will be performed, e. g. concerning strong field variations at small distances, surface waves, skin-effect dispersive media. A new technique involving semi-empirical models in the numerical code will also be investigated.

The code will also be tested on applications such as antenna problems, EMC, electromagnetic effects on human tissue, etc.

Related SSF programs:

Projects admitted 1997

1. Quantum transport in nanodevices

Graduate student: Torbjörn Blomquist

Responsible advisor: Karl-Fredrik Berggren, Dept. of Physics and Measurment Technology, Linköping University. Assistant advisor: Igor V. Zozoulenko, same department.

Abstract:
The great interest in electron properties and applications of semiconductor structures and devices of reduced dimensionality have been motivated by recent breakthroughs in nanofabrication technology. Today, the forefront of the field has moved down to dimensions of the order of tens of nanometers. Emerging novel semiconductor manufacturing technologies already allow one to achieve an unprecedented control of the spatial pattern with the precision down to a few nanometers. Because of the decrease of spatial dimensions, the classical description of electron transport eventually fails and quantum effects take over, which implies fundamental changes in electronic and transport properties of semiconductor devices. This naturally imposes limits on utilization of conventional classical modelling of semiconductor devices like FETs, bipolar transistors, etc. Thus we intend to investigate electric transport and the functioning of nanoscaled semiconductor structures and electronic devices at a basic quantum mechanical level. Nanoscale devices are also highly topical because they provide a new probe of the manifestation of chaos in quantum systems. This conceptually difficult and incompletely understood area, particularly the relation between quantum and classical dynamics, will be explored. Because of the richness and compexity of the physics involved, computer visualization will be a very essential aspect of our project.

Related SSF programs:
"Microelectronics: Photonics and Nano-Science"and "Nanostructure Consortium", Lund University

2. Methods of Electrical Treatment of Cellular Animal Foods

Graduate student: Sven Isaksson

Responsible advisor: Thomas Ohlsson, The Swedish Institute for Food and Biotechnology

Abstract:
Treatment of food by methods of electricity offers interesting possibilities to instantly heat the whole food volume. The heat treatment can this way be made quicker, and be controlled, so that desired temperature profiles are more easily achieved without the limits of traditional heating. The methods are often divided into four (each demanding different techniques): ohmic (50 - 60 Hz), radio frequency (13 MHz), and microwave (400, 900 or 2450 MHz) heating. Industrial usage demands knowledge of how to design the heating method in order to achieve desired temperature profile, and how to integrate old technique with old.

SIK has examined electrical heating methods for some years now. One tool has been computer simulations and calculations. One such tool has been the FDTD (Finite Difference Time Domain) method. The main target is to be able to analyse the influence of different products and process parameters, and also to be able to synthesise right principle of energy division, outgoing from desired temperature profiles and temperature rise, so that an equipment design can be made to fulfil these. Simulation software working in the time domain are interesting for various reasons in this work, e.g. situations where more than one frequency is present.

One such situation is by the use of high electrical pulses (HELP) in order to perforate the walls of cells. HELP has been used in areas such as genetic research, but is now also studied as a method of microbiological decontamination. It is of interest to be able to simulate the method, as well in the equipment design as in the situations for the microbiolocical organisms and cell walls.

Primarily the concerns of the project are focused on development and/or choice of synthesis tools. These are to be used in the choice of process conditions, process design, and frequency domains, with knowledge about the product as "input".

The project is focused on cellular animal foods, minced or whole meat. Secondly, the project is focused on further development of the methods, in order to synthesise HELP (again applied on cellular minced or whole meat). The calculations will be verified by investigating the effects of different levels of strong electrical pulses. Microbiological decontamination will be investigated for different field strengths pulse programs. The effect of the pulses may be expected to strongly depend on the temperature at which the treatment is carried out. Therefore it may be likely to use a controlled volume preheating by using the methods developed in the primary step of the project.

Related SSF programs:
LiFT (50% financing)

3. Numerical methods for Aerodynamic Optimization

Graduate student: Oskar Enoksson

Responsible advisor: Per Weinerfelt, Dept. of Mathematics, Linköping University

Abstract:
The research deals with algorithms for aerodynamic optimization. The work has during 1997 been directed towards a deeper understanding of the state of art algorithms for aerodynamic optimization. For industrial applications it is necessary to enable optimization of complex geometries. As a first step towards a general optimization code for aerodynamic applications a multi block Euler solver, based on domain decomposition, for structured grids has been written. An object oriented language (C++) was chosen in order to get a modular program, which easily can be extended, and to simplify the treatment of the multi block boundary conditions. To ensure a high efficiency on vector and parallel computers the most time consuming routines were written in FORTRAN. The program has been tested and validated on different computers like SGI work stations and the super computer Cray C90.

During 1998 the work will be focused on the following topics: 1. Investigation of different gradient based optimization algorithms like MMA (Method of moving asymptots), Quasi-Newton methods. 2. Study combinations of genetic algorithms and gradient based algorithms. 3. Derive optimization algorithms based on a B-spline or Bezier polynomial rep resentation of the geometries to be optimized. 4. Investigation different types of constrains on the flow solution and the design variables. 5. Parallelization using MPI.

Related SSF programs:

4. Computational Topology Design of Mechanical Structures

Graduate student: Thomas Borrvall

Responsible advisor:Joakim Petterson, Department of Mechanical Engineering, Linköping University. Assistant advisor: Anders Klarbring, same department

Abstract:
Topology optimization of mechanical structures is a promising tool for obtaining more efficient and lighter structures. The method concerns optimal distribution of material in a given domain with prescribed boundary conditions, and it allows for very general design modifications such as the introduction of holes and adjustment of shapes of structural parts.

The Finite Element Method (FEM) is used to discretize the original problem, so a mathematical program is obtained. In addition to the obvious use of optimization algorithms, this project is concerned with the invocation of numerical analysis in topology optimization. The design of efficient numerical methods in large-scale environments will be considered, and the way to perform the FE discretizations will be analysed. This is of qualitative importance, e.g. to avoid anomalies such as "checkerboard patterns", but also of quantitative importance since it is of interest to, given a certain degree of design resolution, construct a mathematical program with minimum size. The project also aims at establishing convergence of FE solutions to exact ones, error bounds, and adaptive element techniques.

Related SSF programs:
NTM (50% financing), IVS

5. Barrier properties of polymers as revealed by molecular dynamics simulations

Graduate student: Gunnar Karlsson

Responsible advisor: Ulf W. Gedde, Department of Polymer Technology, Royal Institute of Technology, Mikael S. Hedenqvist, Department of Polymer Technology, Royal Institute of Technology and Packforsk - Swedish Packaging Research Institute

Abstract:
Diffusion- and solubility-related problems are from a practical point of view of enormous importance, e.g. for food/medical packaging materials, barriers in long-term installations (for power electronics, energy distribuition and telecommunication), and for gas separation technology. Molecular dynamics (MD) modelling provides a new (highly likely) route to obtain relevant macroscopic data for polymer materials. This project will in particular focuse (a) polymer/small-molecule penetrant systems with specific interaction (e.g. hydrogen bonding) and (b) polydimethylsiloxane/ polydimethylsiloxane networks. Most MD simulation work sofar has been concerned with small penetrant molecules (methane, CO2). It is here planned to study the effect of size and shape of the penetrant molecule. By establishing relationships between penetrant size and shape and transport properties using MD it may be possible by extrapolation to obtain reasonably accurate data for larger penetrant molecules. The idea of further developing the MD technique along these lines is to have, in the near future, a tool with a data bank of universal atom sizes and force-fields which enables the synthesis of any "theoretical" material and the prediction of its properties before it has actually been made.

Related SSF programs:

6. Hydrodynamic aspects on the flow along surfaces with biofouling resistance

Graduate student: Stefan Nilsson

Responsible advisor: Prof. Lars Larsson, Department of Naval Architecture and Ocean Engineering, Chalmers University of Technology

Abstract:
This project is part of a multi-disciplinary research program, MASTEC, with participants from biology, chemistry and naval architecture. The primary purpose is to develop various non-toxic anti-fouling techniques. However, the first part of the present project has a somewhat different objective, namely to numerically study the food concentration in the sea-water above a large colony of bivalves. The bivalves feed on organic material obtained by filtering the water. All bivalves may be assumed buried into the sea-bed and may be represented by two holes in the bottom, one for inflow and one for outflow. The interaction between these flows and the sea-bed boundary layer will be modelled using the incompressible Navier-Stokes equations with a suitable turbulence model. With the velocities obtained from these equations the transport of organic material in the boundary layer will be computed via the solution of a convection- diffusion equation.The computational domain will be a rectangular box and the equations will be written in Cartesian coordinates. Since a large number of bivalves will have to be represented the problem is suitable for parallel processing. This investigation will be presented as a licentiate thesis and the plan is to finish this by the end of 1998.

The numerical technique developed will be useful also for studying the concentration of agents released for antifouling purposes from surfaces submerged into a flow. Within the MASTEC programme the slow release principle is a possible antifouling alternative and the new technique will be used during the second part of the project for investigating this alternative. It should be mentioned that the slow release principle is also of interest in connection with drag reduction.

Related SSF programs:
This project is funded 100% by the SSF program MASTEC

7. Numerical simulation of ship flow at full scale Reynolds numbers

Graduate student: Lars Carlsson

Responsible advisor: Doc. Anders Petersson, Dept of Naval Architecture and Ocean Engineering, Chalmers University of Technology. Assistant advisor: Prof. Lars Larsson, same department.

Abstract:
During the past few years several methods for predicting the flow around ship hulls have been presented. How ever, so far, the major part of the work has been focused on solving the Reynolds averaged Navier-Stokes equations at model scale Reynolds numbers (Re = 10^7). Attempts to predict a stern flow at full scale Reynolds number (Re = 10^9) have been sparse. One of the major difficulties of these calculations is to maintain stability of the numerical solution in the near-wall region. The very highly stretched grid needed to resolve the innermost part of the turbulent boundary layer means that the solver must handle cell aspect ratios of the order of 10^3-10^4 at model scale, and 10^5-10^6 for the full scale problem.

To alleviate the problems new numerical techniques together with overlapping (Chimera) grids will be employed. An appealing compromise between explicit and implicit methods for solving the momentum equations is to use a semi-implicit time-integration approach, where only the stretched direction (normal to the hull) is treated implicitly. The time-step is then governed by the terms in the unstretched directions (tangential to the hull) and may thus be much larger than for a purely explicit method, still avoiding the need for solving very large systems of equations as in a purely implicit method. To avoid problems with the pressure Poisson equation a preconditioner will be developed that solves the equations in the stretched direction by a direct (tri-diagonal) solver during each iteration. Another possibility is to employ an unstretched grid for the pressure equation. This is motivated by boundary layer theory, which shows that the pressure is approximately constant through the boundary layer.

Related SSF programs:

8. Adaptive FEM for Maxwell's equations and multiphysics in 3D

Graduate student: Thomas Rylander

Responsible advisor: Anders Bondeson, Department of Electromagnetics, Chalmers. Coadvisor: Claes Johnson, Department of Mathematics, Chalmers

Abstract:
Computational electromagnetics is a rapidly growing field. At present, the Finite-Difference Time-Domain (FDTD) method is the most popular method for high-frequency applications in Computational Electromagnetics. However, the FDTD is tied to structured, Cartesian grids and is unsuitable for structures that are either geometrically complex or small compared with the wave-length. (FDTD encounters difficulties already with oblique surfaces and wires.) The goal of the present project is to develop and test Finite Element Methods (FEM) for Maxwell's equations to enable treatment of complex geometry. The FEM algorithm will be used for application problems with complex geometry in Electromagnetic Compatibility (EMC), antenna design and analysis and electromagnetic scattering. The FEM scheme will use unstructered tetrahedral meshes in 3D, generated by an existing grid generator. The project will be carried out in close collaboration between the Departments of Electromagnetics and Chalmers' Finite Element Center, Phi.

Related SSF programs:
National Network in Mathematics

9. Finite Element Methods for Maxwell's Equations

Graduate student: Rickard Bergström

Responsible advisor: Claes Johnsson, Mathematics, Chalmers.

Abstract:
We want to find a finite element method for Maxwell's equations that is general and uses adaptivity in order to gain flexibility and efficiency. Generality makes it possible to work with a large area of problems, including coupled multiphysics, and the adaptivity could use a secondary, integral quantity instead of the primary variables. The goal is to perform computations for time-dependent problems in 3D with moving boundaries, so part of the project work will deal with questions concerning how this affects the mesh and the computations.

Related SSF programs:
National Network in Applied Mathematics

10. Simulation of polyelectrolyte-surfactant solutions

Graduate student: Marie Jonsson

Responsible advisor: Per Linse, Dept. of Physical Chemistry 1, Lund University

Abstract:
Polyelectrolytes and surfactants are ubiquitous in technical processes and in different procedures as drugs, foodstuffs, paints, paper etc. Besides experimental investigations, different theoretical approaches have been applied to further examine such systems. The aim of this project is to model solutions containing charged surfactant micelles and oppositely charged polyelectrolytes by performing large-scale Monte Carlo/molecular dynamics simulations using a simple model with focus on the electrostatic interactions. We are using an in-house integrated Monte Carlo/molecular dynamic package MOLSIM, recently ported to parallel computers with distributed memory. We anticipate to gain detailed knowledge of the structure and stability of such systems which will be of fundamental value for assessing the accuracy of more simplified approached and to guide the experimentalists and the industrial producers to achieve systems with the desired properties.

Related SSF programs:
Colloid and Interface Technology

11. Structure and stability in complex biological systems: mathematical analysis of model ecosystems

Graduate student: Charlotte Borrvall

Responsible advisor: Bo Ebenman, Biology/IFM, Linköping University. Assistant advisor: Peter Munger, HPC-lecturer, Theoretical Physics/IFM, Linköping University.

Abstract:
Biological systems, and especially ecological ones, are often very complex and with parts operating on greatly different time scales. The complexity and size of the systems makes it very time-consuming and difficult to performe controlled experiments. The study of the relationship between the structure and dynamics of complex ecological systems can therefore benefit immensely from contributions from mathematics and computational science.

The primary goal with this project is to investigate how the structure of ecosystems affect their dynamics and stability. How will structural properties affect ecosystem resilience (return rate to the original state following a disturbance) and reactivity (short-term response to a disturbance)? How does environmental stochasticity influence the structure and dynamics of ecosystems? Are ecosystems with certain structures more vulnerable to species extinctions than others with a different structure? Are there structures that are particularly likely to collapse irreparably following the temporary loss of one or more species?

These and related questions will be investigated by the formulation and analysis of mathematical models of ecosystems. We use equations of Lotka-Volterra type that take species characteristics and environmental variability into account. Stochastic models are used to investigate extinction risks of species in ecosystems with different structures. Both analytical and numerical methods (using high performance computer systems) will be applied.

Related SSF programs:

12. Monte Carlo modelling of metal oxidation and other surface reactions

Graduate student: Dogan Oner

Responsible advisor: Bengt Kasemo

Abstract:
Oxidation of metals is a complex process involving several reaction steps: dissociation of oxygen or other oxidant molecule, 2D diffusion of dissociation fragments , growth of a chemisorption layer, 3D diffusion of oxidant atoms and metal ions and eventually oxide formation and its growth. Since the oxide growth is essentially a combination of the source term (adsorption from the gas phase) coupled to a complex transport mechanism (of O-atoms and/or metal atoms) a set of coupled differntial equations would result, if an analytical approach was tried, with a number of rate constants. Such a mean field approach lacks the capacity to include important variations in energetics and kinetics. This is basically why Monte Carlo approach is required. The development of the Monte Carlo model includes several steps: The first one aims at simulating the build up of a 2D chemisorption layer of oxygen atoms. Different mechanisms of the dissociative events will be simulated: The options include one of the two fragment atoms to occupy the impact site and the second one to perform a ballistic motion and to land on a distant site, or to escape from the surface, etc. Short range as well as long range lateral interactions between the adsorbed oxygen atoms will be included in the simulation. This type of interaction is central in the oxide growth and includes both electronic and elastic interaction. The validity of the assumptions in the model (e.g. the bond energy of a particular site for certain occupation of the neighbor sites) can not be tested directly by an experiment, since they are related to the "microscopic" nature of the processes. However, the applicability of the model can be estimated on the basis of the global results itproduces, i.e., coverage dependence of the sticking coefficient, etc. Comparisons with experimental results may lead to reconsidering of the description of the adsorption events and O-atoms behavior on the surface. The above essentially concerns the 2D stage of oxidation Further development of the model will include the formation of a 3D oxide film on the surface. The possible events contributing to multilayers formation are determined by analyzing the experimental results. Conceptual studies show that the growth mode of a thin oxide film is mainly determined by energy parameters rather than deposition conditions. Simulations here will be performed for different ratios of the adsorbate/adsorbate and adsorbate/substrate interaction energies. The consequence of weak long range forces (e.g. elastic interaction) will be investigated. Different mechanisms of multilayer formation will be explored, e.g. via subsequent jumps of already adsorbed O atoms to a second, third layer etc. Inclusion of nucleation theory is unavoidable since the transformation from chemisorbed and/or dissolved oxygen to oxide nuclei is a nucleation process. Additionally, the model can include alteration of the surface structure during the adsorption process, e. g. O atoms to become buried in the substrate lattice, thus changing the adsorption conditions for the next impinging molecules.

Related SSF programs:
Graduate School in Materials Science

13. Investigation of the wavelet transforms in filtering and compression of multivariate chemical process and sequence data combined with multivariate data analysis.

Graduate student: Johan Trygg

Responsible advisor: Svante Wold and Michael Sjöström, Organic chemistry, Umeå university

Abstract:
The project will investigate the use of the wavelet transform in filtering and compression of multivariate data in calibration and process modeling of multivariate time series. Further, the use of wavelets will be investigated on large data sets, which result from on-line process control of continuous processes and batch processes. Resulting algorithms will be implemented in Matlab.

Other application areas for the wavelet transform that will be investigated are classification and modeling of DNA, RNA and protein sequences. Large amounts of sequences are currently being generated, and earlier we have shown that sequences can be described as multivariate sequence series andtreated as multivariate time series. The wavelet transform is an interesting preprocessing method for such sequences as it extracts both local and global properties.

Related SSF programs:
Advanced utilization of selected wood fibres from the boreal forest ecosystem; with the program Skogsbioteknik och kemi.

14. Molecular dynamics of disordered materials.

Graduate student: Erik Wensink

Responsible advisor: Göran Wahnström, Inst. för tillämpad fysik, Chalmers

Abstract:
The dynamics of disordered materials differ in many respect from their crystalline counterparts. Some generic features have been observed: a slow alpha-process connected to the viscous flow; additional fast and slow beta-processes; and an enhanced density of vibrational states at low frequencies, compared with the ordinary contribution from acoustical sound waves, the so-called ``boson-peak''. These features show up in most glass forming substances: molecular van der Waals liquids, strong network forming glasses, metallic glasses and in polymers.

By studying polymeric systems (poly(propylene oxide)) with a relatively simple morphology using semi-empirical force fields we will investigate in detail various properties as function of chain length. We will study the dependence of these quantities by doing molecular dynamics simulations for the monomer (a molecular van der Waals system) to a long polymer chain. The time and length scales involved place these computations at the forefront of high performance computing.

The above model will be extended and a doped polymeric system will be considered. Electronic structure calculations will be used to investigate some key steps in the ion conductivity mechanism for doped polymer systems. Polymer ionics is a hot area in materials science and they have very high technological potentials.

The computational work will be done in close collaboration with experimentalists, performing NMR, light and neutron scattering studies on the same systems.

Related SSF programs:
Molecular Engineering in Polymer Science, Graduate School in Materials Science, Polymer Ionics

15. Chemical dynamics on a femtosecond timescale

Graduate student: Åsa Petersson

Responsible advisor:Hans Karlsson, Dept. of Quantum Chemistry, Uppsala University. Assistant advisor: Sverker Holmgren, HPC-lecturer, Dept. of Scientific Computing, Uppsala University.

Abstract:
The goal of this project is to provide a theoretical understanding of the electron transfer from the excited electronic state of an adsorbed dye molecule to the TiO2 surface in a dye-sensitized solar cell. This electron transfer occurs on a time-scale of a few femto-seconds, as shown by recent ultrafast laser spectrocsopy experiments.

To this end we intend to initially study similar processes for smaller model systems using numerical modelling, solving an explicit time-dependent Schrodinger equation. Several different numerical schemes will be analysed and tested in terms of efficiency, reliability and ease of implementation. In view of the application in mind, special emphasis will be on generalizability to larger systems and possibility of using parallell computers.

Related SSF programs:

Projects admitted 1998

1. Simulation of elastohydrodynamic lubrication

Graduate student: Torbjörn Almqvist

Responsible advisor: Roland Larsson, Department of Mechanical Engineering, Luleå University of Technology

Abstract:
Elastohydrodynamic lubrication is found in most common machine components. To improve life of components and to reduce friction it is necessary to enable predictions of the lubricating conditions, e.g., contact pressure and lubricant film thickness. This can be done by using numerical simulations and a computational tool for such simulations is described in the project proposal. the objectives for the proposal are:

Enable simulation of the lubrication conditions, e.g., contact pressure and lubricant film thickness, in machine components such as rolling element bearings, gears, cams, clutches and piston rings.
Investigate the influence of lubricant properties such as non-Newtonian rheological properties, compressibility and thermodynamical properties on film thickness, friction, contact stress etc.
Account for thermal effects, i.e., enable estimations of temperature rise in lubricated contacts and investigate its influence on lubricant film thickness and friction.

The computational tool will be used by designers to improve the lubrication conditions in order to prolong life of components and to reduce friction.

Related SSF programs:
SSF JIG (50% financing). High Performance Mechanical Components (HiMeC), proposal to SSF.

2. Graph Theory and statistical physics

Graduate student: Daniel Andrén

Responsible advisor: Roland Häggkvist, Dept of Mathematics, Umeå University

Abstract:
From data generated from small graphs we (in this case the Mathematics group in Umeå centered around myself and the Theoretical Physics group at KTH centered around Anders Rosengren) find scaling relations that enable us to construct an approximate Ising polynomial for larger graphs, where exact calculations are not feasible. We have already generated large amounts of data (the result of some 400 000 CPU-hours at national parallel computing centra during the 12 month period the project has been running) for some graph families (notably the square and cubic lattices), but need more (for instance data on the diamon lattice, fcc, bcc etc). Reliable data for graphs with up to 30 000 vertices are highly desirable if the asymptotics is to be found with any precision. This requires, and gets, a large amount of computer time and extremely economical programming. For the distribution of the coefficients tin the Ising polynomial we have found a heuristic scaling relation between the distributions for different fixed magnetizations, and this relation we intend to study further. We also intend to apply string tools from probalistic combinatorics to determine the central part of the distribution for fixed magnetization. Presently we have a solution for this part which we strongly believe is asymptotically exact. The approach we have developed is applicable to other spin models such as the Potts model, but we will initially restrict our investigations to the Ising model. the graduate student position we hereby seek money for is within an environment where massive large scale parallel computing is the rule and where exceptional mathematical ability is taken for granted.

Related SSF programs:

3. Proteins and peptides in aqueous solution

Graduate student: Ana Caballero-Herrera

Responsible advisor: Lennart Nilsson, Center for Structural Biochemistry, Karolinska Institutet

Abstract:
The surroundings (aqueous solution or the hydrophobic interior of a lipid bilayer membrane) are crucial for protein stability. We propose to use molecular dynamics simulations to study the stability, dynamics, and solvation of oligopeptides of different sequence, but with the same composition, for which it is known that the sequence infulences the conformational stability in solution. In particular the denaturing effects of urea will be studied. In the second phase of the project these investigations will be extended to interactions between proteins and DNA, where often water molecules are found in the interfaces between the macromolecules. The central question here concerns the role of the solvent for specificity and cooperativity, and it will be adressed in simulations of nuclar hormone receptors and homeodomain proteines (wild-type and mutants) in complex with cignate and non-cognate DNA sequences. The detailed interactions between amioacid sidechains and DNA bases will be characterized by using a potential of mean force calculations of different amino acids and base pairs in a reduced system, where the protein is represented by an alpha-helix and only a small fragment of the DNA is present. In order to obtain statistically significant results for the different systems simulations of several nanoseconds length will be required, but this is achievable with parallel workstation clusters, or on machines such at the IBM SP2 at PDC/KTH.

Related SSF programs:
Structural Biology Network (SBNet)

4. Genetic analysis of complex traits

Graduate student:Örjan Carlborg

Responsible advisor: Leif Andersson, Dept of Animal Breeding and Genetics, Swedish University of Agricultural Sciences

Abstract:
Many biological traits show a complex genetic inheritance. This means that they are controlled by multiple genes in combination with environmental factors. This is the case for most traits of agricultural significance in plants and animals as well as for many major disease syndromes in humans like diabetes and cancer. The development of methods for genetic analysis of complex traits is therefore of broad interest. Methods for multivariate analyses and the exploration of gene interaction need to be developed. is should be stressed that the rapid development of molecular methods for large-scale genetic screenings need to be matched by a similar development of more rapid computational methods. moreover, the analysis of complex traits is computationally very demanding and available computer packages are inflexible and not user friendly.

The objective of this project is to implement and further develop methods for genetic analyses of complex traits. The methods will be applied to experimental data developed at the Department of animal breeding and genetics. The major project involves a cross between the Red Jungle Fowl and an egg-producing line of domestic chicken. The pedigree will be an intercross comprising about 1000 F2-animals. the Red Jungle Fowl is the wild ancestor of the domestic chicken and dramatic phenotypic differences have developed due to selective breeding. Phenotypic data for a range of complex traits (e.g. behavior, reproduction, growth and body composition) will be collected as well as genetic data on about 100 DNA markers evenly spaced in the genome. The material constitutes an excellent model for the analysis of the genetic basis for complex traits.

Related SSF programs:
National Centre for Genome Research. Bioinformatics in Sweden (proposal to SSF)

5. Simulation of defects, dislocations and heterogenous systems

Graduate student:Karin Carling

Responsible advisor: Göran Wahnström, Dept of Applied Physics, Chalmers University of Technology

Abstract:
Substantial advances have been made in recent years in calculating atomic, electronic and magnetic properties for bulk solids and interfaces from first principles. Real materials, however, are often highly heterogeneous; they consist of grains stuck together and contain a large amount of impeerfections as dislocations and voids. Material properties such as the toughness and strength depend seneitively on the concetration and beaviour of these boundaries and imperfections and the link between the atomic structure and material properties is not direct.

We would like to consider the local hydrogen concentrations in the close vincinity of a dislocation. We are interested in the interaction between dislocations and the hydrogen induced effect on that elastic interaction. The infulence of temperature and externally applied stresses as well as the motion of the two dislocations with and without hydrogen present will be investigated. The latter willbe studied at low and high temperatures and we will investigate to what extent the hydrogen surrounding can foloow the dislocation motion. To answer these questions an atomistic description is required and we make use of the molecular dynamics technique to incorporate time and temperature effects and first principles electron structure calculations to determine key parameters in the description of the inter-atomic interaction

Related SSF programs:
Computer Simulation of Studies in Condensed Matter Physics, NFR contract granted by SSF (50% financing).

6. Computational wave propagation in solids

Graduate student:Torbjörn Ekevid

Responsible advisor: Nils-Erik Wiberg, Dept of Structural Mechanics, Chalmers University of Technology

Abstract:
The project will deal with transient/dynamic problems with application to wave-propagation with linear as well as nonlinear behavior due to plastic flow using the FE-technique. Spatial localization of the solution will be studied. Suitable error estimation in space, time and plastic dissipation will be applied in an overall adaptive scheme. Both explicit and implicit schemes are used. Two industrial applications are primarily the target; investigate the wave-propagation applied to rock drilling (Sandvik) and wave propagation in solids due to rapid moving loads resulting in shock waves (Vägverket and Geologi, Chalmers). The project will utilize the supercomputer facilities recently obtained at Chalmers. the project will also utilize the advanced visualization tools at the Department of Structural Mechanics to produce videos in order to understand complicated non-linear physical processes. Cooperation will be established with Chalmers Finite Element Center.

We have contact in adaptivity with Prof. Olek Zienkiewicz and in parallel solvers with Prof Papadrakakis, Athens, one of the partners in an ESPRIT-project.

Related SSF programs:
National Network in Applied Mathematics (NTM) (15 % financing)

7. Computer modelling of interaction between electromagnetic radiation and plasma

Graduate student:Bengt Eliasson

Responsible advisor:Bo Thidé, Wave group, Swedish Institute of Space Physics

Abstract:
The project aims at a novel approach, based on a non-Fourier PDE to ODE conversion technique, for deriving, from the Navier-Stokes equations for electromagnetic interactions, generalise Zakharov equations describing non-linear plasma turbulence resulting from radio wave-plasma interactions, and the use of numerical and algebraic computer techniques to solve these equations. The project is motivated by the need to simulate the extremely complex interaction between string radio waves and plasma in order to better understand the mechanisms of electromagnetic radiation from the turbulent space plasmas first discovered by our group, and to space and communications technology. The recently established AIM graduate school provides a good environment for the strengthening of the use of scientific computing in industry.

Related SSF programs:
Advance Instrumentation and Measurment (AIM)

8. Population dynamics in a heterogenous landscape: Crop management strategies for effective biological control of pest populations

Graduate student:Lars Westerberg

Responsible advisor: Uno Wennergren, Dept of Biology, Linköping university

Abstract:
This project will focus on the dynamics/flow of pest species in farming and their natural enemies. The main goal is to show how different management strategies will effect the size of (damage from) pest populations be the means of changes in habitat quality and availability of natural enemies. This work includes both a spatial and a temporal component: (i) the size of each field and what fields/habitats that surrounds it, and (ii) the longevity of each field and the growth/decline and migration of populations. The habitat quality is a function of abiotic and biotic factors such as the weather, resources as food and nesting sites per individual, and presence of predators. The dynamics of such a system includes feedback from competition and predation within and between populations and a large portion of variation and stochasticity from weather etc. Analytical solutions are not possible and hence simulation techniques are the ultimate solution. The project is linked to a research group at Swedish University of Agricultural Sciences and is made in collaboration with farmers.

Related SSF programs:

9. Model for numerical analysis of deformation and fracture in natural wood

Graduate student:Lars-Olof Jernkvist

Responsible advisor: Per Ståhle, Dept of Solid Mechanics, Malmö University

Abstract:
The project will develop engineering models for studies and prediction of deformation and cracking in wood. An important part of the project is to develop a numerical model for the fracture process. The model must be simple to allow numerical treatment and yet capture the most important characteristics of fracture processes. In particular, the interaction between the wood heterogeneous anatomy and the materials fracture behaviour will be investigated.

Related SSF programs:
Träteknik

10. Mechanical defibration of wood

Graduate student:Svante Widehammar

Responsible advisor: Per Gradin, Dept of Chemistry and Process Technology, MidSweden University

Abstract:
The production of pulp by using thermomechanical pulping is more friendly to the environment than methods based predominantly on chemical pulping technology. However, mechanical pulping consumes large quantities electric energy and continual improvements in the design of process equipment are vital if this form of technology it to remain economically viable

The central ingredient in the process is the refiner. the refiner consist of two circular discs, placed concentrically, one of which rotates (or both rotating in opposite directions). the distance between the discs is of the order of a millimeter. Wood chips enter at the center and are transported out in the radial direction between the discs by centrifugal forces and are defibrated into fibers by contact with the radial bars on the discs.

The scientific goal here is to obtain a better understanding of a complex process by developing and implementing physical, mathematical and computational models of the refiner. A big problem in developing and evaluating the models is the lack of material data. Wood is a very complex viscoelastic material. the long term aim of this project is to fill some of the gaps regarding the material data for wood and to use that data in mathematical models of a refiner.

Related SSF programs:

12. Theoretical studies of photoionization and excitation processes in vacuum, and the effects of surrounding material

Graduate student:Magnus Jansson

Responsible advisor: Leif A. Eriksson, Department of Quantum Chemistry, Uppsala University

Abstract:
The aim of the present project is to gain detailed insight into photoionization and photoexcitation processes of importance in e.g. atmospheric and synthetic organic chemistry. Due to the extreme conditions and short lifetimes of the intermediates occurring in these processes, accurate quantum chemistry calculations are a key tool to understand reaction mechanisms, possible products and their properties. The project consists of three parts: (1) Quantum chemical studies of reactions as outlined above; (2) investigate the accuracy of modern density functional approaches relative to conventional CIS methods, primarily for excited states; (3) Programming of g-tensors for radicals, within the DFT framework.

Related SSF programs:

13. Vortex effects in high-Tc superconductors

Graduate student:Kateryna Medvedyeva

Responsible advisor:Peter Olsson and Petter Minnhagen, Department of Physics, Umeå University

Abstract:
One of the limiting factors for the technological use of the new high-Tc superconducting materials is the dissipation caused by moving vortices. The attempts to overcome this limitation has spurred an intensive research effort into the understanding of "Vortex Physics". Advanced computer simulations have been one of the successful ways of advancing this understanding. The questions attacked in the present project concerns both the nature of the phase transitions caused bet the vortices as well as the complex impedance of the high -Tc superconductor caused by the moving and fluctuating vortices. the research is based on the strong experience of the research group has in advanced simulations of this type.

Related SSF programs:

Projects admitted 1999

1. Accurate finite difference methods for simulation of turbulent flow

Graduate student: Arnim Brügger

Responsible advisor: Dan Henningsson, Department of Mechanics, KTH

Abstract:
In this project high-order discretizations of partial differential equations will be developed for turbulent and transitional flow problems. Time-dependent solutions in two and three space dimensions will be studied. The computational meshes are of curvilinear compo site structured type. High order methods are more efficient and less memory demanding than fi rst and second order methods used in industry today.

The reason why high order methods are not used more frequently is that they i ntroduce a number of complications that have not been fully investigated and it is more complicated to construct stable numerical boundary conditions for high order schemes.

A solver for the incompressible Navier-Stokes equations will be developed for direct numerical simulation of turbulent flow. This code can later form a basis for inclusion of large eddy simulation (LES) models.

2. Diffuse interface methods for multi-component phase change

Graduate student: Irina Loginova

Responsible advisor: Gustav Amberg, Department of Mechanics, KTH

Abstract:
The project is focused on efficient modelling and simulation of large non-li near systems of diverse Itô:s stochastic differential equations: stochastic ordinary differe ntial, stochastic partial differential, stochastic partial integro-differential, etc. These eq uation systems with the corresponding initial and boundary conditions may also include nonstochastic equations and present the most popular stochastic models in applied research and engineeri ng design. The efficiency is achieved due to the following two factors: (i) the simulated e quations are the specifically derived deterministic differential equations for "figures of me rit" of the stochastic- equation solutions like expectations, variances, covariances, spectral densi ties, signal-to-noise ratios, coherence functions; (ii) the developed simulation environment prese nts the low-cost, easy-to-use-and-maintain parallel computing based on the commodity personal computers under Windows 95/98/NT and advanced version 3.4.1 of the public-domain messa ge-passing Parallel Virtual Machine (PVM) software. The proposed project is planned for four years. It is multidisciplinary and closely related to a few current (and some coming) SSF projects. Its outcomes (like a prototype software for the above large stochastic systems) are intended for application in many branches of research and engineering design, for instanc e, semiconductor transistors and circuit, flexible mechanical manipulators, mathematical econ omics and option pricing, chemical kinetics, theory of fluids and phase transitions, biologic al systems, etc. The project provides a lot of highly stimulating, synthetic topics for the stude nt works in scientific and engineering computing. They are related to various fields (parallel comp uting, programming, numerical method, stochastic theory, theoretical physics, and m any others) and ideally suited for student education and research. Participation of students is planned to be the core development component.

3. Molecular modelling of charged chemical gels

Graduate student: Stefanie Schneider

Responsible advisor: Per Linse, Physical Chemistry 1, Lund University

Abstract:
During the last decade, the research and use of charged chemical gels in wat er solution has been intensified. Several academic groups are presently performing fundament al research on the properties of new gel systems and on their response to varying external conditions. However, the theoretical understanding of chemical gels is limited and stems back from the work of Flory. Presently, scaling approaches and simpler analytic theories a re use to describe properties of chemical gels. The aim of this project is to improve the molecular understanding of charged chemical gels by. On the basis of the interaction energies, full statistical mechanical av erages will be performed and all properties are readily extracted. The gel will be modelled within the so-called primitive model, where the cross-linked gel will be represented by charged beads connected with harmonic springs, the salt by small charged spheres, charged surfactant micelles by large charged spheres, and the solvent by its relative permeability. With such a simple model, we will be able to vary gels properties as (i) topology, (ii) mesh size, (iii) linear charge density of the links, (iv) rigidity of the links, (v) the effect polyamfolytic links, and to investigate the response of these gels on (i) addition of salt, (ii) addition of charged surfactants, and (iii) changing the solvent. The project has been discussed with representatives from AstraZeneca and SCA Research and the result of the project would be of direct technological use.

5. Simulation of cellular reactions upon mechanical activation: a spatial and temporal analysis

Graduate student: Charlotte Immerstrand

Responsible advisor: Karl-Eric Magnusson, Division of Medical Microbiology

Abstract:
The objective of this project is to use advanced computer simulations and modelling to analyse cellular behaviour. We will focus on cellular processes, such as receptor-mediated signalling, initiated by mechanical cell activation. Modelling and simulation of cellular events would provide a powerful tool in getting a better understanding of how cells process and respond to different types of stimuli, in our case mechanical stimuli. Mechanical stimulation has the advantage that stimuli application can be defined in time and space, as compared to soluble stimuli, which can be focused in time but not in space. Furthermore, the understanding of cellular response to mechanical forces is highly relevant as cells normally are exposed to, and are dependent on a balanced mechanical load. Tools that enable modelling and simulation of cellular systems would be of great value because of the tremendous complexity of cells, the large number of unknown parameters and interaction between signalling pathways. A model also allows us to gradually increase the complexity of the simulated system.

6. Stress mapping of the human heart - finding appropriate constitutive relations using optimization methods

Graduate student: Jonas Stålhand

Responsible advisor: Anders Klarbring, Department of Mechanical Engineering, Linköping University

Abstract:
Quantification of normal and abnormal motion patterns in the human heart is a quite new research field which has been possible during the last years due to the rapid development in non-invasive imaging techniques and computer hardware and software. Utilising the latest engineering methods from as different fields as applied mechanics and image processing enables us to make a significant impact in the every day clinical practise.

During the last 20 years cardiac investigations has developed from risky and elaborate invasive procedures involving cardiac catheterisation to non-invasive investigations with echo-Doppler and Magnetic Resonance Imaging (MRI). The clinical and socio-economic impact is very large - more accurate diagnoses can be performed faster to a far lower risk and cost. Furthermore, for the next generation in diagnosis tools the trend will be to go from qualitative to quantitative.

The heart is a complex three-dimensional structure with mechanical properties that are anisotropic, non-linear and time-dependent. The thick walls of the left ventricular walls have a helical fibrous structure which varies from a left-handed helix on the epicardium to a right- handed helix on the endocardium. During the cardiac cycle the heart muscle (myocardium) undergoes very large elastic deformations as a consequence of the active muscle contraction along the muscle fibers and their relaxation, respectively.

7. Parallelization of interactive ray-tracing algorithms

Graduate student: Jonas Lext

Responsible advisor: Per Stenström, Department of Computer Engineering, Chalmers

Abstract:
With the emergence of virtual reality applications comes challenging demands concerning realism of computer-generated images. In particular, rendering algorithms such as ray-tracing impose challenging performance demands and are currently not well-suited for animation. The goal of this project is to explore how such algorithms can be transformed so that parallelism can be exploited in commercially available multiprocessor systems. Since multiprocessors leverage on the performance growth of microprocessors, parallelization techniques will be of growing importance i virtual reality applications.

This project will develop parallel algorithms for ray-tracing that particularly focuses on the issues concerned with animation. Important issues to adress are how to adjust the parallelism so that a high average as well as worst-case performance can be obtained across a wide range of 3D scenes. We will evaluate the performance of these algorithms on simulation platforms of multiprocessor systems as well as available platforms at Chalmers (Origin 2000 and Sun Enterprise 10000)

8. Theoretical studies of soft magnetic materials

Graduate student: Massimiliano Colarieti-Tosti

Responsible advisor: Olle Eriksson, Department of Physics, Uppsala University

Abstract:
In applications of magnetic materials energy losses are generated by several mechanisms, the most severe one being due to the anisotropic property of magnetism, i.e. the magnetisation has a certain preferred direction, the so called easy axis. This directional dependence, which is often called the magneto crystalline anisotropy, causes irreversible magnetisation processes in the material and energy dissipates in form of heat, causing large energy losses in for instance transformers and reactors. The microsopical understanding of this phenomenon is not understood at all, although van Vleck at an early stage realised that the directionality must be caused by the relativistic spin-orbit coupling, since it connects spin and space degrees of freedom. Recently tremendous advaces in the theoretical description of these phenomena has been made, and so called first principles calculations (based on density functional theory - DFT) have proven to have a high ability to reproduce the magneto crystalline anisotropy. However, improvements in the numerical algorithms, such as parallelisation of computer codes, resolved k-space convergence etc. must be done to establish a solid theoretical tool for studies of MAE and such an undertaking is part of the proposed project. In particular it is expected that a k-point parallelization of the so called Kohn-Sham equation can be made but also parallellisation algorithms for diagonalization will be persued. The project also aims at studying various alloys, theoreticaly, and search for combinations of materials that produce a low MAE and hence a low loss material for use in the next generation of transformers and reactors. Fe alloyed with Si has experimentally been proven to be effective in reducing the MAE but unfortunately this alloy is brittle at too high Si concentrations and additional substituents for enhancing the brittlenes while keeping the magnetic softness must be studied and this on of the here proposed projects. In addition magnetism under high pressure and magneto optical properties are proposed.

9. Dielectric properties of composite polymeric structures

Graduate student: Yuriy Serdyuk

Responsible advisor: Stansilaw Gubanski, Department of Electrical Engineering, Chalmers

Abstract:
One of the interesting applications in the electrical insulation systems is polymer matrix with inorganic or organic inclusions. Size and weight of the insulation system play an important role in its cost. Possibility of manufacturing light and cheap composites with the same electric properties as the traditionally used materials would be advantageous. In order to tailor the composite materials for electric applications better, it is important to know which properties of their constituents, i.e. bulk and surface electric conductivity, dielectric permittivity, filler concentration, spatial distribution and shape of filler particles, etc. are critical in the overall electrical properties of the composite. These can only be taken into account by means of numerical calculations and literature data on such attempts are not available today. Therefore, it is important to apply new computational techniques in this area. Effects of filler concentrations close to the geometrical limit of volyme fraction will be studied in 3D. A more realistic approach to the modelled electrical parameters will be adopted as well, i.e. the frequency dependence of both conductivity and permittivity, which especially in the low frequency range show different types of behaviour. Finally, interfacial phenomena at the borders between the phases will be taken into account.

The project is jointly supervised by Stanislaw Gubanski (Dept. of Electric Engg) and Anders Bondeson (Electromagnetic Field Theory)

10. Computer simulation studies of enzyme catalysis

Graduate student: Isabella Feierberg

Responsible advisor: Johan Åkvist, Department of Cell and Molecular Biology, Uppsala University

Abstract:
This project is focused on two types of problems in the field of enzyme catalysis. (1) To explain the general principles as well as specific reaction mechanisms of enzyme catalyzed keto-enol isomerization reactions, which involve proton abstraction from very non-acidic carbon atoms. (2) To explore the scope of combining centroid path integral calculations with molecular dynamics based simulation techniques for simulating enzyme reactions, where the reliability of evaluating kinetic isotope effects and tunnelling contributions to rate constants will be assessed. The project will also involve optimization and further development of our recently developed software.

11. Visualization in structural dynamics and coupled problems

Graduate student: Anders Olsson

Responsible advisor: Göran Sandberg, Division of Structural Mechanics, Lund University

Abstract:
The proposed project is concerned with scientific visualisation in computational mechanics, particularly visualisation of coupled phenomena i.e. fluid-structure interaction. Other keywords are distributed strategies for monitoring ongoing computations or real-time visualisation. The project will be based on an open class library for visualisation.

Special emphasis will also be put on how visualisation can be performed in order to enhance the understanding of the underlying physics treated in the simulations.

12. Analysis and optimization of conformal base station antennas

Graduate student: Per Ingelström

Responsible advisor: Anders Bondesson, Department of Electromagnetics, Chalmers

Abstract:
This application concerns analysis and optimization of conformal array antennas for next generation of base station antennas. The first part of the project consists of developing analysis tools for electrically very large and geometrically complex electromagnetic problems. We have already developed a hybrid code combining the efficient FDTD in homogeneous regions with the superior ability of unstructured FEM to model complex boundaries. This code will be used for antenna analysis and optimization. Under the present project, a new hybrid between the Method of Moments and FEM will be developed. Such codes can be very efficient for certain geometry's of high relevance to base station antennas.

Base station antennas need spatial adaptivity to increase the ability to distinguish between wanted terminals and interferers. In many cases, an antenna will also have to be conformal to some structure, either to fit into a predesigned environment, or simply to have a sufficiently large field of view. Such conformal antenna arrays are geometrically complex and electrically large and are therefore need highly performant codes for numerical analysis.

The application oriented part of the project will start by a simplified study to find a suitable type of geometry, the number and approximate placing of elements, etc. In this study, we will analyze how the antenna selectivity can be optimized using a moderate number of digital channels, ignoring interelement coupling. Once a suitable basic configuration has been found, detailed CEM modeling and optimization will be made using "exact" tools. An effort will be made to use modern optimization techniques for antenna design.

Analysis and optimization of base station antennas is of utmost importance for Swedish telecommunication industry. The project has been discussed extensively with the Antenna department at Ericsson Microwave Systems and is relevant to the long-term research programmes both of EMW and Allgon.

13. Molecular control systems in living cells

Graduate student: Johan Elf

Responsible advisor: Måns Ehrenberg, Department of Cell and Molecular Biology, Uppsala University

Abstract:
We will use stochastic equations to model biosynthetic pathways in E. coli leading to the production and consumption of amino acids by their incorporation into proteins. We will study the various molecular control mechanisms used by bacteria to steer the process both in steady state and under varying external conditions.

For the stochastic descriptions of these biochemical pathways in single cells we will use Master equations for the probability distributions of numbers of molecules in the various pools. The Master equations will either be intergrated numerically using standard methods like Euler forward or Runge-Kutta integrations. Alternatively, when the dimensionality of the Master equation becomes too high, it is more suitable to solve them with modified versions of the so-called Gillespie algorithm.

14. Mathematical models and computational tools for the analysis of microbial genomes

Graduate student: Daniel Dalevi

Responsible advisor: Siv Andersson, Department of Molecular Evolution, Uppsala University

Abstract:
We are actively engaged in a variety of genome sequencing projects on microorganisms that live in close association with higher eukarytes, such as for example parasites of humans (Rickettsia and Bartonella) and symbionts of insects (Buchnera and Wolbachia). The analysis of complete genome sequence data places demands on the mathematical and computational competence of the genome sequencing teams. The development of efficient methods and user- friendly tools for the analysis of complete genome sequence data is therefore of broad interest to the biological community.

The objectives of this project is to further develop methods for the identification and annotation of genes in newly sequenced microbial genomes as well as to refine existing methods for studies of genomic architectures. The methods will be applied to the sequencing data generated by us within the SSF-funded Program for Genome Research. One of the projects will be to analyse microbial genome sequence data collected from 4 different species of the genus Buchnera. These genomes constitute an excellent model system for studies aimed at calibrating the rates at which genomes rearrange. Such calibrated measures of deletion, insertion, inversion, translocation and horizontal transfer events can be used to create realistic models of how microbial genomes envolve in both the short-term and the long-term. The availability of realistic models for the evolution of microbial genomes is of importance for our understanding of the factors which determine fixation rates for horisontally transfered genes, such as for example antibiotic resistance genes, in microbial populations.

15. Visualization and computational steering of space plasma simulations

Graduate student: Patric Ljung

Responsible advisor: Anders Ynnerman, Department of Science and Technology Norrköping, Linköping University

Abstract:
A new research group in scientific visualization is headed by the applicant. The present proposal builds on an interdisciplinary interaction between space plasma simulations and development of techniques for visualization. The project will thus lead to important results in both these areas. A parallel version of a Particle-In-Cell (PIC) 2-dimensional electromagnetic simulation code has been developed by the applicant and its use has already led to important results in space plasma physics. In the present project the code will be used to study some key problems involving the sounding of plasmas and studies of noise levels in PIC simulations using advanced visualization techniques such as volume rendering of phase space structures and computational steering in visual environments. Additionally the simulations will be used to study topics in visualization such as interaction environments for 3-dimensional graphics and use of virtual reality techniques in scientific visualization. The research will be performed in collaboration with the space and plasma physics group at the University of Warwick, England.

Projects admitted 2002

1. Modelling column packing for chromotagraphy

Graduate student: Lars Gustavsson

Responsible advisor: Alf-Erik Almstedt, Department of Thermo and Fluid Mechanics, Chalmers

2. Dynamic models of genome growth and shrinkage

Graduate student: Mats Pettersson

Responsible advisor: Otto Berg, Evolutionary Biology Centre, Uppsala University

3. Optimization methods for electromagnetic devices

Graduate student: Tomas Halleröd

Responsible advisor: Anders Bondesson, Department of Electromagnetics, Chalmers

4. Electronic structure theory of solids and surfaces

Graduate student: Björn Skubic

Responsible advisor: Olle Eriksson, Department of Physics, Uppsala University

5. Continental-scale modelling of the water balance

Graduate student: Elin Widén

Responsible advisor: Sven Halldin, Department of Earth Sciences, Uppsala University

6. Contact and crack mechanics

Graduate student: Per Heintz

Responsible advisor: Peter Hansbo, Department of Applied Mechanics, Chalmers

7. Efficient numerical methods for integro-differential equations - with application to ski friction

Graduate student: Anderas Almqvist

Responsible advisor: Roland Larsson, Division of Machine elements, Luleå University of Technology

8. Developing covariation and other novel methods for the analysis of protein family evolution

Graduate student: Lena Milchert

Responsible advisor: David Liberles, Stockholm Bioinformatics Centre

9. Solution of the master equation in molecular biology

Graduate student: Paul Sjöberg

Responsible advisor: Per Löstedt, Department of Scientific Computing, Uppsala University

10. Structure and dynamics of protein-nucleic acid complexes

Graduate student: Katarina Lindberg

Responsible advisor: Lennart Nilsson, Department of Biosciences at Novum, Karolinska Institutet

11. Structural dynamics and stability of RNA

Graduate student: Boel Nyström

Responsible advisor: Lennart Nilsson, Department of Biosciences at Novum, Karolinska Institutet

12. Numerical treatment of photodissociation reactions

Graduate student: Erik Abrahamsson

Responsible advisor: Gunnar Nyman, Department of Physical Chemistry, Göteborg University

13. Microwave tomography for breast cancer detection

Graduate student: Andreas Danielsson

Responsible advisor: Mikael Persson, Department of Electromagnetics, Chalmers

14. Modelling the atmosphere of the coastal zone

Graduate student: Karin Törnblom

Responsible advisor: Ann-Sofi Smedman, Department of Earth Sciences, Uppsala University

16. Solid state proton conductors

Graduate student: Per Sundell

Responsible advisor: Göran Wahnström, Department of Applied Physics, Chalmers

17. Computational road mechanics-a coupled thermo-hydro-mechanical problem

Graduate student: Gustav Engström

Responsible advisor: Nils-Erik Wiberg, Department of Structural Mechanics, Chalmers

18. Methods for visualization of climate change

Graduate student: Björn Olsson

Responsible advisor: Anders Ynnerman, Department of Science and Technology Norrköping, Linköping University

19. Temporal electron dynamics in semiconductor nanostructures

Graduate student: Martin Evaldsson

Responsible advisor: Igor Zozoulenko, Department of Science and Technology Norrköping, Linköping University

20. Theoretical simulations of drug solubility

Graduate student: Laban Pettersson

Responsible advisor: Hans Ågren, Department of Theoretical Chemistry, KTH

21.

Graduate student:Jenny Lindau

Responsible advisor:Department of Chemical Engineering & Environmental Science, Chalmers

22.

Graduate student:Karljohan Lundin

Responsible advisor:Anders Ynnerman, Department of Science and Technology Norrköping, Linköping University

Sections

Personal tools

NGSSC Project Catalog

NGSSC graduate student projects

Projects admitted 1996

Projects admitted 1997

Projects admitted 1998

Projects admitted 1999

Projects admitted 2002 (No abstracts available yet)

Projects admitted 1996

Projects admitted 1997

Projects admitted 1998

Projects admitted 1999

Projects admitted 2002

Document Actions