Diploma and Project Work Opportunities

Updated: 2017-08-12Contact point for these projects: ricar [at] ifm.liu.se

I can offer a large range of project work related to my own research. This page contains shorter projects first, and then (below) lists projects suitable on Master's theis level.

A short overview of my research for non-experts: I'm working in the field of high-throughput computation of material properties. I'm both involved in developing new theoretical methods for such computations, and in using these methods to discover (or predict discovery) of new materials using theoretical methods.

In essence, we make up 'theoretical' crystal structures (for example, by adding / removing / exchanging atoms in known materials). Then we run computer software that solves the underlying quantum mechanical problem for the electrons in these materials, and this way we can predict if any of these materials could possibly be created in reality, and if so, what properties they would have. This way we can search for materials that could be created for use in, for example, the next generation of solar cells, batteries, and similar. This is, in a way, a brute-force method for discovering new materials.

Smaller Projects (Suitable for project courses, 1-2 weeks of work or more.)

Software-related projects

Contact point for these projects: ricar [at] ifm.liu.se

I lead the initative behind the Open Materials Database, a searchable archive of computational results for materials. It is built upon a scientific software framework of our own design written in python, the High-Throughput Toolkit . This is a tool we actively use in our research.

We have a large list of smaller projects for extending this framework and the website which belong to the interface between science and software development. These are subject to change, but a present list of possible project ideas follows:

• Connect our toolkit to improved visualization tools. We already have some basic visualization provided via Jmol. But, our research group is connected to the developers of the Interactive Visualization Workshop (Inviwo), and this project is to set up a first 'connector' between our framework and Inviwo that provides a simple crystal structure visualization.
• Visualization of crystal structures in ray-tracing software. Another visualization feature that would be immensely useful for our framework is the ability to generate extremly high-quality figures using ray-tracing imaging. This project is to build a 'connector' from our framework to generate visualizations in the ray-tracing software package powray.
• Better support for molecules in our framework. Our framework was built to treat periodic crystals but it has a basic support for non-periodic systems. This project is to look at how to how molecues should be represented in our database, what tools will be needed to be built to operate on molecules with the same flexibility as we have for crystals, and how to visualize them with the same tools that we use for periodic crystals.
• Support for surface physics. Presently the framework is mostly focused on bulk crystals and bulk crystal properties. This project starts to look at the extension of our tools into generating surface models and to interface with computational software to find their properties. While this in total is a major undertaking, clear small project tasks can be formulated within this effort.
• Implement a new type of computation for predicting materials properties. Today there exists a large amount of computational software for predicting a wide range of properties of materials. So far support for only a limited set is supported via our framework. We are continously looking to expand this set so that new types of materials property calculations can be generated and analyzed. This project requires one to dig down into the rather complex software that we use for running fault-tolerant computational tasks across supercomputers, and then implement a new calculation type from our wishlist of the more 'easy' types.
• Automatic phase diagrams. The framework presently has some support for phase diagrams, and the database contains information needed to generate phase diagrams for a number of materials. This project is to write software to automatize generation of phase diagrams for a large set of our materials and make them available on the website.
• Extended search capabilities. Presently the website exposes a tiny fraction of the possibilities for 'searching' that exist within our database. This project is to look at extending the search types in a user and researcher-friendly way. Note that this is not meant to be software development-only project. To understand what and how this information should be searched requires understanding both of what the information means and how researchers are using it.
• A first basic GUI for automatized calculations. The framework presently allows for setting up computations, retreive the results and analyze them with command-line tools. This project is to start up the development of a simple GUI with similar capabilitires. Note that this is not meant to be software development-only project. To understand how to build a useful tool and the typical workflow will require a basic understanding of these physics-based software packages.
• Expose our data in an IOS or Android app. The website works fine for mobile users, but it is a goal for universities in Sweden to expose research to the broader public. This project starts looking at creating an "app" that exposes materials science data and visualization with a certain 'wow' factor for the general public. By utilizing the website and one of several "quick app developer" tools available, we expect getting a first prototype up and running fits within the time-frame of a smaller project.
• Extend the import / export capabilities. We are presently rather limited in the types of file formats that can be read/written via the framework. This project is for someone techincally-inclined who would find it interesting to dig down in various crystal structure and electronic structure file formats and look at writing converters / readers / writers for such data. (To understand scientific data formats typically requires a level of understanding of the science beind them.)
• Any idea of your own! Take a look at what we offer at the website and with the toolkit. If you have an idea of some feature that would be interesting to extend, or add, feel free to discuss your idea for a project with us!

Larger Projects (Master's thesis, etc.)

Projects in applied high-throughput computation

Contact point for these projects: ricar [at] ifm.liu.se

These projects involve running large-scale database-driven computations, using state-of-the art computational software and working with collections of thousands of material candidates. The project will involve work on software development on our high-throughput computational framework for automatic job creation, submission, retrieval, collection into databases, and automatic analysis.

1. High-throughput search of promising piezoelectric fluoride perovskites. A vast chemical space of possible perovskite-type structures will be screened for materials with beneficial properties to work as modern high-performing peizoelectric materials. Modern computational methods can identify known perovskite oxides currently used or proposed as piezoelectric materials. However, the space of perovskite oxides have more or less been exhaustively explored. This project aims at investigating the closely related, but much less known, space of perovskite fluorides. The project may further expand into investigating other properties of perovskites, e.g., band gaps suitable for solar applications in both oxides and fluorides. The project will be done together with Rickard Armiento at IFM and in collaboration with Igor Abrikosov at IFM, Marco Fornari at Central Michigan University, USA, and Boris Kozinsky at Robert Bosch LLC, Cambridge, USA.

2. High-throughput estimate of formation energies for rapid assessment of new materials. To be able to quickly estimate the formation energy of a system to find what materials are at all likely to be formable is a key component of high-through computational methods. The project aims at evaluating a few very fast methods for such estimates. We will test a few effective potential-based methods, orbital-free DFT, and exciting new machine-learning methods. For a PhD project this could either expand into further developing and improving such techniques and/or their application to our high-throughput problems. The project will be done together with Rickard Armiento at IFM and in collaboration with Igor Abrikosov at IFM, and, for the machine-learning part, in collaboration with Anatole von Lilienfeld at Argonne National Laboratory, Illinois, USA.

Projects in Density Functional Theory, Functional Development

These projects involve in-depth theoretical work in quantum physics, mathematical physics, and to some extent numerical methods with the aim of improving the theoretical methods used to, e.g., predict material properties using computer simulations. Some projects also involve programming and running modern computational software to implement and test the theoretical results.

This category of projects change frequently, email me for suggestions.