Press "Enter" to skip to content

Research and personal interests

This is the serious part of my site concerning my research topics and interests. I’m (generally) happy to talk about them so feel free to ask me anything. I also use this page as a bibliographical repository for future works.

Molecular simulation

My initial background is in physics and statistical physics. I also enjoy programming and a direct application of the latter to the former is molecular simulation, either through Monte-Carlo simulation1 I find amazing that this paper cited more than 42k time has only 2 references. or Molecular dynamics. Both of these methods are about exploring the phase space of particles systems in different ways which are, in my opinion amazing intellectual constructions (both the algorithm and the underlying theory). I tend to focus more on the following topics:

  • Algorithm comparison: My PhD topic was about modelling semi-crystalline polymers’ interphase in order to compute mechanical properties. One way of doing it is through correlations of each term of the strain tensor of the simulation box in the NPT ensemble. However we showed that coupling parameters of the Nosé-Hoover algorithm (as used in LAMMPS) could significantly affect the results. It is much slower to converge (if it converges at all) than in the Monte-Carlo simulation which explores the box shapes in a statistical manner (no explicit barostat). Recently probabilistic barostat algorithms were also proposed for MD simulations. As presented, they are supposed to sample the phase space better than Berendsen (obviously) and Parrinello-Rahman/Nosé-Hoover implementations. A true proof of concept would be to compute anisotropic material mechanical properties with the same reliability than MC simulations.
  • Local mechanical properties: My initial PhD objective was to compute mechanical properties of a biphasic system. We expected to compute « local » mechanical properties in both phases of a semi-crystalline model. However it turns out that, as an intensive quantity, the « local » character of stress (and hence its derivatives with regard to strain) is an ill defined concept, especially at the atomistic level. Discussions have tried to clarify it and offer different mathematical derivations in order to compute a local quantity that could be assimilated to a stress tensor and hence mechanical properties. But this is actually still open and not entirely solved in the molecular simulation community. While mechanical properties of complex materials will be more and more investigated through simulation, this is an interesting topic to investigate.
  • NEMD: I’m also very interested in the work of Billy Todd and Peter Daivis. Especially concerning the implementation and use of NEMD derived equation such as SLLOD algorithm in the following topics

Polymer simulation

  • Polymer are molecules with the particularity to be constituted of the same repetition of monomeric units. This gives them special spaghetti like behaviour and properties which reflects in the materials properties. Unfortunately,

To be continued…