Understanding how atoms and molecules move and interact is crucial for the fundamental understanding of any chemical process. Advanced mathematical algorithms make it possible to simulate and understand the interactions.
A chemical reaction in a laboratory involves billions of atoms and molecules that collide and make new molecular species. If one would be able to zoom in at the Ångstrom level detail or 0.0000000001 meter, one would be able to identify individual molecular collisions.
If the collision happens at the right strength and at exactly the right moment, when the molecules have the perfect orientation, a chemical reaction could occur.
Molecular simulations provide us ‘a virtual microscope’ which allows us to study these collisions. PyRETIS, a project at the NTNU, combines the elements from two main simulation methodologies. This makes it easier to study these phenomena in realistic computer experiments as both standard and rare molecular interactions are simulated.
MD or MC that is the question! ….But why not both?
The research field of molecular simulations consists of two main methodologies: Molecular Dynamics (MD) and Monte Carlo (MC). Molecular Dynamics tries to mimic the natural motion of atoms and molecules. Once the initial atom positions and velocities are set, the simulation evolves largely deterministically. Monte Carlo on the other hand is based on random numbers. Mathematicians called it after the Monaco’s district because of its famous casinos (See picture below from the writer’s last summer holiday).
The association with gambling or rolling the dice reflects the probabilistic nature of the method. At the end of previous millennium, scientists have also realized the strength of combining Monte Carlo with Molecular Dynamics especially for the study of rare events.
Well-done, medium or rare? How rare are “rare events”?
Here, we should probably point out what is meant with a rare event. A rare event in molecular simulations is not so rare as you might think. A chemical process that happens a million times per second is already considered to be ‘rare’ by a scientists working with molecular simulations. That is because there is a huge difference between real time and CPU time.
One molecular simulation describing 1 picosecond real time (or 0.000000000001 seconds) typically takes a day of computation. This number is somewhat arbitrary since it depends very much on the computer hardware that is used, the number of molecules in the system, and the level of accuracy that is applied in order to determine the forces between the molecules. In any case, a reaction, that is expected to happen once every second, will demand centuries of computation time to be produced in an accurate simulation.
This doesn’t mean that nothing can be learned from the computer-produced molecular movies. Even in a virtual movie of only a few picoseconds one can see a lot of molecular collisions. Unfortunately, the chance that a collision produces a chemical reaction is extremely small…but if it happens…it happens fast.
Pirates …no…PyRETIS …of the Caribbean
Replica Exchange Transition Interface Sampling or RETIS is the algorithm that can generate transition trajectories of the molecules and compute the rates. The approach consists of a set of different Monte Carlo moves in combination with Molecular Dynamics.
The nice thing of the method is that it provides the same information as endlessly long Molecular Dynamics simulations. This can be explained as follows: Just consider a movie that is absolutely boring in 99.9999% of the time, but sometimes there is an interesting scene (for example a murder or the main actors falling in love) that only takes a split second. The RETIS approach allows you to cut out the interesting scenes and gather them by a statistical method, which also tells you how likely it is that the scene will occur.
In my group we have started the PyRETIS project for developing an open-source computer program based on Python. Python is an easy-to-learn programming language, which allows constructing readable complex object-oriented computer codes. At the NTNU, Python is presently also more and more used in student courses for programming exercises. It was, therefore, a logical development to make the PyRetis project not only a research-oriented project, but also a project that could serve teaching purposes.
We were, therefore, very happy to hear that the Olav Thon foundation decided to sponsor our project. I tell you more about this in another blogpost.
This blogpost is written by Titus van Erp, Associate Professor at the Department of Chemistry