Modeling Concepts - Conformational Search

We have seen that Minimizations can lead to local minima. Coordinate Scans help and are great for molecules with one or two key variables but what what about molecules with multiple rotatable bonds? It turns out that coordinate scans are really inefficient.

Let's say you have a molecule with 10 rotatable bonds and you choose to scan them. If you chose 10 degree increments, then each bond would be scanned for 360/10 = 36 increment steps. For 10 bonds, that is 36^10 individual calculations! Part of the problem with scans is that all states, low energy and high energy, are considered equally.

So how do we find the global minimum energy geometry? Indeed, we are often interested to know all of the states accessible to a molecule in a given temperature range. The best tool for both of these goals is a Monte Carlo Conformational Search.
Potential Energy Surface
First recall our basic PES. The MC protocol is an iterative process. You do step 1 and the algorithm does the other steps interatively:

1. Input an initial geometry.
2. Rotate a dihedral angle (or several) by random amounts.
3. Minimize the new structure.
4. If the new structure is really high in energy, toss is out.
5. Compare with prior structures. If the new structure was found already, toss it out. If it is unique, keep it.
6. Go to step 2.

Potential Energy Surface - Monte Carlo
The calculation iterates for a specified number of cycles. This process is essentially a statistical sampling method and one must iterate extensively enough to allow adequate sampling of the PES. In practice, there are empirical guidelines as to how many iterations are adequate, based on the number of rotatable bonds. When you use this protocol, you must evaluate any program default settings and decide if the specified number of cycles is adequate.

One of the reasons we use the Monte Carlo Conformational Search method is that there is a bias to spending more time in low energy regions, and the calculations are thus more efficient than coordinate scans.

Excercise 8. How do you know if a Monte Carlo conformational search has actually found the global minimum energy structure?
When you are finished, return to the main page to review or find any other sections you may have missed. There is also a summary of exercises.