I'm trying to carry out metadynamics for a system, where my collective variable is bond length. Is there way to decide how to set values for height and width ? I tried with two different values and plots are completely different .
Thank you.
Smaller width and height should give more accurate results, at the cost of efficiency. Without knowing further details of your system, I'd recommend reducing the parameters until the plot (of free energy, I assume) doesn't change significantly.
In well-tempered metadynamics, the height and width of the Gaussian potential are dynamically adjusted during the simulation based on the history of the system. This helps to enhance sampling of the free energy landscape by reducing the bias introduced by the added Gaussian potentials.
When you start the simulation, you need to specify initial values for the Gaussian height and width. The choice of these values can affect the convergence rate and efficiency of the simulation, but there is no general rule on how to set them. It is recommended to perform some test simulations with different values of height and width to determine the optimal values for your system.
Typically, the Gaussian width is chosen to be larger than the expected range of the collective variable fluctuations. A larger width helps to smooth out the Gaussian bias and enhance sampling. The Gaussian height, on the other hand, should be chosen such that the added potential is not too strong to prevent the system from exploring other regions of the free energy landscape.
You may also consider setting an adaptive Gaussian height, which adjusts the height based on the bias potential accumulated during the simulation. This can be done by setting a negative value for the height in the mdp file, e.g. "g = -0.1", which will trigger the adaptive Gaussian height feature.
In summary, the choice of Gaussian height and width in well-tempered metadynamics depends on the system and the level of sampling required. It is recommended to perform multiple test simulations with different values to determine the optimal parameters for your system.
Satyendra Singh, Jyri Kimari Thank you so much. While I understand what you are saying, the problem I'm facing is there is no way any convergence in my plots. I'm not even sure for how long I have to run the simulations. With run time varying, plots are changing. With width and height change, plots are changing. Any suggestions regarding that ?
- Badges
- Science topic
- Similar topics
- Gels