[gmx-users] refining a TI calculation - additional lambda points

[Hannes Loeffler]

On Tue, 26 Jan 2016 15:47:03 +0100
Oliwia Maria Szklarczyk <oliwias at ethz.ch> wrote:

> Dear All,
> 
> I was wondering if you have tips on how to easily create additional
> lambda points in a thermodynamic integration calculation in gromacs
> 5. I have 21 lambda points already simulated, from L_0 to L_20, and
> now I need six more lambda-points between L_16 and L_19. 

As long as you are only interested in TI you can just add points as you
want.  You could, for example, just make a new mdp file with a new
lambda vector only having these points or reuse the old one with adding
the new points in.  That doesn't really matter much.

The practical issue is more how you combine the resulting data such
that your analysis program of choice can use it.  But it shouldn't be
too difficult to quickly script a solution to do that yourself.   You
will want to write the new TI data into a separate xvg file.

If you intend to do a FEP style analysis, however, you need to consider
that you would need to compute energies for at least the nearest
neighbours.  For BAR e.g. you would need to recalculate the windows
bordering L16 and L19 for full MBAR, in principle, all windows...


Cheers,
Hannes.