Re: [AMBER] values of λ in Thermodynamic Integration

From: David A Case <>
Date: Fri, 25 Jul 2014 07:17:04 -0400

On Fri, Jul 25, 2014, Sun wrote:

> In Thermodynamic Integration calculations, several MD simulations with
> different λ (lambda) values will be performed, for example, for a
> 5-point TI calculation , the values of λ could be 0, 0.1127, 0.5,
> 0.88729 and 1 ( this is the stratege applied in the paper-- "Proton
> Binding to Proteins: pKa Calculations with Explicit and Implicit Solvent
> Models", JACS 2004, 126, pp4167-4180).However,when λ equals to 0 or
> 1, the cooresponding weight used in Gaussian intergration is 0, so
> these two points actually do not contribute to the final result.

I think you are misreading Table 1 of this paper. The CHARMM calculations
used lambda at 0, 0.5 and 1, and did the integration with the trapezoidal
rule. The Amber calculations used 3-point Gaussian integration with lambda
values of 0.112, 0.5 and 0.883. With Gaussian integration, you don't need to
use the endpoints.

> ps: In the Amber online tutorial ( tutotial A6), it is said that the end
> point simulations will be helpful to debug when things go wrong. How to
> understand this ? What useful informations will the end point simulatons
> provide when dubug ?

The endpoint calculations are equivalent to "standard" calculations of a
physical system. If there were some problem with the TI part of the code, it
might be helpful to compare with (say) a completely different code for the
endpoint simulations. In practice, we (now) have extensive experience with TI
calculations, and it is unlikely that this sort of debugging would be needed.


AMBER mailing list
Received on Fri Jul 25 2014 - 04:30:02 PDT
Custom Search