[gmx-developers] accessing atoms position with domain decomposition

[Massimiliano Bonomi]

Dear Berk,

thanks for your suggestion.
I will examine the pull code in detail!

> But if you operate on very many atoms, this can be an expensive  
> operation, since it requires
> access to a global index over the whole system.
This should not be the case. Except for water-coordination collective  
variables,
few atoms positions should be required...

> If you are lucky, your algorithms might work similar to the pull code,
> so you can look at how the pull code is implemented with domain  
> decomposition.
>

Thanks again,
Massimiliano