Re: [AMBER] Question regarding ntx and irest

From: David Case <>
Date: Wed, 25 Sep 2019 13:43:35 +0000

On Wed, Sep 25, 2019, Gustaf Olsson wrote:
>I recently got involved in another persons project and we were going to re-evaluate some old simulations. In doing so we basically ran the old jobs though I found it interesting that there were “impossible” settings that stil ran.
>In the old files these were the settings:
> irest = 1,
> ntx = 7,
>Then I noticed
>that switching to ntx=1 produced the same exact error while using ntx=5
>produced results as ntx=7.

Nowadays, ntx=5 and 7 are identical: the code figures out if it needs
box information or not based on other information. Also, the format is
auto-detected, so the old "even/odd" distinction is no longer relevant

Bottom line, use ntx=1 to read just coordinates, ntx=5 to read
coordinates and velocities.

The irest variable determines *how* those input velocities are used: if
irest=0, even though you read in velocities, they will be ignroed; if
irest=1, the input velocities must be present (i.e. you must also set
ntx=5), and will be used to continue the simulation from the end point
of the last run.

....hope this helps....dac

p.s.: I have no real idea what happens when ntx is not 1 or 5 (which are
the only two values documented in the Reference Manaul (section 17.6.2).
We really should not be allowing any other values, although I know we
allow 7 for backwards compatibility with older versions of the code.
Also, I'd have to study code to say what used to happen in older versions.

AMBER mailing list
Received on Wed Sep 25 2019 - 07:00:02 PDT
Custom Search