Re: [AMBER-Developers] namelist

From: Carlos Simmerling <>
Date: Wed, 17 Mar 2010 18:20:15 -0400

that's what I meant when I said to set a default and see if it's changed. I
put that in for now- but ideally we'd have this for about 20 variables so
it's just tedious. for now i decided to only do it to the few that must be
checked (default values depend on other variables in the namelist), and for
the others make it clear in the manual that some variables will have no
effect depending on others. we do that already, I guess, since you can set
coupling constants without a thermostat, but in this case I think people are
more likely to be confused since some of the GB variables will not have the
intended effect in a different GB model.

On Wed, Mar 17, 2010 at 6:09 PM, case <> wrote:

> On Wed, Mar 17, 2010, Carlos Simmerling wrote:
> > does anyone know if there is a way to know if a user specified a variable
> in
> > the mdin namelist? I'm wondering about making it bullet proof- make sure
> a
> > user doesn't set something that they should not.
> Well, you can check to see if the variable has changed from its default
> value,
> which can only happen if the user set it. Here's some pseudo-code:
> ! set defaults
> gbalphaSB = 5.2d0
> ....
> read (5,cntrl)
> ....
> if (igb .eq. 8) then
> if (gbalphaSB .ne. 5.2d0) then
> ! print warning/error message, exit if this is fatal
> ! and/or reset the value back to its default, etc.
> endif
> endif
> This won't catch the fact that the user might have set gbalphaSB to its
> default value, but that shouldn't do any harm.
> > or should I assume that if I say in the manual that a
> > variable is ignored unless you run a certain type of calculation, that
> it's
> > ok to silently ignore a user's request to change the value?
> We certainly don't trap all irrelevant variables at present, so I wouldn't
> feel bad if you add some new "silently ignore" situations.
> [I'm not sure if the above is what you are thinking about or not....]
> Once the dust settles from Amber11, a good task for someone (!?!) would be
> to add more organization and more intelligence to the parts of mdread.f
> that
> try to find bad input. We could have a "expert" mode that would prevent
> lots
> of changes to the defaults unless it were set. There's a little bit of
> this
> in the manual, where users are warned about changing parameters that are
> not
> in bold-face, but I think few people either know about that, or try to
> follow
> it.
> ....dac
> _______________________________________________
> AMBER-Developers mailing list
AMBER-Developers mailing list
Received on Wed Mar 17 2010 - 15:30:03 PDT
Custom Search