compute temp-Temp/region

Dear Lammps-users,
Hi,
I have a question about computing temperature in lammps.
I need to calculate temperature of rigid water in a region of the simulation box.
Therefore I need to define temp in the way that exclude some degree of freedom.It seems to me that after this definition temp/region command still uses the default temp definition.
Because with and without my temp definition thermo_style prints out the same temperature.
Does anyone have any suggestion ?
The related lines of input is as follows:

From the compute temp/region doc page:

Unlike other compute styles that calculate temperature, this compute does not subtract out degrees-of-freedom due to fixes that constrain motion, such as fix shake and fix rigid. This is because those degrees of freedom (e.g. a constrained bond) could apply to sets of atoms that straddle the region boundary, and hence the concept is somewhat ill-defined. If needed the number of subtracted degrees-of-freedom can be set explicitly using the extra option of the compute_modify command.

If you can estimate the number of water molecules in the region, you

can use compute_modify extra to subtract the appropriate DOF.

If someone can suggest a better way to do this, I suppose an alternative

could be implemented.

Steve

From the compute temp/region doc page:

Unlike other compute styles that calculate temperature, this compute does
not subtract out degrees-of-freedom due to fixes that constrain motion,
such as *fix shake* <http://lammps.sandia.gov/doc/fix_shake.html> and *fix
rigid* <http://lammps.sandia.gov/doc/fix_rigid.html>. This is because
those degrees of freedom (e.g. a constrained bond) could apply to sets of
atoms that straddle the region boundary, and hence the concept is somewhat
ill-defined. If needed the number of subtracted degrees-of-freedom can be
set explicitly using the *extra* option of the *compute_modify*
<http://lammps.sandia.gov/doc/compute_modify.html> command.

If you can estimate the number of water molecules in the region, you
can use compute_modify extra to subtract the appropriate DOF.

If someone can suggest a better way to do this, I suppose an alternative
could be implemented.

​one way to do this would be to add a "dof" ​property to each atom
defaulting to 3.0 (or 2 in 2d sims) and then subtract the DOF fraction that
is removed from it (so 0.5 DOF for a bond contstraint, 1/3 for an angle, it
gets a bit messy for (linear) rigid bodies...)

axel.

Hi Steve and AXel,
Thank you for your advice.

I did not understand what Axel mean as I am new in lammps.

I think Steve means the below modification, in hot region for example, that it does not work for me.

Hi Steve and AXel,
Thank you for your advice.
I did not understand what Axel mean as I am new in lammps.
I think Steve means the below modification, in hot region for example,
that it does not work for me.

​nobody can help you based on such vague terms as "it doesn't work (for
me)" ​and a minimal fraction of your input.

if you believe that LAMMPS is not doing something the way it is advertised
in the documentation (or the way you think it should be done),
then you have to provide a small demosntration input that illustrates your
claim and then explain what you see and what you think you should see and
why you believe that it should be like this. double check with the
documentation.

axel.

Dear Axel,
Thank you again.
Yes you are right and I did not send you enough data.
Regards