[lammps-users] Thermal conductivity computation

Please post your Qs to the mail list, not me directly,
so others can comment.

1) the header of the file tells you what's in the columns:

# TimeStep Number-of-layers
# Layer Coordinate Natoms v_temp
1000 20
1 0.025 150.9 1.48848

4 columns are layer, coord, # of atoms, v_temp

2) You should get a V-shaped profile at any point in
time. You are getting one from 1.4 to 1.6 roughly.
You're not swapping enough KE to get a stronger V.
See the MP papers for more discussion.
You typically have to average the profile over many
outputs to get a smooth profile.

3) The heat flux is something LAMMPS computes
for you and stores with the fix. See the fix thermal/conductiity
command for how to get at that value and print it out.

Steve


Steve,
I increased the kinetic energy swap rate to 10 timesteps (50 timesteps). The V-shape is shown in the attached Excel file for two time instants (25000 and 30000 th time steps). Would you agree they are starting to be more linear?
I am aiming to replicate some of the 1997 J.Chem. Phys. Muller-Plathe paper results. In that paper, an orthorhombic periodic cell is used. In order to create such a cell, would a “custom” lattice style have to be used?

Thanks,
Sreekant

— On Tue, 1/20/09, Steve Plimpton [email protected] wrote:



> From: Steve Plimpton [email protected]
> Subject: [lammps-users] Fwd: Thermal conductivity computation
> To: “Lammps” [email protected]
> Date: Tuesday, January 20, 2009, 8:32 AM
>
> <br>> ---------- Forwarded message ----------<br>> From: Steve Plimpton <[email protected]><br>> Date: Tue, Jan 20, 2009 at 7:57 AM<br>> Subject: Re: [lammps-users] Thermal conductivity computation<br>> To: Sreekant Narumanchi <[email protected]...><br>> <br>> Please post your Qs to the mail list, not me directly,<br>> so others can comment.<br>> <br>> 1) the header of the file tells you what's in the columns:<br>> <br>> # TimeStep Number-of-layers<br>> # Layer Coordinate Natoms v_temp<br>> 1000 20<br>> 1 0.025 150.9 1.48848<br>> <br>> 4 columns are layer, coord, # of atoms, v_temp<br>> <br>> 2) You should get a V-shaped profile at any point in<br>> time. You are getting one from 1.4 to 1.6 roughly.<br>> You're not swapping enough KE to get a stronger V.<br>> See the MP papers for more discussion.<br>> You typically have to average the profile over many<br>> outputs to get a smooth profile.<br>> <br>> 3) The heat flux is something LAMMPS computes<br>> for you and stores with the fix. See the fix thermal/conductiity<br>> command for how to get at that value and print it out.<br>> <br>> Steve<br>> <br>> On Mon, Jan 19, 2009 at 8:44 PM, Sreekant Narumanchi<br>> <[email protected]...> wrote:<br>> > Steve,<br>> > I incorporated your suggestion in the input file (attached) and ran a<br>> > simulation - the log file and the temperature profile file (tmp.profile)<br>> are<br>> > also attached.<br>> > I have some questions about the tmp.profile.<br>> ><br>> > 1) The first is regarding the various columns that appear in the<br>> > tmp.profile. I was trying to understand what columns 3 and 4 are (is one<br>> of<br>> > them v_temp?).<br>> > 2) If I am understanding the Muller-Plathe algorithm, I should be getting<br>> a<br>> > linear temperature profile in the 20 layers that I have in the<br>> z-direction.<br>> > If I should be looking at column 4 of tmp.profile, it is not clear that<br>> > there is a linear profile at any timestep. This is a little confusing to<br>> me<br>> > at present.<br>> > 3) In order to extract a number for thermal conductivity, I also have to<br>> > compute the heat flux. I am not sure how to extract this heat flux number.<br>> > Would I have to use the thermo_style command to extract the energy?<br>> > Actually, in the input file, ke/atom is already being computed.<br>> > In the near future, I would like to take a shot at computing bulk thermal<br>> > conductivity of silicon (perhaps with stillinger weber potential) at 300K<br>> > and compare to the experimental value (~ 150 W/mK). All this is still<br>> > primarily aimed at understanding MD and LAMMPS better.<br>> ><br>> ><br>> > Thanks,<br>> > Sreekant<br>> ><br>> ><br>> > --- On Mon, 1/19/09, Steve Plimpton <[email protected]> wrote:<br>> ><br>> > From: Steve Plimpton <[email protected]><br>> > Subject: Re: [lammps-users] Thermal conductivity computation<br>> > To: "Sreekant Narumanchi" <[email protected]...><br>> > Cc: [email protected]<br>> > Date: Monday, January 19, 2009, 8:32 AM<br>> ><br>> > You want<br>> > variable temp atom c_ke[]/1.5<br>> > not<br>> > variable temp atom c_ke/1.5<br>> ><br>> > Read the variable doc page. You also have fix nve commented out.<br>> > Your atoms will not move.<br>> ><br>> > Steve<br>> ><br>> > On Sat, Jan 17, 2009 at 10:26 PM, Sreekant Narumanchi<br>> > <[email protected]...> wrote:<br>> >> Steve,<br>> >> I was trying to run a thermal conductivity example by modifying<br>> > bench/in.lj<br>> >> (attached file).<br>> >> I get an error message as shown in the attached file. Please advise as<br>> to<br>> >> what I maybe doing incorrectly.<br>> >> I was also not sure if I need a thermo_style command in the input<br>> file.<br>> >> I am trying to extract a thermal conductivity number from the<br>> simulation.<br>> >> I will also go through the Muller-Plathe paper.<br>> >><br>> >> Thanks,<br>> >> Sreekant<br>> >><br>> >> --- On Mon, 1/12/09, Steve Plimpton <[email protected]> wrote:<br>> >><br>> >> From: Steve Plimpton <[email protected]><br>> >> Subject: Re: [lammps-users] Thermal conductivity computation<br>> >> To: [email protected]...<br>> >> Cc: [email protected]<br>> >> Date: Monday, January 12, 2009, 7:56 AM<br>> >><br>> >> I don't have an example,but aside from the fix itself the<br>> >> only command you need (to monitor the KE flow) are<br>> >> those listed on the fix thermal/cond doc page, with<br>> >> fix ave/spatial. You could add all that to bench/in.lj<br>> >> and it should work.<br>> >><br>> >> Steve<br>> >><br>> >> On Sun, Jan 11, 2009 at 9:08 PM, Sreekant Narumanchi<br>> >> <[email protected]...> wrote:<br>> >>> I am a new LAMMPS user and am working through some of the<br>> examples.<br>> >>> It would be very helpful if someone could send me a simple input<br>> > script<br>> >> file<br>> >>> illustrating the use of the fix thermal/conductivity command to<br>> > compute<br>> >> the<br>> >>> thermal conductivity in a material (any material and any<br>> potential).<br>> >>><br>> >>> Thanks,<br>> >>> Sreekant<br>> >>><br>> >>> --- On Tue, 12/30/08, Sreekant Narumanchi<br>> >> <[email protected]...><br>> >>> wrote:<br>> >>><br>> >>> From: Sreekant Narumanchi <[email protected]...><br>> >>> Subject: Thermal conductivity in solids<br>> >>> To: [email protected]<br>> >>> Date: Tuesday, December 30, 2008, 7:24 AM<br>> >>><br>> >>> I am a new user of LAMMPS and molecular dynamics codes in<br>> general. My<br>> >>> interest is in thermal transport. To get started, I was planning<br>> to<br>> > use<br>> >>> LAMMPS to compute thermal conductivity in a solid (e.g. silicon).<br>> I<br>> > was<br>> >>> wondering if anyone has used LAMMPS for thermal conductivity<br>> > computations<br>> >>> and had suggestions on ways to proceed.<br>> >>><br>> >>> Thanks,<br>> >>> Sreekant Narumanchi<br>> >>><br>> >>><br>> >>><br>> >><br>> ><br>> ------------------------------------------------------------------------------<br>> >>> Check out the new SourceForge.net Marketplace.<br>> >>> It is the best place to buy or sell services for<br>> >>> just about anything Open Source.<br>> >>> http://p.sf.net/sfu/Xq1LFB<br>> >>> _______________________________________________<br>> >>> lammps-users mailing list<br>> >>> [email protected]<br>> >>> https://lists.sourceforge.net/lists/listinfo/lammps-users<br>> >>><br>> >>><br>> >><br>> >><br>> ><br>> ><br>> <br>> ------------------------------------------------------------------------------<br>> This SF.net email is sponsored by:<br>> SourcForge Community<br>> SourceForge wants to tell your story.<br>> http://p.sf.net/sfu/sf-spreadtheword<br>> _______________________________________________<br>> lammps-users mailing list<br>> [email protected]<br>> https://lists.sourceforge.net/lists/listinfo/lammps-users<br>> <br>>

|

processingofdata-01-20-09-swaprate-10steps.xls (58.5 KB)

You'll have to judge what is linear enough. Re: non-cubic cells,
I've never used MP with those, but I can't think of any
immediate problem, so long as you put your swap bins
in a direction that is not tilted. E.g. have an xy tilt and
bin in the y dir.

Steve


Steve,
I have been trying to extract the kinetic energy that is being exchanged between the “hot” and “cold” layer in the thermal conductivity computation. The documentation page for the fix thermal/conductivity command mentions that the cumulative kinetic energy being exchanged between the hot and cold layer is stored as a scalar quantity. But I am not clear on what is the name for this scalar and how I would access it. I looked at fix_thermal_conductivity.cpp file and see a variable “ke” being computed. But I am not sure if “ke” is the variable I should be printing out via a thermo_style command. Also, I assume I should only be looking for the kinetic energy being swapped and not the kinetic energy of the whole system.
In order to understand fix_thermal_conductivity.cpp, are there any other files that I should also look at?


Thanks,
Sreekant


— On Wed, 1/21/09, Steve Plimpton [email protected] wrote:



> From: Steve Plimpton [email protected]
> Subject: Re: [lammps-users] Fwd: Thermal conductivity computation
> To: “Sreekant Narumanchi” <sreekant_narumanchi@…16…>
> Cc: “Lammps” [email protected]
> Date: Wednesday, January 21, 2009, 7:24 AM
>
> <br>> You'll have to judge what is linear enough. Re: non-cubic cells,<br>> I've never used MP with those, but I can't think of any<br>> immediate problem, so long as you put your swap bins<br>> in a direction that is not tilted. E.g. have an xy tilt and<br>> bin in the y dir.<br>> <br>> Steve<br>> <br>> On Tue, Jan 20, 2009 at 8:52 PM, Sreekant Narumanchi<br>> <[email protected]...> wrote:<br>> > Steve,<br>> > I increased the kinetic energy swap rate to 10 timesteps (50 timesteps).<br>> > The V-shape is shown in the attached Excel file for two time instants<br>> (25000<br>> > and 30000 th time steps). Would you agree they are starting to be more<br>> > linear?<br>> > I am aiming to replicate some of the 1997 J.Chem. Phys. Muller-Plathe<br>> paper<br>> > results. In that paper, an orthorhombic periodic cell is used. In order to<br>> > create such a cell, would a "custom" lattice style have to be<br>> used?<br>> ><br>> > Thanks,<br>> > Sreekant<br>> > --- On Tue, 1/20/09, Steve Plimpton <[email protected]> wrote:<br>> ><br>> > From: Steve Plimpton <[email protected]><br>> > Subject: [lammps-users] Fwd: Thermal conductivity computation<br>> > To: "Lammps" <[email protected]><br>> > Date: Tuesday, January 20, 2009, 8:32 AM<br>> ><br>> > ---------- Forwarded message ----------<br>> > From: Steve Plimpton <[email protected]><br>> > Date: Tue, Jan 20, 2009 at 7:57 AM<br>> > Subject: Re: [lammps-users] Thermal conductivity computation<br>> > To: Sreekant Narumanchi <[email protected]...><br>> ><br>> ><br>> > Please post your Qs to the mail list, not me directly,<br>> > so others can comment.<br>> ><br>> > 1) the header of the file tells you what's in the columns:<br>> ><br>> > # TimeStep Number-of-layers<br>> > # Layer Coordinate Natoms v_temp<br>> > 1000 20<br>> > 1 0.025 150.9 1.48848<br>> ><br>> > 4 columns are layer, coord, # of atoms, v_temp<br>> ><br>> > 2) You should get a V-shaped profile at any point in<br>> > time. You are getting one from 1.4 to 1.6 roughly.<br>> > You're not swapping enough KE to get a stronger V.<br>> > See the MP papers for more discussion.<br>> > You typically have to average the profile over many<br>> > outputs to get a smooth profile.<br>> ><br>> > 3) The heat flux is something LAMMPS computes<br>> > for you and stores with the fix. See the fix thermal/conductiity<br>> > command for how to get at that value and print it out.<br>> ><br>> > Steve<br>> ><br>> > On Mon, Jan 19, 2009 at 8:44 PM, Sreekant Narumanchi<br>> > <[email protected]...> wrote:<br>> >> Steve,<br>> >> I incorporated your suggestion in the input file (attached) and ran a<br>> >> simulation - the log file and the temperature profile file<br>> (tmp.profile)<br>> > are<br>> >> also attached.<br>> >> I have some questions about the tmp.profile.<br>> >><br>> >> 1) The first is regarding the various columns that appear in the<br>> >> tmp.profile. I was trying to understand what columns 3 and 4 are (is<br>> one<br>> > of<br>> >> them v_temp?).<br>> >> 2) If I am understanding the Muller-Plathe algorithm, I should be<br>> getting<br>> > a<br>> >> linear temperature profile in the 20 layers that I have in the<br>> > z-direction.<br>> >> If I should be looking at column 4 of tmp.profile, it is not clear<br>> that<br>> >> there is a linear profile at any timestep. This is a little confusing<br>> to<br>> > me<br>> >> at present.<br>> >> 3) In order to extract a number for thermal conductivity, I also have<br>> to<br>> >> compute the heat flux. I am not sure how to extract this heat flux<br>> number.<br>> >> Would I have to use the thermo_style command to extract the energy?<br>> >> Actually, in the input file, ke/atom is already being computed.<br>> >> In the near future, I would like to take a shot at computing bulk<br>> thermal<br>> >> conductivity of silicon (perhaps with stillinger weber potential) at<br>> 300K<br>> >> and compare to the experimental value (~ 150 W/mK). All this is still<br>> >> primarily aimed at understanding MD and LAMMPS better.<br>> >><br>> >><br>> >> Thanks,<br>> >> Sreekant<br>> >><br>> >><br>> >> --- On Mon, 1/19/09, Steve Plimpton <[email protected]> wrote:<br>> >><br>> >> From: Steve Plimpton <[email protected]><br>> >> Subject: Re: [lammps-users] Thermal conductivity computation<br>> >> To: "Sreekant Narumanchi"<br>> <[email protected]...><br>> >> Cc: [email protected]<br>> >> Date: Monday, January 19, 2009, 8:32 AM<br>> >><br>> >> You want<br>> >> variable temp atom c_ke[]/1.5<br>> >> not<br>> >> variable temp atom c_ke/1.5<br>> >><br>> >> Read the variable doc page. You also have fix nve commented out.<br>> >> Your atoms will not move.<br>> >><br>> >> Steve<br>> >><br>> >> On Sat, Jan 17, 2009 at 10:26 PM, Sreekant Narumanchi<br>> >> <[email protected]...> wrote:<br>> >>> Steve,<br>> >>> I was trying to run a thermal conductivity example by modifying<br>> >> bench/in.lj<br>> >>> (attached file).<br>> >>> I get an error message as shown in the attached file. Please<br>> advise as<br>> > to<br>> >>> what I maybe doing incorrectly.<br>> >>> I was also not sure if I need a thermo_style command in the input<br>> > file.<br>> >>> I am trying to extract a thermal conductivity number from the<br>> > simulation.<br>> >>> I will also go through the Muller-Plathe paper.<br>> >>><br>> >>> Thanks,<br>> >>> Sreekant<br>> >>><br>> >>> --- On Mon, 1/12/09, Steve Plimpton <[email protected]><br>> wrote:<br>> >>><br>> >>> From: Steve Plimpton <[email protected]><br>> >>> Subject: Re: [lammps-users] Thermal conductivity computation<br>> >>> To: [email protected]...<br>> >>> Cc: [email protected]<br>> >>> Date: Monday, January 12, 2009, 7:56 AM<br>> >>><br>> >>> I don't have an example,but aside from the fix itself the<br>> >>> only command you need (to monitor the KE flow) are<br>> >>> those listed on the fix thermal/cond doc page, with<br>> >>> fix ave/spatial. You could add all that to bench/in.lj<br>> >>> and it should work.<br>> >>><br>> >>> Steve<br>> >>><br>> >>> On Sun, Jan 11, 2009 at 9:08 PM, Sreekant Narumanchi<br>> >>> <[email protected]...> wrote:<br>> >>>> I am a new LAMMPS user and am working through some of the<br>> > examples.<br>> >>>> It would be very helpful if someone could send me a simple<br>> input<br>> >> script<br>> >>> file<br>> >>>> illustrating the use of the fix thermal/conductivity command<br>> to<br>> >> compute<br>> >>> the<br>> >>>> thermal conductivity in a material (any material and any<br>> > potential).<br>> >>>><br>> >>>> Thanks,<br>> >>>> Sreekant<br>> >>>><br>> >>>> --- On Tue, 12/30/08, Sreekant Narumanchi<br>> >>> <[email protected]...><br>> >>>> wrote:<br>> >>>><br>> >>>> From: Sreekant Narumanchi<br>> <[email protected]...><br>> >>>> Subject: Thermal conductivity in solids<br>> >>>> To: [email protected]<br>> >>>> Date: Tuesday, December 30, 2008, 7:24 AM<br>> >>>><br>> >>>> I am a new user of LAMMPS and molecular dynamics codes in<br>> > general. My<br>> >>>> interest is in thermal transport. To get started, I was<br>> planning<br>> > to<br>> >> use<br>> >>>> LAMMPS to compute thermal conductivity in a solid (e.g.<br>> silicon).<br>> > I<br>> >> was<br>> >>>> wondering if anyone has used LAMMPS for thermal conductivity<br>> >> computations<br>> >>>> and had suggestions on ways to proceed.<br>> >>>><br>> >>>> Thanks,<br>> >>>> Sreekant Narumanchi<br>> >>>><br>> >>>><br>> >>>><br>> >>><br>> >><br>> ><br>> ------------------------------------------------------------------------------<br>> >>>> Check out the new SourceForge.net Marketplace.<br>> >>>> It is the best place to buy or sell services for<br>> >>>> just about anything Open Source.<br>> >>>> http://p.sf.net/sfu/Xq1LFB<br>> >>>> _______________________________________________<br>> >>>> lammps-users mailing list<br>> >>>> [email protected]<br>> >>>> https://lists.sourceforge.net/lists/listinfo/lammps-users<br>> >>>><br>> >>>><br>> >>><br>> >>><br>> >><br>> >><br>> ><br>> ><br>> ------------------------------------------------------------------------------<br>> > This SF.net email is sponsored by:<br>> > SourcForge Community<br>> > SourceForge wants to tell your story.<br>> > http://p.sf.net/sfu/sf-spreadtheword<br>> > _______________________________________________<br>> > lammps-users mailing list<br>> > [email protected]<br>> > https://lists.sourceforge.net/lists/listinfo/lammps-users<br>> ><br>> ><br>> <br>>

|

Many fixes store scalar or vector quantities that can be accessed
by other commands in LAMMPS. See section 4.15 of the manual
for an overview. In this case, you probably want to use
the thermo_style custom command to print the output - see
the doc page for that command for more details on accessing a fix.

Steve


Steve,
I am able to extract the kinetic energy exchange between the “hot” and “cold” layer (via f_2 in the script below). I think the thermal conductivity I am extracting for the Lennard-Jones system (with a simple cubic structure) is close to what is reported in the Muller-Plathe paper.
However, I have been struggling to replicate the domain used in that paper. In that paper, they simulate an orthorhombic periodic cell of size 10.0587X 10.0587X30.1762 cubic sigma (in Lennard-Jones reduced units), with a number density of 0.849. The number of atoms is 2592.
I have tried a few things with the lattice “custom” command given in the script below - but I am not able to replicate the domain mentioned above and the number of atoms.
Can you please advise how I can get the above-mentioned domain?

Thanks,
Sreekant


*


# 3d Lennard-Jones







boundary p p p



variable x index 1



variable y index 1



variable z index 1







variable xx equal 10.0587
$x



variable yy equal 10.0587
$y



variable zz equal 10.0587
z<br><br><br><br><br><br><br><br>units lj<br><br><br><br>atom_style atomic<br><br><br><br><br><br><br><br>lattice custom 1 a1 1 0 0 a2 0 1 0 a3 0 0 3 basis 0.0 0.0 0.0<br><br><br><br>region box block 0 {xx} 0 {yy} 0 {zz}



create_box 1 box



create_atoms 1 box



mass 1 1.0







velocity all create 0.694 8700 loop geom







pair_style lj/cut 3.0



pair_coeff 1 1 1.0 1.0 3.0







neighbor 0.3 bin



neigh_modify delay 0 every 20 check no



fix 1 all nve



fix 2 all thermal/conductivity 30 z 20 swap 2



dump 1 all atom 50 dump.lj



compute ke all ke/atom



variable temp atom c_ke[]/1.5



fix 3 all ave/spatial 10 10 1000 z lower 0.05 v_temp file tmp.profile units reduced



thermo_style custom step f_2



timestep 0.002



run 5000






— On Mon, 1/26/09, Steve Plimpton [email protected] wrote:



> From: Steve Plimpton [email protected]
> Subject: Re: [lammps-users] Fwd: Thermal conductivity computation
> To: sreekant_narumanchi@…16…
> Cc: “Lammps” [email protected]
> Date: Monday, January 26, 2009, 8:26 AM
>
> <br>> Many fixes store scalar or vector quantities that can be accessed<br>> by other commands in LAMMPS. See section 4.15 of the manual<br>> for an overview. In this case, you probably want to use<br>> the thermo_style custom command to print the output - see<br>> the doc page for that command for more details on accessing a fix.<br>> <br>> Steve<br>> <br>> On Sat, Jan 24, 2009 at 4:16 PM, Sreekant Narumanchi<br>> <[email protected]...> wrote:<br>> > Steve,<br>> > I have been trying to extract the kinetic energy that is being exchanged<br>> > between the "hot" and "cold" layer in the thermal<br>> conductivity computation.<br>> > The documentation page for the fix thermal/conductivity command mentions<br>> > that the cumulative kinetic energy being exchanged between the hot and<br>> cold<br>> > layer is stored as a scalar quantity. But I am not clear on what is the<br>> name<br>> > for this scalar and how I would access it. I looked at<br>> > fix_thermal_conductivity.cpp file and see a variable "ke" being<br>> computed.<br>> > But I am not sure if "ke" is the variable I should be printing<br>> out via a<br>> > thermo_style command. Also, I assume I should only be looking for the<br>> > kinetic energy being swapped and not the kinetic energy of the whole<br>> system.<br>> > In order to understand fix_thermal_conductivity.cpp, are there any other<br>> > files that I should also look at?<br>> ><br>> ><br>> > Thanks,<br>> > Sreekant<br>> ><br>> > --- On Wed, 1/21/09, Steve Plimpton <[email protected]> wrote:<br>> ><br>> > From: Steve Plimpton <[email protected]><br>> > Subject: Re: [lammps-users] Fwd: Thermal conductivity computation<br>> > To: "Sreekant Narumanchi" <[email protected]...><br>> > Cc: "Lammps" <[email protected]><br>> > Date: Wednesday, January 21, 2009, 7:24 AM<br>> ><br>> > You'll have to judge what is linear enough. Re: non-cubic cells,<br>> > I've never used MP with those, but I can't think of any<br>> > immediate problem, so long as you put your swap bins<br>> > in a direction that is not tilted. E.g. have an xy tilt and<br>> > bin in the y dir.<br>> ><br>> > Steve<br>> ><br>> > On Tue, Jan 20, 2009 at 8:52 PM, Sreekant Narumanchi<br>> > <[email protected]...> wrote:<br>> >> Steve,<br>> >> I increased the kinetic energy swap rate to 10 timesteps (50<br>> timesteps).<br>> >> The V-shape is shown in the attached Excel file for two time instants<br>> > (25000<br>> >> and 30000 th time steps). Would you agree they are starting to be more<br>> >> linear?<br>> >> I am aiming to replicate some of the 1997 J.Chem. Phys. Muller-Plathe<br>> > paper<br>> >> results. In that paper, an orthorhombic periodic cell is used. In<br>> order to<br>> >> create such a cell, would a "custom" lattice style have to<br>> be<br>> > used?<br>> >><br>> >> Thanks,<br>> >> Sreekant<br>> >> --- On Tue, 1/20/09, Steve Plimpton <[email protected]> wrote:<br>> >><br>> >> From: Steve Plimpton <[email protected]><br>> >> Subject: [lammps-users] Fwd: Thermal conductivity computation<br>> >> To: "Lammps" <[email protected]><br>> >> Date: Tuesday, January 20, 2009, 8:32 AM<br>> >><br>> >> ---------- Forwarded message ----------<br>> >> From: Steve Plimpton <[email protected]><br>> >> Date: Tue, Jan 20, 2009 at 7:57 AM<br>> >> Subject: Re: [lammps-users] Thermal conductivity computation<br>> >> To: Sreekant Narumanchi <[email protected]...><br>> >><br>> >><br>> >> Please post your Qs to the mail list, not me directly,<br>> >> so others can comment.<br>> >><br>> >> 1) the header of the file tells you what's in the columns:<br>> >><br>> >> # TimeStep Number-of-layers<br>> >> # Layer Coordinate Natoms v_temp<br>> >> 1000 20<br>> >> 1 0.025 150.9 1.48848<br>> >><br>> >> 4 columns are layer, coord, # of atoms, v_temp<br>> >><br>> >> 2) You should get a V-shaped profile at any point in<br>> >> time. You are getting one from 1.4 to 1.6 roughly.<br>> >> You're not swapping enough KE to get a stronger V.<br>> >> See the MP papers for more discussion.<br>> >> You typically have to average the profile over many<br>> >> outputs to get a smooth profile.<br>> >><br>> >> 3) The heat flux is something LAMMPS computes<br>> >> for you and stores with the fix. See the fix thermal/conductiity<br>> >> command for how to get at that value and print it out.<br>> >><br>> >> Steve<br>> >><br>> >> On Mon, Jan 19, 2009 at 8:44 PM, Sreekant Narumanchi<br>> >> <[email protected]...> wrote:<br>> >>> Steve,<br>> >>> I incorporated your suggestion in the input file (attached) and<br>> ran a<br>> >>> simulation - the log file and the temperature profile file<br>> > (tmp.profile)<br>> >> are<br>> >>> also attached.<br>> >>> I have some questions about the tmp.profile.<br>> >>><br>> >>> 1) The first is regarding the various columns that appear in the<br>> >>> tmp.profile. I was trying to understand what columns 3 and 4 are<br>> (is<br>> > one<br>> >> of<br>> >>> them v_temp?).<br>> >>> 2) If I am understanding the Muller-Plathe algorithm, I should be<br>> > getting<br>> >> a<br>> >>> linear temperature profile in the 20 layers that I have in the<br>> >> z-direction.<br>> >>> If I should be looking at column 4 of tmp.profile, it is not clear<br>> > that<br>> >>> there is a linear profile at any timestep. This is a little<br>> confusing<br>> > to<br>> >> me<br>> >>> at present.<br>> >>> 3) In order to extract a number for thermal conductivity, I also<br>> have<br>> > to<br>> >>> compute the heat flux. I am not sure how to extract this heat flux<br>> > number.<br>> >>> Would I have to use the thermo_style command to extract the<br>> energy?<br>> >>> Actually, in the input file, ke/atom is already being computed.<br>> >>> In the near future, I would like to take a shot at computing bulk<br>> > thermal<br>> >>> conductivity of silicon (perhaps with stillinger weber potential)<br>> at<br>> > 300K<br>> >>> and compare to the experimental value (~ 150 W/mK). All this is<br>> still<br>> >>> primarily aimed at understanding MD and LAMMPS better.<br>> >>><br>> >>><br>> >>> Thanks,<br>> >>> Sreekant<br>> >>><br>> >>><br>> >>> --- On Mon, 1/19/09, Steve Plimpton <[email protected]><br>> wrote:<br>> >>><br>> >>> From: Steve Plimpton <[email protected]><br>> >>> Subject: Re: [lammps-users] Thermal conductivity computation<br>> >>> To: "Sreekant Narumanchi"<br>> > <[email protected]...><br>> >>> Cc: [email protected]<br>> >>> Date: Monday, January 19, 2009, 8:32 AM<br>> >>><br>> >>> You want<br>> >>> variable temp atom c_ke[]/1.5<br>> >>> not<br>> >>> variable temp atom c_ke/1.5<br>> >>><br>> >>> Read the variable doc page. You also have fix nve commented out.<br>> >>> Your atoms will not move.<br>> >>><br>> >>> Steve<br>> >>><br>> >>> On Sat, Jan 17, 2009 at 10:26 PM, Sreekant Narumanchi<br>> >>> <[email protected]...> wrote:<br>> >>>> Steve,<br>> >>>> I was trying to run a thermal conductivity example by<br>> modifying<br>> >>> bench/in.lj<br>> >>>> (attached file).<br>> >>>> I get an error message as shown in the attached file. Please<br>> > advise as<br>> >> to<br>> >>>> what I maybe doing incorrectly.<br>> >>>> I was also not sure if I need a thermo_style command in the<br>> input<br>> >> file.<br>> >>>> I am trying to extract a thermal conductivity number from the<br>> >> simulation.<br>> >>>> I will also go through the Muller-Plathe paper.<br>> >>>><br>> >>>> Thanks,<br>> >>>> Sreekant<br>> >>>><br>> >>>> --- On Mon, 1/12/09, Steve Plimpton <[email protected]><br>> > wrote:<br>> >>>><br>> >>>> From: Steve Plimpton <[email protected]><br>> >>>> Subject: Re: [lammps-users] Thermal conductivity computation<br>> >>>> To: [email protected]...<br>> >>>> Cc: [email protected]<br>> >>>> Date: Monday, January 12, 2009, 7:56 AM<br>> >>>><br>> >>>> I don't have an example,but aside from the fix itself the<br>> >>>> only command you need (to monitor the KE flow) are<br>> >>>> those listed on the fix thermal/cond doc page, with<br>> >>>> fix ave/spatial. You could add all that to bench/in.lj<br>> >>>> and it should work.<br>> >>>><br>> >>>> Steve<br>> >>>><br>> >>>> On Sun, Jan 11, 2009 at 9:08 PM, Sreekant Narumanchi<br>> >>>> <[email protected]...> wrote:<br>> >>>>> I am a new LAMMPS user and am working through some of the<br>> >> examples.<br>> >>>>> It would be very helpful if someone could send me a simple<br>> > input<br>> >>> script<br>> >>>> file<br>> >>>>> illustrating the use of the fix thermal/conductivity<br>> command<br>> > to<br>> >>> compute<br>> >>>> the<br>> >>>>> thermal conductivity in a material (any material and any<br>> >> potential).<br>> >>>>><br>> >>>>> Thanks,<br>> >>>>> Sreekant<br>> >>>>><br>> >>>>> --- On Tue, 12/30/08, Sreekant Narumanchi<br>> >>>> <[email protected]...><br>> >>>>> wrote:<br>> >>>>><br>> >>>>> From: Sreekant Narumanchi<br>> > <[email protected]...><br>> >>>>> Subject: Thermal conductivity in solids<br>> >>>>> To: [email protected]<br>> >>>>> Date: Tuesday, December 30, 2008, 7:24 AM<br>> >>>>><br>> >>>>> I am a new user of LAMMPS and molecular dynamics codes in<br>> >> general. My<br>> >>>>> interest is in thermal transport. To get started, I was<br>> > planning<br>> >> to<br>> >>> use<br>> >>>>> LAMMPS to compute thermal conductivity in a solid (e.g.<br>> > silicon).<br>> >> I<br>> >>> was<br>> >>>>> wondering if anyone has used LAMMPS for thermal<br>> conductivity<br>> >>> computations<br>> >>>>> and had suggestions on ways to proceed.<br>> >>>>><br>> >>>>> Thanks,<br>> >>>>> Sreekant Narumanchi<br>> >>>>><br>> >>>>><br>> >>>>><br>> >>>><br>> >>><br>> >><br>> ><br>> ------------------------------------------------------------------------------<br>> >>>>> Check out the new SourceForge.net Marketplace.<br>> >>>>> It is the best place to buy or sell services for<br>> >>>>> just about anything Open Source.<br>> >>>>> http://p.sf.net/sfu/Xq1LFB<br>> >>>>> _______________________________________________<br>> >>>>> lammps-users mailing list<br>> >>>>> [email protected]<br>> >>>>> https://lists.sourceforge.net/lists/listinfo/lammps-users<br>> >>>>><br>> >>>>><br>> >>>><br>> >>>><br>> >>><br>> >>><br>> >><br>> >><br>> ><br>> ------------------------------------------------------------------------------<br>> >> This SF.net email is sponsored by:<br>> >> SourcForge Community<br>> >> SourceForge wants to tell your story.<br>> >> http://p.sf.net/sfu/sf-spreadtheword<br>> >> _______________________________________________<br>> >> lammps-users mailing list<br>> >> [email protected]<br>> >> https://lists.sourceforge.net/lists/listinfo/lammps-users<br>> >><br>> >><br>> ><br>> ><br>> <br>>

|

Just use the lattice command with a reduced density of 0.849 and
some number of unit cells like 10 10 30. The 10.0587 is a
result from the reduced density being converted to box units.

Steve