[EXTERNAL] Question about shock propagation

Please direct all correspondence to the LAMMPS mailing list.

Q1: The timestep must be sufficiently tiny so that relative changes in force are small from one timestep to the next. How small is small? Small enough so that your results to not depend on the timestep. You might want to start out looking at weak shocks. 25 km/s is a little extreme.
Q2: thermo_temp will give a temperature for the entire system based on the velocity of each atom. You can subtract out the center-of-mass contribution using compute temp/com. But this will not be accurate if different parts of the system are moving a different speeds, as is usually the case in a shock simulation. For accurate results, you can slice your system up in to layers, as the follow snippet demonstrates (courtesy of Ray Shan):

region reg1 block {xbox1} {xbox2} INF INF INF INF units box side in
region reg2 block {xbox2} {xbox3} INF INF INF INF units box side in
region reg3 block {xbox3} {xbox4} INF INF INF INF units box side in
region reg4 block {xbox4} {xbox5} INF INF INF INF units box side in
region reg5 block {xbox5l} {xbox6} INF INF INF INF units box side in

group reg1 region reg1
group reg2 region reg2
group reg3 region reg3
group reg4 region reg4
group reg5 region reg5

compute temp1 reg1 temp/com
compute temp2 reg2 temp/com
compute temp3 reg3 temp/com
compute temp4 reg4 temp/com
compute temp5 reg5 temp/com