One simple way to do this is to have lammps output xyz and ke to a dump file then use a script to convert it to a pdb file. In the script you want normalize kinetic energy to a value between 0 and 1 and output it to the charge column in the pdb file. Then in vmd you can load your molecule and select color by charge. The only problem is that vmd only reads the charge from the first frame of a series of coordinate snap shots. There are some tcl scripts available which can read in and update the charges for each frame. In principle you can dump any variable you might choose (stress, centrosymmetry, etc) and color it in this fashion using the same or a slightly modified script.
Another tip: Make sure you use time averaged quantities, expecially for kinetic energy, because the data can be quite noisy otherwise. When you look at it in vmd it will look all speckled and hard to decipher.
Dave Schall