Emc2 gets an negative entropy and NaN phi

Alloy Theoretic Automated Toolkit (ATAT)
1

Hello, I met some problems when I used emc2 to do an MC simulation (NVT).

  1. If the initial T0 is 0, phi will be NaN
    emc2 -gs=-1 -T0=0 -T1=2000 -dT=20 -k=8.617e-5 -dx=1e-4 -aq=0 -er=25 -sigdig=12 -innerT -x=-0.5 -cm -o=out1
    the outcome likes this:

T mu E-mu*x x phi
0.000000 0 -0.020950 0.000000 0.000000 0.000000
20.000000 0 -0.021461 0.000000 -nan 0.000000
40.000000 0 -0.021895 0.000000 -nan 0.000000
60.000000 0 -0.022043 0.000000 -nan 0.000000
80.000000 0 -0.022026 0.000000 -nan 0.000000
…

  1. If the initial T0 is not 0, then phi seems normal, but If I plot phi vs. T, the slope is positive, so entropy is negative. (because it’s an NVT MC, chemical potential mu = 0, in fact phi = F)

emc2 -gs=-1 -T0=10 -T1=2000 -dT=20 -k=8.617e-5 -dx=1e-5 -aq=0 -er=25 -sigdig=12 -innerT -x=-0.8 -cm -o=out0.1
[image deleted]

2

In canonical mode (-cm), features are more limited.
In particular, the code does not do thermodynamic integrations from the disordered state (-gs=-1). It can do it from an ordered phase a low T.
This is source of your issues.
I’ve updated the help files to make this clearer.