Emc2...huge values for Phi

Alloy Theoretic Automated Toolkit (ATAT)
1

Dear Prof. Axel,
When I run MC simulation(emc2) for a 8 atoms unit cell of Nb-Sn system, I get this enormous value for Phi(mc.out column 5). And the code stops running after few iterations…does not scan through the given temperature and the chemical potential range.

I am running this on the latest atat 3.31 beta version. Similar results in older 3.23 version as well.

the emc2.in that I used is as follows:
emc2 -T0=323 -T1=973 -dT=25 -mu0=-1.5 -mu1=0.5 -dmu=0.05 -er=50 -gs=1 -k=8.617e-5 -o=mc.out -dx=0.005
Note: in this particular instance I gave temperature in Kelvin but I have tried with celsius as well and I get similar value for Phi.

Output (1st line in mc.out):
323.000000 -0.534857 -0.582545 -0.750032 340282346638528859811704183484516925440.000000 0.000000 0.000004 3402823466385288598117041834845169254
40.000000 -0.875000 340282346638528859811704183484516925440.000000 -0.582545 -0.750122 -0.587368 -0.582050 -0.998535 -0.582217 0.875
066 1.000000 -1.000000 -0.000127 1.000000 0.000127 1.000000 -0.000127 -1.000000 -0.000127 -1.000000

2

This huge value (defined as maxfloat in C) is a marker to say that the low-temperature expansion (used as the starting value for phi by default) is used outside of its range of validity (usually, for a chemical potential that stabilizes a different ground state than the one given by -gs=…).
Solutions:

  1. either try a different ground state as the initial config
  2. specify your own initial value of phi (-phi0=…) . Note that if the initial value is arbitrary (say 0 or 1), the phi will be correct up to a rigid shift that you may later correct.