problem with calculating equilibirum about pycalphad HOT 3 CLOSED

Hi Jing,

Thank you for your message. You should definitely trust the result from Thermo-Calc and OpenCalphad. pycalphad's energy minimization routine currently has trouble with tiny site fractions, so sometimes it cannot achieve good convergence when they are smaller than 1e-2 or so. Sometimes it works, sometimes it doesn't. I have attempted some solutions for this but they make the calculation very slow.

Unfortunately I do not have time to fix this yet, but I am hoping to solve the problem by the end of the summer.

from pycalphad.

If you want you can also try manually adjusting the point density for the calculation, like

# default 'pdens' is 100
equilibrium(db, ['AL', 'NI', 'VA'], my_phases, {v.X('AL'): 0.2, v.T: 873.15, v.P: 101325}, calc_opts={'pdens': 200})

from pycalphad.

I raise the point density to 1000 and it finally reach a result the same as the opencalphad and pycalphad.

Here are the result,
X = array([[[[[ 0.23357854, 0.76642146],
[ 0.10570285, 0.89429715]]]]])
Y = array([[[[[ 1.58639953e-04, 9.99841360e-01, 9.33838246e-01,
6.61617537e-02],
[ 1.05702851e-01, 8.94297149e-01, 1.00000000e+00,
nan]]]]])

I used to compute the equilibrium by calling the opencalphad using subprocess. It seems that when proper point density is chosen I have no need to do this any more. I would like to try something else.

Thanks!

from pycalphad.