SCF Fails to start for MoO3

Having trouble to get an SCF (and eventual GeoOpt) to start for MoO3.
the calculation seems to stall at the begining of the SCF, even with 28 cores and exciting_purempi.

I pulled a cif file of MoO3 off of materials project

and wrote an input.xml file with ase’s exciting writer.

I found that the calculation does not start, and instead hands on the begining of the SCF calculation.

Naively, it seems that muffin tin radii sum for O+Mo is greater than the distance between atoms, so i set the Mo radius to be 1.9 and the O radius to be 1.2 using a kpt sampling of [8,8,8]

Is there something i am missing?

Here is the input.xml

<?xml version="1.0" ?>
<input xmlns:xsi="" xsi:noNamespaceSchemaLocation="">
	<title>MoO3 Geo Opt</title>
	<structure speciespath="exciting/exciting/species" autormt="false" tshift="true">
			<basevect>6.95667380701739 0.00000000000000 0.00000000000000</basevect>
			<basevect>0.00000000000000 7.40534651109514 0.00000000000000</basevect>
			<basevect>0.00000000000000 0.00000000000000 26.88905760858228</basevect>
		<species speciesfile="Mo.xml" chemicalSymbol="Mo" rmt = "1.9">
			<atom coord="0.25000000000000 0.42300961000000 0.60205717000000"/>
			<atom coord="0.75000000000000 0.57699039000000 0.39794283000000"/>
			<atom coord="0.75000000000000 0.92300961000000 0.89794283000000"/>
			<atom coord="0.25000000000000 0.07699039000000 0.10205717000000"/>
		<species speciesfile="O.xml" chemicalSymbol="O" rmt = "1.2">
			<atom coord="0.25000000000000 0.46632673000000 0.71834359000000"/>
			<atom coord="0.75000000000000 0.53367327000000 0.28165641000000"/>
			<atom coord="0.75000000000000 0.96632673000000 0.78165641000000"/>
			<atom coord="0.25000000000000 0.03367327000000 0.21834359000000"/>
			<atom coord="0.25000000000000 0.49853619000000 0.43558467000000"/>
			<atom coord="0.75000000000000 0.50146381000000 0.56441533000000"/>
			<atom coord="0.75000000000000 0.99853619000000 0.06441533000000"/>
			<atom coord="0.25000000000000 0.00146381000000 0.93558467000000"/>
			<atom coord="0.25000000000000 0.98268220000000 0.58709074000000"/>
			<atom coord="0.75000000000000 0.48268220000000 0.91290926000000"/>
			<atom coord="0.75000000000000 0.01731780000000 0.41290926000000"/>
			<atom coord="0.25000000000000 0.51731780000000 0.08709074000000"/>
	<groundstate outputlevel="normal"
                 ngridk="8 8 4" 
                 gmaxvr="15.0" />

Hi strategos,

Some general points:

  • Don’t use exciting_purempi, you won’t have any threading support (it’s for development purposes only). Use smp_and_mpi.

  • W.r.t the input file, the lattice vectors look like Bohr and atomic positions look fractional, which is correct

  • If the muffin tin radii are nearly touching, you’ll still have problems with the basis. Try a RMT_O + RMT_Mo = 0.8 * bond length and go from there.

  • rgkmax should really be specified, this will affect the basis quality (but can be low for testing)

Can you give the INFO.OUT and any errors warnings?



Ok i will recompile for exciting_mpiandsmp.

I did try a smaller rmt for Mo and O and found that yes, rgkmax was the problem.

I think that solved the error.