Encountering Crashes during Stress Calculations with Hybrid Functionals

[jiweiz]

Dear Francois,

I am in the process of conducting structure optimizations utilizing various xc functionals, where stress calculations are requisite. However, upon acquainting myself with the test example `test/si4stress` and attempting various hybrid functionals such as PBE0 or HSE, I encounter a crash during the calculation. I am uncertain if there's an error on my part. Attached are the input files for my test:

gs.i:

Code: Select all

# Si4 ground state
set cell 16 0 0  0 16 0  0 0 16
set ref_cell 18 0 0  0 18 0  0 0 18
species silicon /project/gagalli/jiaweiz/ONCVPSP/xml/Si_ONCV_PBE-1.2.xml
atom Si1 silicon  3.700 -0.100  0.300
atom Si2 silicon -0.100  2.800 -0.200
atom Si3 silicon -3.700  0.100  0.300
atom Si4 silicon -0.100 -2.800 -0.200
strain 0.02 0.04 0.06 0.03 0.05 0.07
params.i
set wf_dyn PSDA
set stress ON
set debug STRESS ON
randomize_wf
set scf_tol 1.e-9
run 0 300
save gs.xml

params.i:

Code: Select all

set ecut 12
set ecuts 10
set xc HSE

output:

Code: Select all

...
[qbox] <cmd>run 0 300</cmd>
 EnergyFunctional: <np0v> 44 </np0v>  <np1v> 44 </np1v>  <np2v> 44 </np2v>
 EnergyFunctional: vft->np012(): 85184
<wavefunction ecut="6.00000000" nspin="1" nel="16" nempty="0">
<cell a="16.320000 0.480000 1.120000"
      b="0.480000 16.640000 0.800000"
      c="1.120000 0.800000 16.960000"/>
 reciprocal lattice vectors
 0.387014 -0.009958 -0.025088
 -0.009958 0.378710 -0.017206
 -0.025088 -0.017206 0.372939
<refcell a="18.000000 0.000000 0.000000"
         b="0.000000 18.000000 0.000000"
         c="0.00
free(): corrupted unsorted chunks
free(): corrupted unsorted chunks

I appreciate your help with this. Looking forward to hearing from you soon.

Best,
Jiawei

[fgygi]

It seems that this particular combination of parameters escaped testing. This is fixed in release 1.76.1.

[jiweiz]

Thank you very much Francois.