Dear QE developers,<br><br>I would like to point out a few problems in the examples for PW in espresso-5.0.1.<div><br></div><div>example05 has a run_example script which is inconsistent with the README and reference directory, namely the script has calculations on Ni whereas the README and reference do not. </div>
<div><br></div><div>Moreover, running with ifort 12.1.5, default configuration returned by configure, using MKL, one of the Ni calculations does not converge. results/Ni_gamma_d9s1.out ends with:<br>
<br><div> iteration #100 ecut= 27.00 Ry beta=0.25</div><div> Davidson diagonalization with overlap</div><div> ethr = 3.63E-10, avg # of iterations = 2.5</div><div><br></div><div> negative rho (up, down): 0.000E+00 0.169E-04</div>
<div><br></div><div> total cpu time spent up to now is 236.0 secs</div><div><br></div><div> total energy = -85.54366480 Ry</div><div> Harris-Foulkes estimate = -85.54366495 Ry</div>
<div> estimated scf accuracy < 0.00000015 Ry</div><div><br></div><div> total magnetization = 2.00 Bohr mag/cell</div><div> absolute magnetization = 2.00 Bohr mag/cell</div><div><br>
</div><div> End of self-consistent calculation</div><div><br></div><div> convergence NOT achieved after 100 iterations: stopping</div><div><br></div><div>By contrast, with openmpi 1.4, on 2 procs, it does converge.</div>
I have nothing to compare the run with otherwise, since it does not appear in the reference directory.</div><div><br><div><br></div><div>cluster_example is unable to download the pseudo (the pseudo's were successfully downloaded for me for all the other examples):</div>
<div><br></div><div><div>Downloading H.pbe-kjpaw.UPF to /home/dstrubbe/espresso-5.0.1/pseudo...</div><div>ERROR: /home/dstrubbe/espresso-5.0.1/pseudo/H.pbe-kjpaw.UPF not existent or not readable</div><div>Aborting</div></div>
<div><br></div><div><br></div><div>Sincerely,</div><div>David Strubbe</div><div>MIT</div>
</div>