<div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div>Dear qE community</div><div><br></div><div>I am encountering a problem with neb.x.</div><div><br></div><div>The code executes fine for the 1st iteration only compile in 6.4 and above. I am currently focussing on the 6.4.1 distribution.</div><div><br></div><div>neb.x calculates the first iteration with no problems but then fails to restart for the second, i.e. the 'out' file looks like:</div><div><br></div><div><br></div><div><br></div><div><br></div><div>    Program NEB v.6.4.1 starts on 13Apr2019 at 22:49:32 </div><div>     This program is part of the open-source Quantum ESPRESSO suite<br>     for quantum simulation of materials; please cite<br>         "P. Giannozzi et al., J. Phys.:Condens. Matter 21 395502 (2009);<br>         "P. Giannozzi et al., J. Phys.:Condens. Matter 29 465901 (2017);<br>          URL <a href="http://www.quantum-espresso.org">http://www.quantum-espresso.org</a>", <br>     in publications or presentations arising from this work. More details at<br>     <a href="http://www.quantum-espresso.org/quote">http://www.quantum-espresso.org/quote</a></div><div>     Parallel version (MPI), running on   360 processors</div><div>     MPI processes distributed on    15 nodes<br>     R & G space division:  proc/nbgrp/npool/nimage =     360</div><div>     parsing_file_name: <a href="http://test.in">test.in</a><br>     Reading input from <a href="http://pw_1.in">pw_1.in</a><br>               file C.pz-n-kjpaw_psl.1.0.0.UPF: wavefunction(s)  2S 2P renormalized<br>               file H.pz-kjpaw_psl.1.0.0.UPF: wavefunction(s)  1S renormalized<br>               file Pt.pz-n-kjpaw_psl.1.0.0.UPF: wavefunction(s)  6S 5D renormalized<br>     Reading input from <a href="http://pw_2.in">pw_2.in</a><br>               file C.pz-n-kjpaw_psl.1.0.0.UPF: wavefunction(s)  2S 2P renormalized<br>               file H.pz-kjpaw_psl.1.0.0.UPF: wavefunction(s)  1S renormalized<br>               file Pt.pz-n-kjpaw_psl.1.0.0.UPF: wavefunction(s)  6S 5D renormalized</div><div>     initial path length           =  7.0802 bohr<br>     initial inter-image distance  =  1.1800 bohr<br> <br>     string_method                 =    neb<br>     restart_mode                  =    from_scratch<br>     opt_scheme                    =    broyden<br>     num_of_images                 =    7<br>     nstep_path                    =    10<br>     CI_scheme                     =    no-CI<br>     first_last_opt                =    F<br>     use_freezing                  =    F<br>     ds                            =    2.0000 a.u.<br>     k_max                         =    0.3000 a.u.<br>     k_min                         =    0.2000 a.u.<br>     suggested k_max               =    0.1542 a.u.<br>     suggested k_min               =    0.1028 a.u.<br>     path_thr                      =    0.1000 eV / A</div><div>     ------------------------------ iteration   1 ------------------------------</div><div>     tcpu =      6.1    self-consistency for image   1<br>     tcpu =   1455.9    self-consistency for image   2<br>     tcpu =   2760.1    self-consistency for image   3<br>     tcpu =   3665.8    self-consistency for image   4<br>     tcpu =   4881.4    self-consistency for image   5<br>     tcpu =   6379.5    self-consistency for image   6<br>     tcpu =   7531.2    self-consistency for image   7</div><div>     activation energy (->) =   0.533763 eV<br>     activation energy (<-) =   0.874959 eV</div><div>     image        energy (eV)        error (eV/A)        frozen</div><div>         1    -990477.5110581            0.022280            T<br>         2    -990477.3291278            0.688721            F<br>         3    -990477.0457248            0.978578            F<br>         4    -990476.9772952            1.201974            F<br>         5    -990477.2260733            1.146443            F<br>         6    -990477.6294736            0.797173            F<br>         7    -990477.8522547            0.023965            T</div><div>     path length          =  7.080 bohr<br>     inter-image distance =  1.180 bohr</div><div>     ------------------------------ iteration   2 ------------------------------</div><div>     tcpu =   8726.4    self-consistency for image   2</div><div> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br>     Error in routine pw_readschemafile (1):<br>     xml data file not found<br> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%</div><div>     stopping …</div><div><br></div><div><br></div><div><br></div><div><br></div><div><br></div><div>I've checked the basics:</div><div>(1) the input file to neb.x has: restart_mode = 'from_scratch'</div><div>(2) the input file to neb.x has:   nosym = .true.,<br><span><span>(in case the k-point distribution has changed though this is a low symmetry system, so should be OK for NEB)<span><span><br><span><span>(3) the first and last images were optimised using pw.x from the same distribution so I am reasonably sure it's not a problem with the geometry or the pseudopotentials, for example.</span></span></span></span></span></span></div><span><span><span><span><span><span></span></span></span></span></span></span></div><span><span><span><span><span><span></span></span></span></span></span></span></div><span><span><span><span><span><span></span></span></span></span></span></span></div></div><div><br></div><div>Attempting the simulation with an earlier distribution but with a similar compile strategy produces similar problems at the start of the second iteration. I'm inclined to think therefore it's an I/O problem. I've tried compiling with hdf5 but that doesn't seem to have fixed the problem. I'm compiling using the Archer HPC based in the UK [ <a href="http://www.archer.ac.uk/about-archer/">http://www.archer.ac.uk/about-archer/</a> ] which is a Cray XC30. My compile script (without hdf5) is:</div><div><br></div><div><br></div><div><br></div><span><p>module swap PrgEnv-cray PrgEnv-intel<br>module load fftw</p><p>export CC=cc<br>export FC=ftn<br>export F77=ftn<br>export F90=ftn</p><p>./configure \<br>    LDFLAGS="-static -I/opt/intel/parallel_studio_xe_2017_ce/mkl/include/  -I/opt/intel/parallel_studio_xe_2017_ce/mkl/include/intel64/lp64/" \<br>    BLAS_LIBS="/opt/intel/parallel_studio_xe_2017_ce/mkl/lib/intel64/libmkl_sequential.a /opt/intel/parallel_studio_xe_2017_ce/mkl/lib/intel64/libmkl_blacs_intelmpi_lp64.a -Wl,--end-group" \<br>    LAPACK_LIBS="/opt/intel/parallel_studio_xe_2017_ce/mkl/lib/intel64/libmkl_intel_lp64.a /opt/intel/parallel_studio_xe_2017_ce/mkl/lib/intel64/libmkl_core.a" \<br>    SCALAPACK_LIBS="/opt/intel/parallel_studio_xe_2017_ce/mkl/lib/intel64/libmkl_scalapack_lp64.a -Wl,--start-group" \<br>    FFT_LIBS="" \<br>    --prefix=/work/e05/e05/ian/qe </p><p>make all<br><span><span><br></span></span></p><span><p><br></p><span><p>I've tried a number of variations on the above using various combinations of 'include-hdf5' and associated module inclusions, but nothing seems to be 'catching'. So my questions are:</p><p>(1) Am I going in totally the wrong direction thinking hdf5 will solve these problems? If I am, what else could I try?</p><p>(2) If I'm not in the wrong direction, has anyone got any suggestions of what adjustments I could try for my input file?<br></p><span><p><br></p><span><p><br></p><p>With kindest thanks</p><p>Ian Shuttleworth</p><p>(Physics / Nottingham Trent University)</p><p><br></p><span><p><br></p><div dir="ltr"><br><span><span><br></span></span></div></span></span></span></span></span></span></div></div></div>