<html><head><meta http-equiv="Content-Type" content="text/html charset=utf-8"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;" class=""><br class=""><div><blockquote type="cite" class=""><div class="">On 26 Jan 2016, at 20:02, Paolo Giannozzi <<a href="mailto:p.giannozzi@gmail.com" class="">p.giannozzi@gmail.com</a>> wrote:</div><br class="Apple-interchange-newline"><div class=""><div dir="ltr" class=""><br class=""><div class="gmail_extra"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr" class=""><div class="">One should obtain the same numbers - within the errors due to roundoff, though - in serial, OpenMP, MPI execution, and on different machines, and with different compilers and mathematical libraries.</div></div></blockquote><div class=""><br class=""></div><div class="h5">... actually, only at exact convergence, at least when using default wavefunction initialization: small differences come from different randomization of atomic wavefunctions. These however disappear as convergence is made tighter (but sometimes they don’t</div></div></div></div></div></blockquote><div><br class=""></div><div><br class=""></div></div>Thanks Paolo & Stefano, that’s very useful information.<div class=""><br class=""></div><div class="">The convergence threshold was at 1.0e-9 and when I tightened it to 1.0e-12 it improved the agreement significantly.</div><div class=""><br class=""></div><div class="">Best regards,</div><div class="">Nick</div></body></html>