<html><head><meta http-equiv="Content-Type" content="text/html charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;" class="">Dear Daniele, <div class=""><br class=""></div><div class="">in the LBFGS method convergence used to be decided upon the projected </div><div class="">gradient and the relative force error only, the convergence checks are just </div><div class="">printed for information. You can use the BFGS method instead, which converges </div><div class="">only if all checks are met. However, the implementation details are slightly </div><div class="">different, which is why typically more BFGS steps are required. </div><div class=""><br class=""></div><div class="">Best, </div><div class="">Thomas</div><div class=""><br class=""><div><blockquote type="cite" class=""><div class="">Am 22.02.2019 um 14:39 schrieb Daniele Ongari <<a href="mailto:daniele...@gmail.com" class="">daniele...@gmail.com</a>>:</div><br class="Apple-interchange-newline"><div class=""><div dir="ltr" class=""><div class="">Dear CP2K mailing list,</div><div class=""><br class=""></div><div class="">We experience a strange behavior of the L-BFGS optimization algorithm when dealing with cases where SCF convergence is problematic.</div><div class=""><br class=""></div><div class="">After 20 minimization steps the cp2k printed the following:</div><div class=""><br class=""></div><div class=""><br class=""></div><div class="">-------- Informations at step = 21 ------------</div><div class=""> Optimization Method = LBFGS</div><div class=""> Total Energy = -2345.6430825804</div><div class=""> Internal Pressure [bar] = 23.9760390703</div><div class=""> Real energy change = 0.0028283710</div><div class=""> Decrease in energy = NO</div><div class=""> Used time = 5864.105</div><div class=""><br class=""></div><div class=""> Convergence check :</div><div class=""> Max. step size = 0.0000000000</div><div class=""> Conv. limit for step size = 0.0030000000</div><div class=""> Convergence in step size = YES</div><div class=""> RMS step size = 0.0000000000</div><div class=""> Conv. limit for RMS step = 0.0015000000</div><div class=""> Convergence in RMS step = YES</div><div class=""> Max. gradient = 0.0082916384</div><div class=""> Conv. limit for gradients = 0.0004500000</div><div class=""> Conv. for gradients = NO</div><div class=""> RMS gradient = 0.0009846623</div><div class=""> Conv. limit for RMS grad. = 0.0003000000</div><div class=""> Conv. for gradients = NO</div><div class=""> Pressure Deviation [bar] = -76.0239609297</div><div class=""> Pressure Tolerance [bar] = 100.0000000000</div><div class=""> Conv. for PRESSURE = YES</div><div class="">---------------------------------------------------</div><div class=""><br class=""></div><div class="">--------------------------</div><div class="">OPTIMIZATION STEP: 22</div><div class="">--------------------------</div><div class=""><br class=""></div><div class="">***********************************************</div><div class="">* Specific L-BFGS convergence criteria</div><div class="">* WANTED_PROJ_GRADIENT and WANTED_REL_F_ERROR</div><div class="">* satisfied .... run CONVERGED!</div><div class="">***********************************************</div><div class=""><br class=""></div><div class=""><br class=""></div><div class=""><br class=""></div><div class="">The two things that look strange for us are:</div><div class=""><br class=""></div><div class="">1) Max. step size and RMS step size were 'artificially' set to 0.0000000000</div><div class="">2) Not all checks are showing convergence, while L-BFGS algorithm has decided to stop.</div><div class=""><br class=""></div><div class=""><br class=""></div><div class="">Therefore, we would like to know if this is an expected behavior of the L-BFGS algorithm and if yes, what is the reason behind working this way?</div><div class=""><br class=""></div><div class="">Please see input and output files attached.</div><div class=""><br class=""></div><div class="">With best regards,</div><div class="">Daniele and Sasha</div></div><div class=""><br class="webkit-block-placeholder"></div> -- <br class=""> You received this message because you are subscribed to the Google Groups "cp2k" group.<br class=""> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:cp2k+uns...@googlegroups.com" class="">cp2k+uns...@googlegroups.com</a>.<br class=""> To post to this group, send email to <a href="mailto:cp...@googlegroups.com" class="">cp...@googlegroups.com</a>.<br class=""> Visit this group at <a href="https://groups.google.com/group/cp2k" class="">https://groups.google.com/group/cp2k</a>.<br class=""> For more options, visit <a href="https://groups.google.com/d/optout" class="">https://groups.google.com/d/optout</a>.<br class=""> <span id="cid:f400f734-2996-46ef-923a-cadae074c4ca"><aiida.coords.xyz></span><span id="cid:f03c146d-b8a0-4b53-a0f2-5b9e705c666d"><aiida.inp></span><span id="cid:5f76de14-9f38-41d1-bab4-3eb45f991e05"><aiida.out></span><span id="cid:55a8cc0c-3b59-47be-9b88-12ddd6898eaa"><aiida-pos-1.xyz></span></div></blockquote></div><br class=""></div></body></html>