Hi Oliver<div><br></div><div>Can't really help other than to say I have observed this as well.</div><div><br></div><div><div>Kind Regards</div><div><br></div><div>Sam</div><br></div><div class="gmail_quote"><div dir="auto" class="gmail_attr">On Wednesday, May 18, 2022 at 5:43:13 PM UTC+2 Oliver wrote:<br/></div><blockquote class="gmail_quote" style="margin: 0 0 0 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">Hi cp2k users,<div>I would like to address a concern which i came through recently when my cp2k program crashed for some reason. I'm doing a geometry optimization on an atomic system with the Gradient Conjugate option (CG) and default values for CG keywords. While restarting the program, i discovered that the two "new" first iterations were done with the Steepest Descent method (SD) and that the third one was using CG method. Unfortunately, the System's energy jumps a bit from the last value calculated before cp2k crashed while using CG during this iteration. As a consequence, the atomic configuration is blurred and it seems that the optimization process couldn't return anymore to the optimization path found before the crash. </div><div>I tried to play with keywords Max_Steep_Steps and Restart_Limit, but nothing changes.</div><div>Does anyone could give me some advice on how to force cp2k to only use CG for the "new" iterations when restarting the program.</div><div>Thanks in advance for the help.</div><div>Oliver</div></blockquote></div>
<p></p>
-- <br />
You received this message because you are subscribed to the Google Groups "cp2k" group.<br />
To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:cp2k+unsubscribe@googlegroups.com">cp2k+unsubscribe@googlegroups.com</a>.<br />
To view this discussion on the web visit <a href="https://groups.google.com/d/msgid/cp2k/67e2a49c-a470-4a6b-9967-ccdd79a6785fn%40googlegroups.com?utm_medium=email&utm_source=footer">https://groups.google.com/d/msgid/cp2k/67e2a49c-a470-4a6b-9967-ccdd79a6785fn%40googlegroups.com</a>.<br />