<br><br>On Tuesday, January 24, 2012 10:02:34 PM UTC-5, Chetan Mahajan wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">Hello Friends<div><br></div><div>I am new to cp2k or any quantum calculations, carrying out Born-Oppenheimer molecular dynamics simulations of water solvated acid-base polymer systems. I have some questions on our first output:</div>
<div><br></div><div>
1. Any good reference which would explain all the terms in cp2k output? I have Essentials of Computational Chemistry by Cramer and it is pretty good, but it does not use same or all the terms as cp2k output, so it is confusing for a beginner like me. </div></blockquote><div><br></div><div>the cramer book is very "essential" and does not touch a lot of what is relevant for cp2k.</div><div>a lot of the "slang" related to cp2k comes from the condensed matter physics / electronic</div><div>structure area and is best understood, if you also gain some understanding of pure</div><div>plane wave basis set DFT calculations (e.g. in the form of car-parrinello MD).</div><div><br></div><div>a good introduction that is palatable for people with a chemistry training (there are</div><div>other good ones, but they are more accessible to people with a strong physics</div><div>background), is in my opinion "electronic structure calculations for solid and molecules:</div><div>theory and computational methods" by jorge kohanoff. </div><div><br></div><div>from there on, you can work your way through some of the cp2k papers</div><div>and more specific books. as a rule of the thumb you should keep in mind</div><div>that the integral of all evil things remains a constant, i.e. for every benefit</div><div>there is a disadvantage. which means, that you have to expect to make</div><div>some serious mistakes until you get to know those problems. it is </div><div>therefore highly recommended to not start with something too difficult</div><div>or too complex. that will just increase the number of possibilities for</div><div>errors and decrease the chances to detect them.</div><div><br></div><blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">
<div><br></div><div>2. For a classically MD equilibrated input to AIMD, initial energy is -3128.122 H, but for a random, non-equilibrated input (which by the way was my input to classical MD equilibration) to AIMD, initial energy is -3129.058 H. My question is why is energy decreasing for a non-equilibrated system? I understand we are dealing with ground-states only, but shouldn't quantum mechanical energy be higher if system is not classically equilibrated in general? Is this pointing to any error?</div></blockquote><div><br></div><div>this is very difficult to say without knowing the exact difference between</div><div>your equilibrated and non-equilibrated system. do they have the same </div><div>volume? are they both neutral? if the difference is rather small, it can just</div><div>be that what is equilibrated with a classical model has some too close</div><div>contacts for your DFT calculation. also keep in mind that (simple) DFT</div><div>(e.g. BLYP) is notoriously bad at representing dispersion interactions.</div><div><br></div><div>it could also just mean that you may not have converged one of the </div><div>wavefunctions to the real ground state. there are many things that</div><div>can go wrong and it is difficult judge only from the total energy.</div><div><br></div><div>axel.</div><div><br></div><div><br></div><div> </div><blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;"><div></div><div>Thanks</div><div>Chetan</div>
<div><div><br>
</div>-- <br>Chetan V Mahajan<br>PhD candidate, Ganesan Group<br>Department of Chemical Engineering<br>The University of Texas at Austin<br><br>
</div>
</blockquote>