<div>Matt,</div>
<div> </div>
<div>Just to follow up. Tried your suggestion of SKIP_OPTIMIZATION TRUE but that did not work.</div>
<div> </div>
<div>
<p> preconditioner : FULL_KINETIC : cholesky inversion of T + eS<br> stepsize : 0.15000000<br> energy_gap : 0.20000000<br> eps_taylor : 0.10000E-15<br> max_taylor : 4</p>
<p> ----------------------------------- OT --------------------------------------<br> *<br> *** 03:21:02 ERRORL2 in cp_fm_types:cp_fm_create processor 0 ***<br> *** err=-300 condition FAILED at line 169 ***<br>
*</p>
<p> ===== Routine Calling Stack =====</p>
<p> 7 make_preconditioner_single<br> 6 init_scf_loop<br> 5 scf_env_do_scf<br> 4 qs_energies<br> 3 qs_forces<br> 2 qs_mol_dyn_low<br> 1 CP2K<br>
CP2K| condition FAILED at line 169<br> CP2K| Abnormal program termination, stopped by process number 0<br>[0] [MPI Abort by user] Aborting Program!<br></p><br><br> </div>
<div><span class="gmail_quote">On 3/13/08, <b class="gmail_sendername">Matt W</b> <<a href="mailto:MattWa...@gmail.com">MattWa...@gmail.com</a>> wrote:</span>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid"><br>As to your problem with the 2D distribution, I don't know the code,<br>but the manual perhaps gives you a hint.<br>
<br>located at __ROOT__%FORCE_EVAL%DFT%QS%DISTRIBUTION<br><br>SKIP_OPTIMIZATION TRUE<br> Do not optimize the distribution, go for something very simple. Might<br>be useful if the optimization, which scales quadratically in system<br>
size, is too expensive.<br> This optional keyword expects a logical<br> This keyword behaves as a switch T . Default value: F<br><br>I have no idea how much it might impact on load balancing...<br><br>Matt<br>
<br><br>Reston, VA<br>443-567-8328 (C)<br>410-278-2692 (O)