Okay, I am running "dftb+_s081217.i686-linux", I turned the scc back on, and I still don't get convergence. (and the system looks like it exploded). Would that hint at a bad initial coordinate input? How quickly did your's converge? Why would they be different?<br>
<br>Thanks!<br><br>Here is the end result of 100 cg steps and 100 scc steps:<br><br> 95 -0.72192733E+02 0.21030738E-02 0.16794511E-01<br> 96 -0.72194720E+02 -0.19864693E-02 0.13569232E-01<br> 97 -0.72195137E+02 -0.41683246E-03 0.14380645E-01<br>
98 -0.72192421E+02 0.27155471E-02 0.25102920E-01<br> 99 -0.72192982E+02 -0.56062346E-03 0.16674836E-01<br> 100 -0.72191969E+02 0.10125796E-02 0.17576861E-01<br> <br> Total Energy: -70.920240<br>
Total Mermin free energy: -70.920240<br>WARNING!<br>-> SCC is NOT converged, maximal iterations exceeded<br> Maximal force component: 0.051841<br>>> Charges saved for restart in charges.bin<br>
WARNING!<br>-> !!! Geometry did NOT converge!<br> !!! Geometry did NOT converge!<br><br><br>Here's the input:<br><br>#3x3 unit cell: Does not converge!<br>Geometry = GenFormat {<br> 72 S<br>C H <br>1 1 -3.7979 -0.2582 -6.5783<br>
2 1 -2.5319 0.2584 -5.8473<br>3 1 -2.5319 -0.2582 -4.3855<br>4 1 -3.7979 0.2584 -3.6545<br>5 2 -3.7979 -1.3798 -6.5783<br>6 2 -2.5319 1.3800 -5.8473<br>7 2 -2.5319 -1.3798 -4.3855<br>8 2 -3.7979 1.3800 -3.6545<br>9 1 -1.2659 -0.2582 -6.5783<br>
10 1 0.0001 0.2584 -5.8473<br>11 1 0.0001 -0.2582 -4.3855<br>12 1 -1.2659 0.2584 -3.6545<br>13 2 -1.2659 -1.3798 -6.5783<br>14 2 0.0001 1.3800 -5.8473<br>15 2 0.0001 -1.3798 -4.3855<br>16 2 -1.2659 1.3800 -3.6545<br>17 1 1.2661 -0.2582 -6.5783<br>
18 1 2.5321 0.2584 -5.8473<br>19 1 2.5321 -0.2582 -4.3855<br>20 1 1.2661 0.2584 -3.6545<br>21 2 1.2661 -1.3798 -6.5783<br>22 2 2.5321 1.3800 -5.8473<br>23 2 2.5321 -1.3798 -4.3855<br>24 2 1.2661 1.3800 -3.6545<br>25 1 -3.7979 -0.2582 -2.1927<br>
26 1 -2.5319 0.2584 -1.4617<br>27 1 -2.5319 -0.2582 0.0001<br>28 1 -3.7979 0.2584 0.7310<br>29 2 -3.7979 -1.3798 -2.1927<br>30 2 -2.5319 1.3800 -1.4617<br>31 2 -2.5319 -1.3798 0.0001<br>32 2 -3.7979 1.3800 0.7310<br>33 1 -1.2659 -0.2582 -2.1927<br>
34 1 0.0001 0.2584 -1.4617<br>35 1 0.0001 -0.2582 0.0001<br>36 1 -1.2659 0.2584 0.7310<br>37 2 -1.2659 -1.3798 -2.1927<br>38 2 0.0001 1.3800 -1.4617<br>39 2 0.0001 -1.3798 0.0001<br>40 2 -1.2659 1.3800 0.7310<br>41 1 1.2661 -0.2582 -2.1927<br>
42 1 2.5321 0.2584 -1.4617<br>43 1 2.5321 -0.2582 0.0001<br>44 1 1.2661 0.2584 0.7310<br>45 2 1.2661 -1.3798 -2.1927<br>46 2 2.5321 1.3800 -1.4617<br>47 2 2.5321 -1.3798 0.0001<br>48 2 1.2661 1.3800 0.7310<br>49 1 -3.7979 -0.2582 2.1929<br>
50 1 -2.5319 0.2584 2.9238<br>51 1 -2.5319 -0.2582 4.3857<br>52 1 -3.7979 0.2584 5.1166<br>53 2 -3.7979 -1.3798 2.1929<br>54 2 -2.5319 1.3800 2.9238<br>55 2 -2.5319 -1.3798 4.3857<br>56 2 -3.7979 1.3800 5.1166<br>57 1 -1.2659 -0.2582 2.1929<br>
58 1 0.0001 0.2584 2.9238<br>59 1 0.0001 -0.2582 4.3857<br>60 1 -1.2659 0.2584 5.1166<br>61 2 -1.2659 -1.3798 2.1929<br>62 2 0.0001 1.3800 2.9238<br>63 2 0.0001 -1.3798 4.3857<br>64 2 -1.2659 1.3800 5.1166<br>65 1 1.2661 -0.2582 2.1929<br>
66 1 2.5321 0.2584 2.9238<br>67 1 2.5321 -0.2582 4.3857<br>68 1 1.2661 0.2584 5.1166<br>69 2 1.2661 -1.3798 2.1929<br>70 2 2.5321 1.3800 2.9238<br>71 2 2.5321 -1.3798 4.3857<br>72 2 1.2661 1.3800 5.1166<br>0.0 0.0 0.0<br>
7.5960000000000001 0.0 0.0<br>0.0 20.000000000000000 0.0<br>0.0 0.0 13.156727759999999<br>}<br>Driver = ConjugateGradient {<br> MovedAtoms = Range { 1 -1 }<br> MaxForceComponent [eV/AA] = 1.0e-2<br> MaxSteps = 100<br>
# StepSize = 0.1<br> OutputPrefix = "geo_end"<br>}<br>Hamiltonian = DFTB {<br> SCC = Yes<br> MaxAngularMomentum = {<br> C = "p"<br> H = "s"<br> }<br> SlaterKosterFiles = Type2FileNames {<br>
Prefix = "/home/schall2/Research/splines/slako/pbc-0-2/"<br> Separator = "-"<br> Suffix = ".skf"<br> }<br> KPointsAndWeights = SupercellFolding {<br> 2 0 0<br> 0 2 0<br>
0 0 2 <br> 0.5 0.5 0.5<br> }<br> Eigensolver = Standard{}<br>}<br><br><div class="gmail_quote">On Fri, Mar 19, 2010 at 11:59 AM, <span dir="ltr"><<a href="mailto:dftb-plus-user-request@dftb-plus.info">dftb-plus-user-request@dftb-plus.info</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Send DFTB-Plus-User mailing list submissions to<br>
<a href="mailto:dftb-plus-user@dftb-plus.info">dftb-plus-user@dftb-plus.info</a><br>
<br>
To subscribe or unsubscribe via the World Wide Web, visit<br>
<a href="http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user" target="_blank">http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user</a><br>
or, via email, send a message with subject or body 'help' to<br>
<a href="mailto:dftb-plus-user-request@dftb-plus.info">dftb-plus-user-request@dftb-plus.info</a><br>
<br>
You can reach the person managing the list at<br>
<a href="mailto:dftb-plus-user-owner@dftb-plus.info">dftb-plus-user-owner@dftb-plus.info</a><br>
<br>
When replying, please edit your Subject line so it is more specific<br>
than "Re: Contents of DFTB-Plus-User digest..."<br>
<br>
<br>
Today's Topics:<br>
<br>
1. Convergence issues? (Dave Schall)<br>
2. Re: Convergence issues? (B?lint Aradi)<br>
<br>
<br>
----------------------------------------------------------------------<br>
<br>
Message: 1<br>
Date: Fri, 19 Mar 2010 11:39:36 -0400<br>
From: Dave Schall <<a href="mailto:schall2@oakland.edu">schall2@oakland.edu</a>><br>
Subject: [DFTB-Plus-User] Convergence issues?<br>
To: <a href="mailto:dftb-plus-user@dftb-plus.info">dftb-plus-user@dftb-plus.info</a><br>
Message-ID:<br>
<<a href="mailto:29919bb41003190839i1bd64f03r1263b15dc75723ea@mail.gmail.com">29919bb41003190839i1bd64f03r1263b15dc75723ea@mail.gmail.com</a>><br>
Content-Type: text/plain; charset="iso-8859-1"<br>
<br>
Hello,<br>
I'm trying to do an energy minimization of slab-like planar structure and I<br>
cannot get convergence of steepest descents or cg method. In a separate, but<br>
possibly related issue, I can't get the scc calculation to converge either.<br>
I started with a 1x1 unit cell "slab" of the material, then a 2x2 unit cell<br>
slab. These both converge in around 8 iterations. If I scale up to 3x3 cells<br>
or greater, I do not achieve convergence and the energy oscilates or<br>
actually goes up. Any suggestions for achieving convergence for energy<br>
minimization and / or scc? Ideally I'd like to scale up to a 10x5 unit cell<br>
slab for my calculations.<br>
<br>
Attached are the input files for a 2x2 and 3x3 unit cell energy minimization<br>
calculations with scc off.<br>
<br>
Thanks in advance for your assistance!<br>
<br>
Dave Schall<br>
<br>
#2x2 unit cell: Converges just fine.<br>
Geometry = GenFormat {<br>
32 S<br>
C H<br>
1 1 -2.5319 -0.2582 -4.3855<br>
2 1 -1.2659 0.2584 -3.6545<br>
3 1 -1.2659 -0.2582 -2.1927<br>
4 1 -2.5319 0.2584 -1.4617<br>
5 2 -2.5319 -1.3798 -4.3855<br>
6 2 -1.2659 1.3800 -3.6545<br>
7 2 -1.2659 -1.3798 -2.1927<br>
8 2 -2.5319 1.3800 -1.4617<br>
9 1 0.0001 -0.2582 -4.3855<br>
10 1 1.2661 0.2584 -3.6545<br>
11 1 1.2661 -0.2582 -2.1927<br>
12 1 0.0001 0.2584 -1.4617<br>
13 2 0.0001 -1.3798 -4.3855<br>
14 2 1.2661 1.3800 -3.6545<br>
15 2 1.2661 -1.3798 -2.1927<br>
16 2 0.0001 1.3800 -1.4617<br>
17 1 -2.5319 -0.2582 0.0001<br>
18 1 -1.2659 0.2584 0.7310<br>
19 1 -1.2659 -0.2582 2.1929<br>
20 1 -2.5319 0.2584 2.9238<br>
21 2 -2.5319 -1.3798 0.0001<br>
22 2 -1.2659 1.3800 0.7310<br>
23 2 -1.2659 -1.3798 2.1929<br>
24 2 -2.5319 1.3800 2.9238<br>
25 1 0.0001 -0.2582 0.0001<br>
26 1 1.2661 0.2584 0.7310<br>
27 1 1.2661 -0.2582 2.1929<br>
28 1 0.0001 0.2584 2.9238<br>
29 2 0.0001 -1.3798 0.0001<br>
30 2 1.2661 1.3800 0.7310<br>
31 2 1.2661 -1.3798 2.1929<br>
32 2 0.0001 1.3800 2.9238<br>
0.0 0.0 0.0<br>
5.0640000000000001 0.0 0.0<br>
0.0 20.000000000000000 0.0<br>
0.0 0.0 8.7711518399999999<br>
}<br>
Driver = ConjugateGradient {<br>
MovedAtoms = Range { 1 -1 }<br>
MaxForceComponent [eV/AA] = 1.0e-2<br>
MaxSteps = 100<br>
# StepSize = 0.1<br>
OutputPrefix = "geo_end"<br>
}<br>
Hamiltonian = DFTB {<br>
SCC = No<br>
MaxAngularMomentum = {<br>
C = "p"<br>
H = "s"<br>
# Eigensolver = Standard{}<br>
}<br>
SlaterKosterFiles = Type2FileNames {<br>
Prefix = "/home/schall2/Research/<br>
splines/slako/pbc-0-2/"<br>
Separator = "-"<br>
Suffix = ".skf"<br>
}<br>
KPointsAndWeights = SupercellFolding {<br>
2 0 0<br>
0 2 0<br>
0 0 2<br>
0.5 0.5 0.5<br>
}<br>
Eigensolver = Standard{}<br>
}<br>
<br>
#3x3 unit cell: Does not converge!<br>
Geometry = GenFormat {<br>
72 S<br>
C H<br>
1 1 -3.7979 -0.2582 -6.5783<br>
2 1 -2.5319 0.2584 -5.8473<br>
3 1 -2.5319 -0.2582 -4.3855<br>
4 1 -3.7979 0.2584 -3.6545<br>
5 2 -3.7979 -1.3798 -6.5783<br>
6 2 -2.5319 1.3800 -5.8473<br>
7 2 -2.5319 -1.3798 -4.3855<br>
8 2 -3.7979 1.3800 -3.6545<br>
9 1 -1.2659 -0.2582 -6.5783<br>
10 1 0.0001 0.2584 -5.8473<br>
11 1 0.0001 -0.2582 -4.3855<br>
12 1 -1.2659 0.2584 -3.6545<br>
13 2 -1.2659 -1.3798 -6.5783<br>
14 2 0.0001 1.3800 -5.8473<br>
15 2 0.0001 -1.3798 -4.3855<br>
16 2 -1.2659 1.3800 -3.6545<br>
17 1 1.2661 -0.2582 -6.5783<br>
18 1 2.5321 0.2584 -5.8473<br>
19 1 2.5321 -0.2582 -4.3855<br>
20 1 1.2661 0.2584 -3.6545<br>
21 2 1.2661 -1.3798 -6.5783<br>
22 2 2.5321 1.3800 -5.8473<br>
23 2 2.5321 -1.3798 -4.3855<br>
24 2 1.2661 1.3800 -3.6545<br>
25 1 -3.7979 -0.2582 -2.1927<br>
26 1 -2.5319 0.2584 -1.4617<br>
27 1 -2.5319 -0.2582 0.0001<br>
28 1 -3.7979 0.2584 0.7310<br>
29 2 -3.7979 -1.3798 -2.1927<br>
30 2 -2.5319 1.3800 -1.4617<br>
31 2 -2.5319 -1.3798 0.0001<br>
32 2 -3.7979 1.3800 0.7310<br>
33 1 -1.2659 -0.2582 -2.1927<br>
34 1 0.0001 0.2584 -1.4617<br>
35 1 0.0001 -0.2582 0.0001<br>
36 1 -1.2659 0.2584 0.7310<br>
37 2 -1.2659 -1.3798 -2.1927<br>
38 2 0.0001 1.3800 -1.4617<br>
39 2 0.0001 -1.3798 0.0001<br>
40 2 -1.2659 1.3800 0.7310<br>
41 1 1.2661 -0.2582 -2.1927<br>
42 1 2.5321 0.2584 -1.4617<br>
43 1 2.5321 -0.2582 0.0001<br>
44 1 1.2661 0.2584 0.7310<br>
45 2 1.2661 -1.3798 -2.1927<br>
46 2 2.5321 1.3800 -1.4617<br>
47 2 2.5321 -1.3798 0.0001<br>
48 2 1.2661 1.3800 0.7310<br>
49 1 -3.7979 -0.2582 2.1929<br>
50 1 -2.5319 0.2584 2.9238<br>
51 1 -2.5319 -0.2582 4.3857<br>
52 1 -3.7979 0.2584 5.1166<br>
53 2 -3.7979 -1.3798 2.1929<br>
54 2 -2.5319 1.3800 2.9238<br>
55 2 -2.5319 -1.3798 4.3857<br>
56 2 -3.7979 1.3800 5.1166<br>
57 1 -1.2659 -0.2582 2.1929<br>
58 1 0.0001 0.2584 2.9238<br>
59 1 0.0001 -0.2582 4.3857<br>
60 1 -1.2659 0.2584 5.1166<br>
61 2 -1.2659 -1.3798 2.1929<br>
62 2 0.0001 1.3800 2.9238<br>
63 2 0.0001 -1.3798 4.3857<br>
64 2 -1.2659 1.3800 5.1166<br>
65 1 1.2661 -0.2582 2.1929<br>
66 1 2.5321 0.2584 2.9238<br>
67 1 2.5321 -0.2582 4.3857<br>
68 1 1.2661 0.2584 5.1166<br>
69 2 1.2661 -1.3798 2.1929<br>
70 2 2.5321 1.3800 2.9238<br>
71 2 2.5321 -1.3798 4.3857<br>
72 2 1.2661 1.3800 5.1166<br>
0.0 0.0 0.0<br>
7.5960000000000001 0.0 0.0<br>
0.0 20.000000000000000 0.0<br>
0.0 0.0 13.156727759999999<br>
}<br>
Driver = ConjugateGradient {<br>
MovedAtoms = Range { 1 -1 }<br>
MaxForceComponent [eV/AA] = 1.0e-2<br>
MaxSteps = 100<br>
# StepSize = 0.1<br>
OutputPrefix = "geo_end"<br>
}<br>
Hamiltonian = DFTB {<br>
SCC = No<br>
MaxAngularMomentum = {<br>
C = "p"<br>
H = "s"<br>
# Eigensolver = Standard{}<br>
}<br>
SlaterKosterFiles = Type2FileNames {<br>
Prefix = "/home/schall2/Research/splines/slako/pbc-0-2/"<br>
Separator = "-"<br>
Suffix = ".skf"<br>
}<br>
KPointsAndWeights = SupercellFolding {<br>
2 0 0<br>
0 2 0<br>
0 0 2<br>
0.5 0.5 0.5<br>
}<br>
Eigensolver = Standard{}<br>
}<br>
<br>
<br>
--<br>
J. David Schall, Assistant Professor<br>
Dept. of Mechanical Engineering, Oakland University<br>
130 DHE, Rochester, MI 48309<br>
248-370-2870<br>
-------------- next part --------------<br>
An HTML attachment was scrubbed...<br>
URL: <a href="http://www.dftb-plus.info/pipermail/dftb-plus-user/attachments/20100319/deefa1c6/attachment-0001.htm" target="_blank">http://www.dftb-plus.info/pipermail/dftb-plus-user/attachments/20100319/deefa1c6/attachment-0001.htm</a><br>
<br>
------------------------------<br>
<br>
Message: 2<br>
Date: Fri, 19 Mar 2010 16:59:11 +0100<br>
From: B?lint Aradi <<a href="mailto:balint.aradi@bccms.uni-bremen.de">balint.aradi@bccms.uni-bremen.de</a>><br>
Subject: Re: [DFTB-Plus-User] Convergence issues?<br>
To: User list for DFTB+ related questions<br>
<<a href="mailto:dftb-plus-user@dftb-plus.info">dftb-plus-user@dftb-plus.info</a>><br>
Message-ID: <<a href="mailto:4BA39F4F.8000706@bccms.uni-bremen.de">4BA39F4F.8000706@bccms.uni-bremen.de</a>><br>
Content-Type: text/plain; charset="utf-8"<br>
<br>
Dear Dave Schall,<br>
<br>
> I'm trying to do an energy minimization of slab-like planar structure and I<br>
> cannot get convergence of steepest descents or cg method. In a separate, but<br>
> possibly related issue, I can't get the scc calculation to converge either.<br>
> I started with a 1x1 unit cell "slab" of the material, then a 2x2 unit cell<br>
> slab. These both converge in around 8 iterations. If I scale up to 3x3 cells<br>
> or greater, I do not achieve convergence and the energy oscilates or<br>
> actually goes up. Any suggestions for achieving convergence for energy<br>
> minimization and / or scc? Ideally I'd like to scale up to a 10x5 unit cell<br>
> slab for my calculations.<br>
><br>
> Attached are the input files for a 2x2 and 3x3 unit cell energy minimization<br>
> calculations with scc off.<br>
<br>
Actually I could not reproduce your problem. I took the 3x3 unit cell<br>
you sent, turned SCC on (which you should, as the pbc set is for SCC<br>
calculations) and got a converged geometry. (See attached input/output).<br>
I used dftb+_s081217, which I recommend to use.<br>
<br>
Best regards,<br>
<br>
B?lint<br>
<br>
--<br>
Dr. B?lint Aradi<br>
Bremen Center for Computational Materials Science, University of Bremen<br>
<a href="http://www.bccms.uni-bremen.de/en/%7Eb_aradi" target="_blank">http://www.bccms.uni-bremen.de/en/~b_aradi</a><br>
<br>
-------------- next part --------------<br>
An embedded and charset-unspecified text was scrubbed...<br>
Name: output<br>
Url: <a href="http://www.dftb-plus.info/pipermail/dftb-plus-user/attachments/20100319/7ed23555/attachment.txt" target="_blank">http://www.dftb-plus.info/pipermail/dftb-plus-user/attachments/20100319/7ed23555/attachment.txt</a><br>
-------------- next part --------------<br>
An embedded and charset-unspecified text was scrubbed...<br>
Name: dftb_in.hsd<br>
Url: <a href="http://www.dftb-plus.info/pipermail/dftb-plus-user/attachments/20100319/7ed23555/attachment-0001.txt" target="_blank">http://www.dftb-plus.info/pipermail/dftb-plus-user/attachments/20100319/7ed23555/attachment-0001.txt</a><br>
-------------- next part --------------<br>
A non-text attachment was scrubbed...<br>
Name: signature.asc<br>
Type: application/pgp-signature<br>
Size: 252 bytes<br>
Desc: OpenPGP digital signature<br>
Url : <a href="http://www.dftb-plus.info/pipermail/dftb-plus-user/attachments/20100319/7ed23555/attachment.pgp" target="_blank">http://www.dftb-plus.info/pipermail/dftb-plus-user/attachments/20100319/7ed23555/attachment.pgp</a><br>
<br>
------------------------------<br>
<br>
_______________________________________________<br>
DFTB-Plus-User mailing list<br>
<a href="mailto:DFTB-Plus-User@dftb-plus.info">DFTB-Plus-User@dftb-plus.info</a><br>
<a href="http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user" target="_blank">http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user</a><br>
<br>
<br>
End of DFTB-Plus-User Digest, Vol 27, Issue 7<br>
*********************************************<br>
</blockquote></div><br><br clear="all"><br>-- <br>J. David Schall, Assistant Professor<br>Dept. of Mechanical Engineering, Oakland University<br>130 DHE, Rochester, MI 48309<br>248-370-2870<br>