<html>
<head>
<style>
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
FONT-SIZE: 10pt;
FONT-FAMILY:Tahoma
}
</style>
</head>
<body class='hmmessage'><div style="text-align: left;">Hi,<br><br>I now realized that my fix what completely incorrect.<br>I wanted to multiply the margin of 0.0001 by 2, not the factor 0.5.<br>But my mistake means that is now given an error when the off diagonal<br>element is relatively more than 1 instead of more than 0.5.<br>This is a far too weak check.<br>So I would not expect any fatal errors any more with this incorrect check.<br>Therefore I don't understand why you still get errors.<br><br>Could you try changing BOX_MARGIN to 0.5005, as it currently is in version 4.0?<br><br>Berk.<br><br></div><br><br><hr id="stopSpelling">> Date: Wed, 5 Mar 2008 09:40:58 +0100<br>> From: reich@mpikg.mpg.de<br>> To: gmx-users@gromacs.org<br>> Subject: [gmx-users] Skewed box using Parrinello-Rahman pressure coupling<br>> <br>> Hi Berk,<br>> <br>> some of my simulations crashed again, basically with the same error message:<br>> <br>> "Source code file: ns.c, line: 313<br>> <br>> Fatal error:<br>> Could not correct too skewed box"<br>> <br>> Right now I have factored all occurrences of BOX_MARGIN by 2 in<br>> src/gmxlib/pbc.c but kept the original value of 0.5001 for BOX_MARGIN in<br>> include/pbc.h. Could the bug maybe fixed by setting BOX_MARGIN to 0.5005<br>> while multiplying all instances in src/gmxlib/pbc.c by 2?<br>> <br>> <br>> Best<br>> <br>> Luther<br>> <br>> <br>> > All occurrences of BOX_MARGIN in pbc.c should have the factor of 2:<br>> ><br>> > #ifdef ALLOW_OFFDIAG_LT_HALFDIAG<br>> > else if ((fabs(box[YY][XX])+fabs(box[ZZ][XX])><br>> 2*BOX_MARGIN*box[XX][XX])<br>> > ||<br>> > (fabs(box[ZZ][YY])> 2*BOX_MARGIN*box[YY][YY]))<br>> > #else<br>> > else if ((fabs(box[YY][XX])> 2*BOX_MARGIN*box[XX][XX]) ||<br>> > (fabs(box[ZZ][XX])> 2*BOX_MARGIN*box[XX][XX]) ||<br>> > (fabs(box[ZZ][YY])> 2*BOX_MARGIN*box[YY][YY]))<br>> ><br>> > You can leave BOX_MARGIN at 0.5001.<br>> ><br>> > All these things are not critical for correct results.<br>> > The only annoying thing is that the "old" 3.3 code can give a fatal<br>> error<br>> > due to a rounding issue.<br>> ><br>> > Berk.<br>> ><br>> >> Date: Thu, 14 Feb 2008 14:31:15 +0100<br>> >> From: reich@mpikg.mpg.de<br>> >> To: gmx-users@gromacs.org<br>> >> Subject: [gmx-users] Skewed box using Parrinello-Rahman pressure coupling<br>> >> Hi Berk,<br>> >> thanks again. I am not sure though whether I have understood what<br>> modifications you have suggested. In src/gmxlib/pbc.c there are several<br>> occurrences of BOX_MARGIN, some with and some without a factor of two.<br>> Should each occurrence be multiplied by two (so those that have the<br>> factor<br>> >> already are left as they are and all others get multiplied)? Could you<br>> post the lines 71 to 76 of src/gmxlib/pbc.c with the right corrections<br>> so<br>> >> that I can be sure to implement them correctly?<br>> >> Also, should I leave the value for BOX_MARGIN in include/pbc.h at<br>> 0.5005<br>> >> or change it back to 0.5001 ?<br>> >> Best<br>> >> Luther<br>> >> > That good news.<br>> >> ><br>> >> > While looking at the code again just now,<br>> >> > I realized what the actual problem is.<br>> >> > The check before and after the box correction are both done with<br>> BOX_MARGIN,<br>> >> > which can lead to the problem you observed when the last bit of the<br>> >> box<br>> >> > is rounded in an unlucky fashion. So even with your change the<br>> problem<br>> >> > could still occur.<br>> >> > A real fix is adding a factor 2 (2*) in front of BOX_MARGIN in<br>> src/gmxlib/pbc.c.<br>> >> > I have added this fix for the 3.3.3 release.<br>> >> ><br>> >> > Berk.<br>> >> ><br>> >> >> Date: Wed, 13 Feb 2008 16:37:55 +0100<br>> >> >> From: reich@mpikg.mpg.de<br>> >> >> To: gmx-users@gromacs.org<br>> >> >> Subject: [gmx-users] Skewed box using Parrinello-Rahman pressure<br>> coupling<br>> >> >><br>> >> >> Hi Berk,<br>> >> >><br>> >> >> the change of parameter worked! My simulation finished without<br>> >> crashing.<br>> >> >><br>> >> >><br>> >> >> Thanks for your help<br>> >> >><br>> >> >> Luther<br>> >> >><br>> >> >><br>> >> >> >> Dear Gromacs-users,<br>> >> >> >><br>> >> >> >> I have been performing simulations with cubic and<br>> >> octahedron-shaped<br>> >> >> >> boxes<br>> >> >> >> using Parrinello-Rahman pressure coupling. After some time I get<br>> >> the<br>> >> >> >> error<br>> >> >> >> message "Could not correct too skewed box" and the simulations<br>> >> crash.<br>> >> >> >> I set the parameters for the coupling as listed below:<br>> >> >> >><br>> >> >> >> pcoupltype = isotropic<br>> >> >> >> tau_p = 5.0<br>> >> >> >> compressibility = 5e-5<br>> >> >> >> ref_p = 1.0<br>> >> >> >><br>> >> >> >><br>> >> >> >> I have never encountered the same problems when using Berendsen<br>> >> >> pressure<br>> >> >> >> coupling so I suppose I must have made a bad choice for some<br>> >> >> parameter<br>> >> >> >> for<br>> >> >> >> the Parrinello-Rahman barostat. I tried to equilibrate my system<br>> >> >> using a<br>> >> >> >> Berendsen barostat before switching to Parrinello-Rahman for 1ns<br>> >> (and<br>> >> >> >> tau_p=1.0) but this didnĀ“t help. Do you have any clue about where<br>> >> the<br>> >> >> >> problem originates from?<br>> >> >> >><br>> >> >> >><br>> >> >> >> Best<br>> >> >> >><br>> >> >> >> Luther<br>> >> >> ><br>> >> >> > I guess this is a rounding issue in the box checking/correcting<br>> >> >> routines.<br>> >> >> > Could you change BOX_MARGIN in include/pbc.h from 0.5001 to<br>> 0.5005,<br>> >> >> > recompile gmxlib, mdlib and mdrun and see if the problem still<br>> >> occurs?<br>> >> >> > Please report back, so we can fix this for 3.3.3.<br>> >> >> ><br>> >> >> > Berk.<br>> <br>> <br>> <br>> _______________________________________________<br>> gmx-users mailing list gmx-users@gromacs.org<br>> http://www.gromacs.org/mailman/listinfo/gmx-users<br>> Please search the archive at http://www.gromacs.org/search before posting!<br>> Please don't post (un)subscribe requests to the list. Use the <br>> www interface or send it to gmx-users-request@gromacs.org.<br>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php<br><br /><hr />Express yourself instantly with MSN Messenger! <a href='http://clk.atdmt.com/AVE/go/onm00200471ave/direct/01/' target='_new'>MSN Messenger</a></body>
</html>