2014-12-30 3:22 GMT-07:00 Francesca Stanzione <[email protected]>: > Hi Christoph, > it seems to work now. The first step is started without any error. That is good news.
> I needed the table_bin=0.0005 because I am using the double precision. I > know that for a CG it is not really necessary but I need a CG potential for > an Adaptive Resolution simulation. Remember, that it is enough to interpolate the table to 0.0005, which VOTCA does automatically for step>table_bin. One does not need to sample the distributions and calculate the IBI update for this fine spacing. The only good (and common) reason for double precision is energy conservation, which only is fulfilled by H-AdResS. For the normal AdResS, which is implement in Gromacs, single precision is sufficient as the hybrid region needed to be heavily thermostated. In addition, in AdResS the CG region mainly acts as a simplified system to speedup the simulation, the interesting stuff happens in the atomistic region, which is covered by the atomistic force field. Making a too complicated CG model for an AdResS simulation might take longer than running a brute-force atomistic simulation ;-) However, good luck, Christoph > I want to thank you again for your time. > > Francesca > > On Dec 28, 2014, at 3:07 PM, Christoph Junghans <[email protected]> wrote: > > 2014-12-28 12:10 GMT-07:00 Christoph Junghans <[email protected]>: > > Hi, > > 2014-12-28 4:17 GMT-07:00 Francesca Stanzione > <[email protected]>: > > Hi Christoph, > unfortunately I tried to update the votca version but it is not working. I > have the same error. I changed the spacing in the settings.xml to a > reasonable one (0.002) and I also changed the spacing in the > ABB-angle.dist.tgt. > Any additional suggestion? > > Ok, I looked at the original problem again and there seems to be a > rounding problem with step/gromacs.table_bins=0.0005. > Basically numbers get rounded in each potential conversion step and > hence after a while one gets the "Illegal division by zero" as two x > values become too similar. I still have not found out, where exactly > it goes wrong and add a warning. > > I fixed the original issue: > <https://code.google.com/p/votca/source/detail?r=cc6ea63f95532302c3d97159b6175b6899266bc7&repo=tools> > > Please re-build tools and csg to get it. > > > In addition, Gromacs internally uses a spacing of 0.002 (single > precision) or 0.0005 (double precision), so 0.0005 would only make > sense with double precision Gromacs. But double precision isn't really > needed for coarse-grained simulation with thermostats. Have you tried > using gromacs.table_bins=0.002 ? > > With gromacs.table_bins=0.002 and the following patch, I got your > table running in hexane/ibi_bonded example. > diff -r a4ed014029f5 hexane/ibi_bonded/settings.xml > --- a/hexane/ibi_bonded/settings.xml Fri Oct 10 03:40:35 2014 -0600 > +++ b/hexane/ibi_bonded/settings.xml Sun Dec 28 11:54:25 2014 -0700 > @@ -16,12 +16,12 @@ > <bonded> > <!-- name of the interaction --> > <name>angle</name> > - <min>1.7</min> > + <min>0.9</min> > <max>3.1</max> > - <step>0.02</step> > + <step>0.002</step> > <inverse> > <!-- target distribution --> > - <target>angle.dist.tgt</target> > + <target>ABB-angle.dist.tgt</target> > <gromacs> > <table>table_a1.xvg</table> > </gromacs> > > But anyhow, step=0.002 is still 1100 bins which need to be filling > with statistics, which is a lot, you should aim for something like > 0.02. > Like I said in my last email, you might have better chances generating > a custom initial guess for your angle potential. > > Christoph > > > Thank you, > Francesca > > > 2014-12-19 13:24 GMT-05:00 Christoph Junghans <[email protected]>: > > > 2014-12-18 15:36 GMT-07:00 Francesca Stanzione > <[email protected]>: > > Dear all, > I am experiencing a problem to convert the angle potential in a > table.xvg > I have a polymer chain composed by 7 beads with two beadstypes A and B. > The > polymer sequence is: A1-B1-B2-B3-B4-B5-A2. I have 2 bondtypes (AB and > BB) > and two angletypes (ABB and BBB). > I obtained the dist.tgt. When I try to run the IBI it works for the > bonds > but not for the angles. Because I supposed I was doing something wrong I > tried the hexane and propane tutorials to see if something was wrong > with my > angles or my settings.xml and I am not able to obtained the tabulated > potential for angle neither for them. > This is the error I get: > ERROR: > # > # do_external: subscript > # > # /home/software/votca/share/votca/scripts/inverse/table_to_xvg.pl --max > 1000000 --type angle ABB-angle.pot.shift.YkwA2 table_a0.xvg # > # (from tags convert_potential xvg) failed > # > # For details see the logfile inverse.log > > > THIS IS the DETAILS FROM THE inverse.log > > or_all: run 'do_external convert_potential gromacs > $(csg_get_interaction_property name).pot.cur > $(csg_get_interaction_property > inverse.gromacs.table)' for interaction named 'ABB-angle' > Running subscript 'potential_to_gromacs.sh ABB-angle.pot.cur > table_a0.xvg' > (from tags convert_potential gromacs) dir > /home/software/votca/share/votca/scripts/inverse > Convert ABB-angle.pot.cur to table_a0.xvg > Running critical command 'sed -n s/.*cg_bonded\.\([^[:space:]]*\) > .*/\1/p' > Running critical command 'mktemp ABB-angle.pot.scale.XXXXX' > Running subscript 'table_linearop.pl --on-x ABB-angle.pot.cur > ABB-angle.pot.scale.dHEeQ 57.2957795 0' (from tags table linearop) dir > /home/software/votca/share/votca/scripts/inverse > table_linearop.pl: ABB-angle.pot.cur to ABB-angle.pot.scale.dHEeQ with > x' = > 57.2957795*x + 0 > Running critical command 'mktemp ABB-angle.pot.smooth.XXXXX' > Running critical command 'csg_resample --in ABB-angle.pot.scale.dHEeQ > --out > ABB-angle.pot.smooth.bZWVq --grid 0:0.0286479:180' > Running critical command 'mktemp ABB-angle.pot.extrapol.XXXXX' > Running subscript 'potential_extrapolate.sh --type angle > ABB-angle.pot.smooth.bZWVq ABB-angle.pot.extrapol.3ThKw' (from tags > potential extrapolate) dir > /home/software/votca/share/votca/scripts/inverse > Extrapolate ABB-angle.pot.smooth.bZWVq to ABB-angle.pot.extrapol.3ThKw > Running subscript 'table_extrapolate.pl --function linear --avgpoints 3 > --region leftright ABB-angle.pot.smooth.bZWVq > ABB-angle.pot.extrapol.3ThKw' > (from tags table extrapolate) dir > /home/software/votca/share/votca/scripts/inverse > table_extrapolate.pl: extrapolating the left using linear with gradient > -8.60697674418617 > table_extrapolate.pl: extrapolating the right using linear with gradient > -0.0167414651162821 > Running critical command 'mktemp ABB-angle.pot.interpol.XXXXX' > Running critical command 'csg_resample --in ABB-angle.pot.extrapol.3ThKw > --out ABB-angle.pot.interpol.j2iBK --grid 0:0.0005:180 --comment Created > on > Thu Dec 18 16:38:36 EST 2014 by fstanzio@silicon > called from potential_to_gromacs.sh, version 1.3-dev hgid: 4fce95a62b88 > settings file: > /home/Documents/bigger_box/reaction/ibi_cou_user/settings.xml > working directory: > /home/Documents/bigger_box/reaction/ibi_cou_user/step_001 > Comments from > > /home/Documents/bigger_box/reaction/ibi_cou_user/step_001/ABB-angle.pot.cur:\nCreated > on Thu Dec 18 16:38:30 EST 2014 by fstanzio@silicon > called from prepare_generic_single.sh, version 1.3-dev hgid: > 4fce95a62b88 > settings file: > /home/Documents/bigger_box/reaction/ibi_cou_user/settings.xml > working directory: > /home/Documents/bigger_box/reaction/ibi_cou_user/step_000' > Running critical command 'mktemp ABB-angle.pot.shift.XXXXX' > Running subscript 'potential_shift.pl --type angle > ABB-angle.pot.interpol.j2iBK ABB-angle.pot.shift.YkwA2' (from tags > potential > shift) dir /home/software/votca/share/votca/scripts/inverse > Running subscript 'table_to_xvg.pl --max 1000000 --type angle > ABB-angle.pot.shift.YkwA2 table_a0.xvg' (from tags convert_potential > xvg) > dir /home/software/votca/share/votca/scripts/inverse > Illegal division by zero at > > /home/fstanzio/software/votca/share/votca/scripts/inverse/table_to_xvg.pl > line 96. > > Checking the xxx.pot.shift file I have "o" on the last 5 values, I tried > to > resample, to change the --grid to 179.5 (last value with "i") but > nothing is > working.. > > I have the version 1.3-dev. > > I am sending you in addition the ABB-angle.dist.tgt and the settings.xml > files. > Can you please help me to understand what is the problem and why I am > not > able to have a table_angle? > > Ok, it looks like ABB-angle.dist.tgt gets converted to > ABB-angle.pot.cur in step_000 and the conversion from > ABB-angle.pot.cur to > table_a0.xvg. > > However, with the latest version of VOTCA 1.3-dev (changeset: > 1e124eb53c82) I wasn't able to reproduce this problem, so please as > 0th step re-install VOTCA and see if the problem persists. > (<https://code.google.com/p/votca/wiki/Installing#Development_Version>) > > Independently, after looking at your files, I have a couple of issues: > 1.) a spacing of 0.02 in ABB-angle.dist.tgt might be a bit to fine, > when looking at the curve around 2.35, it very rough. IBI will never > be able to fit that. You will have to smooth the target distribution > first. > 2.) a spacing of 0.0005 for ABB-angle in settings.xml is by far too > fine. You will have to run really really long cg simulation in each > iteration to fill the 4400 (=2.2/0.0005) bins of that interaction with > statistic. > Remember that the stepping for each interaction doesn't have to be the > same as table_bins in gromacs block. > > Cheers, > > Christoph > > > > > > Thank you, > Francesca > > -- > You received this message because you are subscribed to the Google > Groups > "votca" group. > To unsubscribe from this group and stop receiving emails from it, send > an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/votca. > For more options, visit https://groups.google.com/d/optout. > > > > > -- > Christoph Junghans > Web: http://www.compphys.de > > -- > You received this message because you are subscribed to a topic in the > Google Groups "votca" group. > To unsubscribe from this topic, visit > https://groups.google.com/d/topic/votca/wcSWx0n900s/unsubscribe. > To unsubscribe from this group and all its topics, send an email to > [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/votca. > For more options, visit https://groups.google.com/d/optout. > > > -- > You received this message because you are subscribed to the Google Groups > "votca" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/votca. > For more options, visit https://groups.google.com/d/optout. > > > > > -- > Christoph Junghans > Web: http://www.compphys.de > > > > > -- > Christoph Junghans > Web: http://www.compphys.de > > -- > You received this message because you are subscribed to a topic in the > Google Groups "votca" group. > To unsubscribe from this topic, visit > https://groups.google.com/d/topic/votca/wcSWx0n900s/unsubscribe. > To unsubscribe from this group and all its topics, send an email to > [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/votca. > For more options, visit https://groups.google.com/d/optout. > > > -- > You received this message because you are subscribed to the Google Groups > "votca" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/votca. > For more options, visit https://groups.google.com/d/optout. -- Christoph Junghans Web: http://www.compphys.de -- You received this message because you are subscribed to the Google Groups "votca" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/votca. For more options, visit https://groups.google.com/d/optout.
