I likely won't have much time to look at it tonight, but you can see
exactly what the option is doing to the topology. run gmxdump on the
tpr. All of the stuff that couple-intramol does is in grompp, so the
results will show up in the detailed listings of the interactions, and
which ones have
I think the grammar got a little garbled there, so I'm not sure quite
what you are claiming.
One important thing to remember; 1-4 interactions are treated as
bonded interactions right now FOR COUPLE intramol (not for lambda
dependence of the potential energy function), so whether
couple-intramol
. Not that many people have used this
code, thus there are likely ways that it can be improved for better
utility. I generally haven't tried to calculate free energy
differences from the transition matrix, though it should give
consistent results with the other methods.
Best,
Michael Shirts
Hi, all-
Rest essentially scales the solute-solvent interactions, but maintains
the solute-solute interactions. This can be done solely with
Hamiltonian replica exchange, which is in 4.6. It's a bit tricky,
though. We plan on having something that does this automatically in
5.0 or 5.1, but it's
Hi, all-
At this point, any fixes are going to be in the 5.0 version, where the
integrators will be a bit different. If you upload your system files
to redmine.gromacs.org (not just the .mdp), then I will make sure this
gets tested there.
On Fri, Oct 25, 2013 at 10:14 AM, Guillaume Chevrot
Is there a link to the documentation? It's a little difficult to know
exactly what this supposed to be doing. Is it a GUI interface to
gromacs?
In general, it would be great to get these sort of extensions
coordinated with the main gromacs development tree, since otherwise
they would tend to
Hi, Guillaume-
No one can tell you if you did anything wrong if you didn't tell us
what you did! There are literally thousands of combinations of
options in running an NVE simulation, a substantial fraction of which
are guaranteed not to conserve energy.
If you post the files (inputs and
OPLS-AA was generally derived with Monte Carlo, which means that all
bonds were exactly constrained. But read the papers!
On Thu, Oct 10, 2013 at 12:27 PM, Justin Lemkul jalem...@vt.edu wrote:
On 10/10/13 12:17 PM, Martin, Erik W wrote:
Hi, I'm new to both Gromacs and OPLS. I have always
Hmm. This really isn't quite enough information to go on. Can you
file a redmine issue, and include the files used to generate the run
that crashed (.mdp, .gro,. .top), as well as the files that show it
failing (.log)?
http://redmine.gromacs.org/
On Thu, Oct 3, 2013 at 4:32 AM, Jernej Zidar
Sounds like the simulation is blowing up. How soon does it start crashing.
Also, what configurations are you using to start your free energy
simulations at each lambda?
On Wed, Oct 2, 2013 at 10:31 PM, Jernej Zidar jernej.zi...@gmail.com wrote:
Hi all,
I'm trying to determine the free
Just to be clear, is this the expanded ensemble version of the calculation?
On Thu, Sep 26, 2013 at 5:25 PM, Mark Abraham mark.j.abra...@gmail.com wrote:
I found the -multi version of that tutorial a bit temperamental...
Michael Shirts suggested that double precision is more reliable
I thought I had just managed to solve the issue :)
If you look at the soft core parameters, there are two types listed --
one with sc-r-power = 48, and one with sc-r-power = 6. The sc-r-power
are more stable with single precision calculations.
I have changed the files on the website to make the
energy calculation pathways, the
sc-r-power=48 pathway will now be phased out anyway by 5.1.
On Thu, Sep 26, 2013 at 6:37 PM, Michael Shirts mrshi...@gmail.com wrote:
I thought I had just managed to solve the issue :)
If you look at the soft core parameters, there are two types listed --
one
assitance Mark and Michael.
Chris.
-- original message --
Michael Shirts mrshirts at gmail.com
Fri Sep 27 00:41:17 CEST 2013
Previous message: [gmx-users] segfault when running the alchemistry.org
ethanol solvation free energy tutorial
Next message: [gmx-users] segfault when running
There are some starter files here:
http://www.gromacs.org/Documentation/Tutorials/GROMACS_USA_Workshop_and_Conference_2013/An_introduction_to_free_energy_calculations%3a_Michael_Shirts%2c_Session_2A
Which can be used in conjunction with the Alchemistry.org instructions.
But it needs to be
IPS in CHARMM involves additional calculations beyond a simple
homogeneous approximation -- roughly equivalent to PME for dispersion,
though its a bit messier.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2723858/
On Thu, Aug 29, 2013 at 7:56 AM, Justin Lemkul jalem...@vt.edu wrote:
On
The only integrators with stochastic force components are sd and bd.
vrescale has a small stochastic component, but that is for the target
kinetic energy, and is not a random force acting on each particle.
On Thu, Aug 29, 2013 at 3:15 PM, Ali Sinan Saglam
asinansag...@gmail.com wrote:
Hi,
I
Pressure should fluctuate significantly. The estimator for the
pressure that is generally used is very noisy. The question is, do the
pressure averages over, say, 500 ps or 1 ns look about right?
On Sun, Aug 25, 2013 at 5:07 PM, Mark Abraham mark.j.abra...@gmail.com wrote:
On Sun, Aug 25, 2013
Can you clarify - Do you mean that different replicas have different
average pressures?
WITHIN each replica, the +/- 2000 bar changing from step to step is
very common for using an atomic virial like gromacs does.
The AVERAGES of EACH replica should each be the average pressure they
are set as
. Thank you.
On Tue, Aug 20, 2013 at 10:50 AM, Michael Shirts
michael.shi...@virginia.edu wrote:
Dear GROMACS users-
I'd like to remind you all about the 2013 GROMACS USA Workshop and
Conference at the University of Virginia Sept 13th-15th. There are
still registration slots available
,
The 2013 GROMACS USA Workshop and Conference Steering Committee
Michael Shirts (chair)
Angel Garcia
Berk Hess
Yu-Shan Lin
Erik Lindahl
Peter Kasson
--
gmx-users mailing listgmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
* Please search the archive at
http
I don't go beyond 2 fs with either all- bonds or h-bonds. Things like kinetic
energy start being subtly off.
H-bonds has less chance of failing with large numbers of constraints- less
iteration required, especially if bond system cross parallelization boundaries.
If your molecules are 10
This is
Sent from my iPhone
On Aug 14, 2013, at 1:40, Albert mailmd2...@gmail.com wrote:
Does anybody have any idea what's the problem?
I use the tutorial example and I don't know why it doesn't work.
THX
On 08/13/2013 07:19 PM, Albert wrote:
Dear:
I am trying to run plumed with
This is a plumed error, not a gromacs error. Gromacs never handles those
keywords.
Sent from my iPhone
On Aug 14, 2013, at 1:40, Albert mailmd2...@gmail.com wrote:
Does anybody have any idea what's the problem?
I use the tutorial example and I don't know why it doesn't work.
THX
Hi, Sanku-
The way to invoke Hamiltonian replica exchange has changed to be a bit
more flexible. We should go back and make sure that this legacy way
is supported (I thought this invocation was supported, but apparently
it isn't), but what you should be able to do to get it working quickly
is
Hi, all-
That particular redmine was in the case of couple-intramol NOT being
specified. So it may be different.
Can you upload samples files for this test case to redmine.gromacs.org
with specific instructions? Please upload both the 'behaving
correctly' and 'not behaving correctly' versions.
Short answer is anything that has a B state parameter can be included
in in Hamiltonian exchange.
If it's pull code or explicit restraints, it's controlled by restraint lambda.
I went through the manual and couldn't find any definite answers to the
following questions.
First, I wonder if the
I take it that in the position restraint case (not COM-pulling), where
the reference positions are determined by the starting structure instead of
a B-state topology, the reference positions won't be swapped ?
2013/8/3 Michael Shirts-2 [via GROMACS]
ml-node+s5086n5010324...@n6.nabble.com
Hi Dejun-
The basic problem is that for this particular configuration, the
current state is the only state with nonzero weight. Note that the
state with the second highest weight has weight 10^-7. When it tries
to compare weights in single precision, it has a numerical overflow
and fails.
A few
and Conference Steering Committee
Michael Shirts (chair)
Angel Garcia
Berk Hess
Yu-Shan Lin
Erik Lindahl
Peter Kasson
Original announcement:
~~~
We are pleased to announce the 2013 GROMACS USA Workshop and
Conference at the University of Virginia
When I minimize a structure, I can get down to the force max being 0.01
Low-Memory BFGS Minimizer converged to Fmax 0.01 in 6839 steps
Potential Energy = -5.12340607768673e+03
Maximum force = 6.68907856457542e-03 on atom 3029
Norm of force = 2.19978176343026e-03
kJ/nm. However, when
To follow up -- if I try to minimize again using -t, I get the same
low forces as in the minimization in the previous step. So it appears
to be something with what do_nm is doing, not with errors in the
output structure.
On Sat, Jul 20, 2013 at 9:37 AM, Michael Shirts mrshi...@gmail.com wrote
Problem partly addressed. If I run normal modes in -nt 1, then I get
the same force as after the minimization.
I'll file a redmine.
On Sat, Jul 20, 2013 at 9:43 AM, Michael Shirts mrshi...@gmail.com wrote:
To follow up -- if I try to minimize again using -t, I get the same
low forces
Chris, can you post a redmine on this so I can look at the files?
Also, does it crash immediately, or after a while?
On Thu, Jul 18, 2013 at 2:45 PM, Christopher Neale
chris.ne...@mail.utoronto.ca wrote:
Dear Users:
I have a system with water and a drug (54 total atoms; 27 heavy atoms). The
4.5.7 does not support Hamiltonian exchange. It says all properties
are the same, because all the temperatures and pressures are the same
-- it won't switch the umbrellas.
On Wed, Jul 17, 2013 at 3:30 PM, Parisa akhshi...@gmail.com wrote:
Hi Michael,
I think that this is an issue with the
It seems to have something to do with the integrator/pressure-coupling.
that is what I expected based on some preliminary testing earlier.
When
I ran the tutorial on
http://www.alchemistry.org/wiki/GROMACS_4.6_example:_Ethanol_solvation_with_expanded_ensemble,
everything seems fine
OK good!
it's not optimal for data collection as you pointed out. I wonder if it's a
trivial fix that might have been done to add semiiso to MTTK?
Thanks again for your help!
2013/7/17 Michael Shirts mrshi...@gmail.com
It seems to have something to do with the integrator/pressure-coupling
Hi, all-
This not a problem with W-L, but is instead something that is wrong
with a particular combination of mdp options that are not working for
expanded ensemble simulations. W-L can equilibrate to incorrect
distributions because it decreases the weights too fast (more on that
later), but
You need to have different pull parameters at the end states. Right
now, pull-kB1 is not defined in your code, so there is nothing to
interpolate to: it assume pull-kB1 = pull-k1.
Longer scale -- one would want to define reference distances that
change with lambda within the same simulation, but
? (Maybe not in this particular case or in the case of the tutorial
you just posted.) I wonder if we need as many intermediate lambdas as in
the case of replica exchange since the weights compensate the difference in
energy.
Thanks,
Dejun
2013/7/16 Michael Shirts mrshi...@gmail.com
Hi, all
Ah, this is a force field issue -- urey-bradley terms are not
supported free energy calculations. However, since only restraints are
changing, this warning doesn't really need to be there.
It would be relatively simple to put in a check to allow this to work,
but it might take a week or two to
Have you checked out the -ndec option for trjconv? If you have a high
precision format (.trr, or .xtc if they are stored with sufficient
precision) you can print out a .gro file (that gromacs can read) with
higher precision.
Gromacs can read .gro files with increased precisions in the
No.
This is a statistical mechanical issue, not a GROMACS issue. For
interacting systems, entropy is a quantity describing the system as a
whole, and cannot be defined for different parts of the system, at
least not in any way such that the individual components can be added
together.
I'm also
Hi, Sonia-
Gromacs 4.6.2 (some bug fixes vs 4.6.1) with the example files you
point out from Alchemistry.org should work well for expanded ensemble.
David Mobley and I have been validating expanded ensemble and replica
exchange, and the files posted there now are stable for all sizes of
systems,
If you are computing enthaply in the NPT ensemble, P is constant, and
is the applied pressure.
The pressure quantity calculated from the KE and the virial is not
the pressure. It is a quantity that when averaged over time is equal
the pressure. Only the average is meaningful macroscopically.
and Conference Steering Committee
Michael Shirts (chair)
Angel Garcia
Berk Hess
Yu-Shan Lin
Erik Lindahl
Peter Kasson
--
gmx-users mailing listgmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists
or should i be doing U+V*ref_p = H?
More specifically, U + V*ref_p = H
H isn't really meaningful thing. I mean, you can define something
such that H* = H, but that's not really thermodynamics.
example system gives H = -1168 kJ/mol and i find H = -725 kJ/mol either
Interesting. What
An important final point is that you can always see EXACTLY what
grompp is putting into the B state by running gmxdump on the resulting
tpr. It's a LOT of information, but all in text all the interactions
are listed explicitly there.
On Mon, Jun 10, 2013 at 6:20 PM, Justin Lemkul jalem...@vt.edu
It should also be noted (and obvious now that I actually look into it) that
using dispersion correction results in both the latent heat of vapourisation
and density of the alkanes being over estimated (for both Cut-off and
Reaction Field, and by the same amount).
That may not be quite the
that I'd like to sample
all temperature acceptable conformations on the 100 ns trajectory.
But as I understood the chain regime does not compatible with the membrane
protein simulation due to the artifacts arising with MTTK batostat.
James
2013/6/1 Michael Shirts mrshi...@gmail.com
I can't
I can't think of any instance where nose-hoover chains provides an
advantage over nose-hoover in a large system -- all the demonstrations
of superiority are in model systems that are not particularly chaotic.
As the system gets more chaotic, it matters less.
I would go with md, nose-hoover (w/o
Summarizing!
On Fri, May 3, 2013 at 12:31 AM, XAvier Periole x.peri...@rug.nl wrote:
Are confirming that you reproduce the problem with gmx-4.6.1 or simply
summarizing in case we lose track :))
On May 2, 2013, at 23:31, Michael Shirts mrshi...@gmail.com wrote:
So to summarize
Quick check here -- is 4.6 behaving correctly? I actually spent some
time working on REMD in 4.6, and it seems to be behaving correctly in
my hands with temperature and pressure control.
Thanks for any additional info on this!
On Thu, May 2, 2013 at 8:18 AM, Mark Abraham
-4.6.1
On May 2, 2013, at 2:26 PM, Michael Shirts mrshi...@gmail.com wrote:
Quick check here -- is 4.6 behaving correctly? I actually spent some
time working on REMD in 4.6, and it seems to be behaving correctly in
my hands with temperature and pressure control.
Thanks for any additional
turned it off during compilation …
You could try to run on particle decomposition to see if you get a problem …
it should one quite quick.
On May 2, 2013, at 2:36 PM, Michael Shirts mrshi...@gmail.com wrote:
Both. So if 4.6.1 doesn't work, I want to know so we can patch it
before 4.6.2
You have to change atom types. For example:
[ atomtypes ]
;name bond_typemasscharge ptype sigma epsilon
h1h1 0. 0. A 2.47135e-01 6.56888e-02
h1_pert h1 0. 0. A 2.47135e-01 3.56888e-02
; perturbed
Hi, Dejun-
Right now, the vector of lambda parameters is simply vdw, coul, bonded,
restraint, temperature. You can't have, say, 2 different coul vectors or
two different restraint vectors for different restraints. But you can
change any of these components.
You define the vector manually by
. Will try to play around with this later
today.
Best,
Joakim
Date: Thu, 7 Mar 2013 08:55:31 -0500
From: Michael Shirts mrshi...@gmail.com
Subject: Re: [gmx-users] Hamiltonian replica exchange umbrella
sampling with gmx 4.6
To: Discussion list for GROMACS users gmx-users@gromacs.org
Hi
Hi, Joakim-
Hamiltonian exchange only should work if there is a lambda coupling
parameter that defines the potential at each state. You need to
define your pulling potential so that the coupling-lambda parameter
can be used to define the different pulling location centers along
your trajectory.
Ah, now perhaps I see that I misread the question - it could have been
phrased more clearly. If Erik understood it correctly, then the
answer to the question is: It depends on the integrator. The
simulation is not constrained to a particular temperature or pressure
- rather, the dynamics are
The coordinates and velocities that are printed (and that are used to
calculate the properties like energy, virial, etc) are always
consistent with the constraints. The exact order of how things are
done often depends on the integrator. For example, velocity scaling
can be done before or after
Hi, Joakim-
Expanded ensemble is still a bit experimental. I don't immediately
see any problem that jump right out, but if you go to
http://redmine.gromacs.org/ and file a bug report, including giving
example files that cause the problem, I can take a look at it.
On Tue, Feb 5, 2013 at 6:00 AM,
I assume PLUMED will be implemented for Gromacs 4.6, as many PLUMED
developers use Gromacs. Perhaps any PLUMED lurkers on the list can
speak up. . . .
On Wed, Jan 16, 2013 at 9:20 AM, Mark Abraham mark.j.abra...@gmail.com wrote:
The GROMACS team has no plans for that. The usual problem here is
Could you tell me is there any difference of different Tau_t ussage (
inverse friction in case of Stochastic dynamics) for simulation of
water-soluble as well as membrane-proteins ? In the first case I'm
using tau_t 2ps that is lower than internal water friction. In the
second case one part
In the absence of PBC, you simply have an infinite system. In a loose
sense, that may be NVT, but V is infinite, so whether or not you can
consider that to be constant or not is theoretical math above what I know :)
A real molecule in vacuum is usually NVE -- it is not coupled to the
Hi, all-
I would recommend using Parrinellio-Rahman + Nose-Hoover md + at least
until 4.6.
A random-walk drift in the conserved energy is actually what MTTK
gives -- it's not as conserved as, say, energy conservation, it just
has an expectation value of zero drift over time, which means that the
Hi, all-
There are some issues with MTTK + constraints that are being worked
out for 4.6. The good thing is, I have developed some sensitive tests
of the correct volume distribution (see
http://arxiv.org/abs/1208.0910) and the errors in PR are very, very
small. I would recommend using md + PR
It's in review with JCTC right now.
On Thu, Nov 22, 2012 at 2:19 PM, ABEL Stephane 175950
stephane.a...@cea.fr wrote:
Hello,
This is a very nice and interesting work, Michael. Thank you for the efforts
you made in writing this paper. I hope you will publish it.
Best
Stephane
scaling info!
Best,
Michael Shirts
Assistant Professor
Department of Chemical Engineering
University of Virginia
michael.shi...@virginia.edu
(434)-243-1821
--
gmx-users mailing listgmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
* Please search the archive
However, the time value (4 in this example) is limited to 6 digits.
Sounds like this should be increased? There's a pending change to
replica exchange, so this could be added to 4.6 without disrupting the
release timing.
On Wed, Sep 26, 2012 at 11:22 AM, Andreas Zink zinka...@googlemail.com
I've done some extensive testing (paper on testing method in the
works) and vrescale gives a very accurate ensemble very well for NVT.
Parrinello-Rahman and MTTK are the only algorithms that are correct
for NPT. Berendsen barostat is not. Note that there is a bug with
vrescale + md-vv + that is
Hi, David-
Perhaps we can work with you to compare the internal g_bar
implementation with our external BAR and MBAR implementation run from
the .dhdl.xvg data output in 4.6. It would probably be easier to run
these calculations after the optimization updates are added in the
next few days(?),
Short answer -- use a dispersion correction if the force field was
parameterized to include one. Make sure you use the cutoff that the
lipid was parameterized for as well.
Long answer -- neither on nor off is correct for a lipid bilayer. A
dispersion corrections corrects for the fact that you
Good question. Short answer, no -- LINCS doesn't play well with a
velocity verlet based pressure control algorithm.
Long answer: MTTK has ended up not being robust because you need to
solve a self consistent set of equations every timestep using the
pressure estimator, which is extremely noisy,
The implementation of BAR in gromacs is pretty hard for me to follow
because of how everything is stored noncompactly in the histogram. In
4.6, both can be computed from the same dhdl.xvg file, so it might be
easier to track down possible bugs.
On Fri, Jun 29, 2012 at 2:24 PM, David van der
It only matters for running on Folding@Home. For other users of
gromacs, it doesn't do anything.
On Tue, Jun 26, 2012 at 3:50 PM, Bao Kai paeanb...@gmail.com wrote:
Hi, all,
I am wondering what the --enable-fahcore option of configure means. I
got the explanation from configure --help of
There's a fair amount of interest for more general support for Monte
Carlo methods in GROMACS 5.0. However, there is no any current
support for configuration bias Monte Carlo (or any other kind of MC)
currently in the code.
On Fri, Jun 1, 2012 at 12:10 PM, Benjamin Haley bha...@purdue.edu wrote:
Gromacs already supports replica exchange -- what particularly are you
implementing?
Equilibration of pressure is always a good idea -- even if you are
running NVT simulations, you want to get them to be at the equilibrium
volume for your system and temperature choice, which will require
4.6 will include Hamiltonian replia exchange functionality built into
the MPI version.
Currently the description of the error is very vague -- if you can
write up what exactly the numbers are, and what they should be, with
files that exactly replicate the error, then I can take a look. But
Hi, Sabine-
If you can go to http://redmine.gromacs.org/projects/gromacs/issues
and file a bug report (including attaching files), I can look at it.
If that's ends up not working well, you can send me the files off of
the list, but it's usually better to have things in the redmine system
so
Sabine, thanks for filing it in redmine! Having a record helps a lot.
Can you also attach all your input files to the redmine filing? It
can only really be debugged if the input files you used are included.
Best,
Michael
On Fri, Mar 23, 2012 at 9:11 AM, Justin A. Lemkul jalem...@vt.edu wrote:
It also depends whether or not you use constant pressure, in which case it
makes sense to increase the time to long enough to let the volume relax.
I still do not understand why people do NVT REMD, because it makes all but
one replica have a pressure that is not the ambient pressure.
If all
Is anyone aware of any benchmark analysis about the implementation of the
amber99sb force field (or any of its variants: 99sb-ildn, 99sb-nmr) in
GROMACS and AMBER. I am interested to know in what extend the energies
correlate and if the results agree with experimental data.
Whether the
Dispcorr is only for homogeneous liquids. It should not be used for
membranes.
More precisely -- the dispersion correction is an analytical
correction to both the pressure and energy that is rigorously correct
in the limit of the radial distribution function g(r)=1 outside the
van der Waals
FEP is a poorly defined term. It can either mean 1) making small
changes 'alchemical' changes in the molecules and computing the free
energies by any method (BAR, TI, etc), or 2) it can mean specifically
computing the free energies by exponentially averaging the potential
energy differences.
So BAR is only
referring to the mathematical code used to calculate the overall free
energy for the FEP, correct?
Yes. The information required is the same, with the exception that
exponential averaging requires energy differences from only one state,
whereas BAR requires energy differences
I don't think this is a question about new free energy code -- I think
this is asking about the fact if you can do a free energy calculation
by specifying the A and B variables in the topology, instead of using
the MDP coupl-moltype arguments. This is actually the way free energy
calculations
work, but it's not quite ready yet.
Michael Shirts
Assistant Professor
Department of Chemical Engineering
University of Virginia
michael.shi...@virginia.edu
(434)-243-1821
--
gmx-users mailing listgmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
Hamiltonian replica exchange is planned for 4.6, and is being beta
tested by some users.
On Tue, Aug 16, 2011 at 2:39 PM, Sanku M msank...@yahoo.com wrote:
Hi,
I was wondering whether hamiltonian replica exchange simulation has been
implemented in latest version of gromacs . Or, is there
quite well -- you might get some drift after 1-2 ns,
but not much. I'm working on developing suggested PME parameters
right now for highly quantitative work, but it's not quite ready yet.
Michael Shirts
Assistant Professor
Department of Chemical Engineering
University of Virginia
I am doing free energy calculation in Gromacs and want to get an error
estimate of my results. Is it possible to compute the autocorrelation time
of dVpot/dlambda in Gromacs using a certain length of trajectory such as 10
ns? Thanks a lot
The amount of simulation time required to compute the
Hi, all-
Have you tried running
constraints = hbonds?
That might eliminate some of the constraint issues. Much less likely
for LINCS to break or have DD issues if only the hbonds are
constrained. 2 fs is not that big a deal for the heteroatom bonds.
Best,
Michael
On Thu, Mar 10, 2011 at
One other thing I would point out is that the solvation free energy is
dependent on concentration. you will get a different result with 4
polymer chains vs 3 vs 3, etc. Make sure you understand the
dependence. Also, the free energy will depend on the polymer chain
length.
Polymer and finite
OK, that is what I was trying to figure out -- is the problem
reproducible on both GPU and CPU. Now, you havent answered the direct
question, if the energies are the same for at least the first 5 steps
are so -- without that knowledge, then here might be different errors
occurring in GPU vs CPU.
Do you get the same effects if you run a normal simulation on CPU and not
GPU? That information would be critical for properly diagnosing what's
going on. If it's not GPU-specific, in all likelihood whatever you're doing
is incorrect somewhere along the way.
Folllowing up on this -- if it's
A few questions:
1) What force field are you using?
2) do you get the same answers with and without GPU acceleration?
3) How long does it take for secondary structure to disappear? 100's
of ps? 10's of ns?
On Mon, Jan 17, 2011 at 8:45 AM, K. Singhal k.sing...@uva.nl wrote:
Hi
I have a
Things to check: were the random numbers initializing the velocities
the same? Looking at the log files for each simulation should be able
to tell you (look at the initialized kinetic energy at the beginning).
Some options choose a different random number automatically for each
simulation.
On
correct there? ;)
Best,
Michael Shirts
Assistant Professor
Department of Chemical Engineering
University of Virginia
michael.shi...@virginia.edu
(434)-243-1821
--
gmx-users mailing listgmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
Please search the archive
method (python implementation from
Michael shirts ), I have to rerun the simulations, in order to get
potential energies for each configuration at every lambda point.
That's correct. Very soon (i.e., the code is in the CVS) it will be
possible to generate these energy differences during single runs
Can you be more specific in the question of what property you want to
compute for which molecule or molecules? Linear Interaction energy
approaches are not always very efficient for coulomb, and is
definitely not exact for the LJ term. So LIE might not be what you
want to be doing.
Michael
1 - 100 of 111 matches
Mail list logo
