The simulated tempering options haven't been as well tested as the
hamiltonian expanded ensemble version. The weights SHOULD be showing up in
the column that says -nan, but clearly they aren't. If you file a redmine
issue, I may be able to take a look, but it might take a while to address.
On We
Hamilton replica exchange is implemented in GROMACS (unless it's been
broken in some way). See
http://www.alchemistry.org/wiki/GROMACS_4.6_example:_Direct_ethanol_solvation_free_energy
for
an older tutorial, but it should be relatively straightforward to adjust to
current call patterns.
On Tue, O
w carrying out expanded
> ensemble at NVT and converting back to NPT on the mailing list. Again I
> appreciate your advice in this area.
>
> Best wishes,
>
> Gregory
>
>
> On 5/8/2019 12:01 PM, Michael Shirts wrote:
>
> Yeah, this is an unfortunately place in the code wher
If you have "unprotected" electrostatic sites (i.e. with nonzero repulsive
terms directly on top of the charge), then there will always be some
configurations with essentially infinitely negative energy. For vanilla
MD, these are essentially impossible to reach kinetically at any reasonable
temper
Yeah, this is an unfortunately place in the code where not all combinations
work - very long story. Hopefully this will be working better in 2020.
What I would recommend is, if possible, performing the expanded ensemble
simulation at NVT. Everything should work fine there (paper coming out
hopef
Does anyone have experience running GROMACS on GPU's on Oak Ridge National
Labs Titan or Summit machines, especially parallelization over multiple
GPUs? I'm looking at applying for allocations there, and am interested in
experiences that people have had. We're probably mostly looking at systems
in
As free energies get larger, then the error gets less accurate. So if it is
reporting 18.51 +/- 2.95 for one of the intervals, that likely suggest
there is very little overlap in that area.
A total free energy difference of in -6.12 +/- 3.19 for the
transformation indicates that the result is no
Hi, Jacob-
Unfortunately, that is correct. We had though we'd have functionality that
could support this combination in 2018 (a Monte Carlo barostat and/or
expanded ensemble in alchemical space working with leapfrog), but those did
not make it in. The support for MTTK with constraints was removed
This is not the right place to ask about MBAR. You could pose the question
on https://github.com/choderalab/pymbar if you have specific questions
about how to implement something there.
https://github.com/MobleyLab/alchemical-analysis is a wrapper for MBAR that
can process gromacs files.
See:
htt
You don't have a well-posed scientific question yet. Come back when you've
thought about the hypotheses you want to test, and have written out what
you are planning to do. Try to do the calculations yourself, and come back
if they are not working. This is a help discussion forum for people who
a
More precisely, the van der Waals energy depends on whether you turn off
the charges at the same time or after. It's a path-dependent quantity, not
a state function.
On Wed, Sep 20, 2017 at 12:33 PM, David van der Spoel
wrote:
> On 20/09/17 19:51, Nikhil Maroli wrote:
>
>> Dear all,
>>
>> How t
Significantly more information is be needed to understand what happened.
* What does the logfile say that was output?
* What command are you using to run on multiple nodes?
* What is the .mdp file?
* How many nodes are you running on?
* What version of the program?
And so forth.
On Mon, Aug 21,
I'm interested carrying out a long simulation of a thin, narrow protein.
Clearly, the volume-minimizing way to do this would be to put it in a thin,
narrow box of solvent. The issue with this is that it could rotate around
and interact with itself. if the orientation of the protein changed
relativ
I'd take a look at the paper describing the tool:
http://dx.doi.org/10.1007/s10822-015-9840-9.
On Tue, Apr 18, 2017 at 12:07 PM, Alex wrote:
> Any idea please !
> Thanks
> Alex
>
> -- Forwarded message --
> From: Alex
> Date: Sun, Apr 16, 2017 at 11:29 AM
> Subject: Error bar i
wrote:
> Dear Michael.
>
> Thanks for the reply. I have already used this python code however I would
> like to calculate myself by using the potential energies.
> Because I need to reweigh the free energy by using exponential
> re-weighting technique.
> > On 10 Feb 2017,
https://github.com/MobleyLab/alchemical-analysis
Takes gromacs dhdl.xvg output and calculate free energies by many different
methods, including BAR, MBAR and Zwanzig. See
http://www.alchemistry.org/wiki/Main_Page for more information.
On Fri, Feb 10, 2017 at 5:26 AM, gozde ergin wrote:
> Dear
. Competitive
candidates will have a track record of scientific success in graduate study
with a strong publication record.
Apply at
https://cu.taleo.net/careersection/2/jobdetail.ftl?job=05979&lang=en&sns_id=mailto#.V5DS-mZVNX8.mailto
Best,
~~~~
Michael Shirts
Associate P
See the following paper
(http://pubs.acs.org/doi/abs/10.1021/ct400109a) for some examples of
diffusion calculations in GROMACS.
The finite size correction for diffusion coefficients is something
different than the long range energy corrections. A discussion of this
(I can't vouch for all the detai
Hi, all-
I wanted to check quickly. I noticed a difference in domain
decomposition when I ran with free energy on and off (using the
coupl-mol functionality), with a relatively large molecule (8 residue
peptide) being decoupled. Domain decomposition failed for the free
energy on 4 cores, and loo
See http://pubs.acs.org/doi/abs/10.1021/acs.jctc.6b00397 for an
example of a PSCP in GROMACS (although it was from crystal to
crystal). It takes series of different .mdps to do it.
On Fri, Jul 22, 2016 at 5:34 AM, Justin Lemkul wrote:
>
>
> On 7/21/16 11:20 PM, kar...@shell.com wrote:
>>
>> Hell
md integrator I do not face
> with this problem.
> Any idea?
>> On 18 Jul 2015, at 04:13, Michael Shirts wrote:
>>
>>> Has problem been solved about molecules drifting in md-vv integrator?
>>
>> We think so, but some additional tests are running to address any
Try ParmEd.
http://parmed.github.io/ParmEd/html/index.html
On Sat, May 28, 2016 at 7:57 AM, Sourav Ray wrote:
> Hello
>
> Currently I have a combination of:
>
> 1. .pdb file (four copies of a monomer)
> 2. .itp file defining the topology of one monomer (applied on all four with
> the help of the
If you are doing solvation in water, I would suggest not bothering
with expanded ensemble. I would just use standard molecular dynamics
at a range of lambda values. We have some information that it can be
useful with binding problems, but with solvation alone, you are
probably just fine running a
ile snippets that fills me
>> with any hope that it would be noticeably faster.
>>
>> We have a pile of infrastructure built and in the works that will lead to
>> being able to offer similarly optimized free-energy kernels, but they won't
>> see the light of day
Hi, Chris: I'm pretty sure that it's because the nonbonded free
energies are much slower than the standard free energies. You state:
> I took a look at gmxlib/nonbonded/nb_free_energy.c in v.5.1.2, but I was
> unable to find a function called "gmx_waste_time_here()" and beyond that I
> was out
Correct, currently a WCA decomposition has not been implemented.
I'm not sure if I understand this comment, though:
> For this reason, I only plan to calculate E_vdw_dispersive and E_elec for my
> system.
E_vdw_repulsive is a gigantic contribution to the solvation free
energy of the protein. W
In theory, https://github.com/GromPy/GromPy can do it
(http://www.ncbi.nlm.nih.gov/pubmed/22370965), but it's old and I've
never tried it. Not in the main code currently.
On Wed, Feb 10, 2016 at 11:38 AM, Irem Altan wrote:
> Hi,
>
> I have two questions:
>
> - Is it possible to do μVT simulation
Also, can you be more precise about the inconsistency in the outputs
you are seeing (with numerical output from GROMACS and analysis code).
On Thu, Feb 4, 2016 at 6:17 AM, Michael Shirts wrote:
> Hi, Dries-
>
> Can you print out what the corresponding section of mdout.mdp look like?
>
fourierspacing = 0.08
>
> ; EWALD/PME/PPPM parameters
> pme_order= 6
> ewald_rtol = 1e-06
> epsilon_surface= 0
>
> Thanks in advance,
>
> Dries
>
> On 4 February 2016 at 00:56, Michael Shirts wrote:
>
>> Hi, D
if this is
happening.
> From: Van Rompaey Dries
Date: Wednesday, February 3, 2016 at 12:34 PM
To: Michael Shirts
Subject: PME settings in free energy calculations
Dear prof. Shirts,
I'm currently trying to figure out the PME settings for a relative
free binding energy simulatio
It introduce how to calculate this
> angle. No code is provided with this
> paper. Can they share this code? I use a smooth surface.
>
>
> Best wishes!
>
>
>
>
>
> ------ Original --
> From: "Michael Shirts";;
> Date:
See:
http://pubs.acs.org/doi/abs/10.1021/acs.langmuir.5b00842
For a study using GROMACS. They code can be downloaded (though it
might not be trivial to use).
On Thu, Jan 21, 2016 at 6:32 PM, HZJ <1047334...@qq.com> wrote:
> Dear GMX users,
>
>
> Recently, I want to analyze the contact angle of
You should file an issue on the alchemical-analysis issue tracker.
https://github.com/MobleyLab/alchemical-analysis
On Tue, Dec 15, 2015 at 8:35 PM, Nathan K Houtz wrote:
> Hello,
>
>
> This request is related to python programs used for analyzing gromacs
> outputs, not gromacs itself. If there
> Yes, I just try to show to Alexey Shvetsov or some gromas users who want
> to do check the conversion.
The issue is that the hardest and most time consuming part is not the
conversion itself, but the testing. Without the testing, the
conversion isn't that useful. I appreciate the work you've d
IF it's NVT, then you should just be able to add harmonic potentials
using the position restraints (not constraints) -- the harmonic
restraints can be controlled by lambda. You should only need to alter
the charge and VDW interactions -- no need to adjust the bonds.
We have been doing some work o
Great! It's important to have all data validated carefully. Have you
put together a set of tests, with coverage of all the atom types and
interaction types, showing that the molecules give the same energies
in both GROMACS and AMBER? Validating conversions is quite important
so that people can tr
Was this not useful?
> Most of this should still be applicable for 5.0.4.
> http://www.alchemistry.org/wiki/GROMACS_4.6_example:_Direct_ethanol_solvation_free_energy
On Mon, Nov 30, 2015 at 8:51 AM, Sanja Zivanovic
wrote:
> Dear users,
>
> Could you please send me some example of inputs for Hami
Most of this should still be applicable for 5.0.4.
http://www.alchemistry.org/wiki/GROMACS_4.6_example:_Direct_ethanol_solvation_free_energy
On Wed, Nov 25, 2015 at 6:23 AM, Sanja Zivanovic
wrote:
> Could you please tell me how I can perform Hamiltonian replica exchange
> simulations in Gromacs
lations are unchanged. Is
> the calculation of dh/dl
re
> ally that expensive, or is this abnormal?
>
> Thank you for the help as always. Regards,
> Nathan H.
>
> - Original Message -
> From: "Michael Shirts"
> To: "Discussion list for GROMACS us
For a pure fluid, G = N \mu. And \mu = (dG/dN)_(T,P). So you only
need to change one molecule to ideal gas to get the change in free
energy. The free energy of transfer of water from liquid to gas is
indeed the free energy of solvation of one water molecule in bath of
water. So there's a reason
> Is there any way to use maestro cms file for md simulations in gromacs.
If it's just proteins / water / ions, then saving as a pdb and using
pdb2gmx is probably the best.
If it contains other things, you can try:
https://github.com/shirtsgroup/InterMol
This is still in alpha, so no guarantee
ParmEd (http://parmed.github.io/ParmEd/html/index.html) is an
interesting new tool that can help automate such conversions so they
don't need to be done as manually in the future.
On Wed, Sep 30, 2015 at 1:37 PM, Simon Dürr wrote:
> Hi,
>
> we had the pleasure to convert the *.mol2 and *.frcmod f
hings. Shyno wasn't very specific, but his samples of output
> include the string "Repl" which suggests the latter, whose data is not
> checkpointed.
>
> Mark
>
> On Thu, Sep 24, 2015 at 11:19 PM Michael Shirts wrote:
>
>> static int do_cpt_df_hist(XD
http://parmed.github.io/ParmEd/html/index.html
https://code.google.com/p/acpype/
On Thu, Sep 24, 2015 at 5:14 PM, Gustavo Avelar Molina
wrote:
> Hi,
>
> I would like to know if there is a way to convert Amber structures and
> topologies to the GROMACS format.
>
> Thank you very much for your tim
ist->Tij_empirical, list); break;
On Thu, Sep 24, 2015 at 5:06 PM, Mark Abraham wrote:
> Hi,
>
> Some things are accumulated there, yes. But not the transition history.
>
> Mark
>
> On Thu, Sep 24, 2015 at 11:04 PM Michael Shirts wrote:
>
>> I take it back, I think
I take it back, I think it is actually accumulated. See (in checkpoint.cpp):
static int do_cpt_df_hist(XDR *xd, int fflags, df_history_t *dfhist, FILE *list)
On Thu, Sep 24, 2015 at 4:59 PM, Michael Shirts wrote:
> Apologies: I missed the part where it was stated that the simulation
&g
27;ll get truncated upon restart-with-append.
>
> Mark
>
> On Thu, Sep 24, 2015 at 10:42 PM Michael Shirts wrote:
>
>> > Now, to determine the acceptance rate for the whole 10ns run I should
>> > average these numbers as shown below?
>>
>> No, the matri
> Now, to determine the acceptance rate for the whole 10ns run I should
> average these numbers as shown below?
No, the matrix at 10ns is the accumulated matrix. You should use the
last matrix.
--
Gromacs Users mailing list
* Please search the archive at
http://www.gromacs.org/Support/Mailing_
Maybe the ligand actually is leaving the binding site and moving
around, and that's what your simulation is telling you! Hard to say
with the posted information.
On Tue, Sep 15, 2015 at 11:24 AM, Homa rooz wrote:
> Dear Justin
> I have followed your advice but it didn't solve my problem.
> Ligan
> The exchange attempt forces replicas to synchronize, and this forces the
> whole simulation system to pay the cost of any localized slow-down (e.g.
> network traffic congestion, internal simulation load imbalance from
> diffusion, even throttling of core clocks from heating effects) at each
> syn
> I am new to doing replica exchange simulations.
> I am trying to understand the empirical transition matrix printed in the
> log files. I tried exchange intervals 2 ps and 0.2 ps. However, later I
> read gromacs recommends exchange interval not to be under 1ps.
I don't know that there are firm r
Chemical potential in the fluid would be almost exactly the sanme to a
solvation calculation -- as long as you can count on it to be
metastable enough to not crystalize / phase separate during the time
of the simulation.
Chemical potential in the crystal is . . . harder. Check out the
pseudo-sup
That could be documented better. It is all of the changes in the
molecule that are not explicitly occurring in their own lambda
variable. For example, in setup 2, then all the changes have separate
dhdl terms, so there is nothing left to be in dVremain/dl. Note that
in setup 1) dVremain/dl is ju
Adjusting the soft core is a gigantic pain. I wouldn't recommend it,
and it's likely not necesssary. What settings are you using now? Note
that the manual describes how it is defined.
Can you post the alchemical-analysis output for N/N_k? There's an
analysis quirk we are looking at where the co
Adding an integrator for your own work is hard but doable, especially
if you don't need it parallelized. Look at the do_update_X routines
for examples. It will be harder for 5th/6th order since you probably
need values of the forces for times in the past, which are not
automatically stored in gr
The identifiers are just strings, so as long as they are unique, there
shouldn't be a problem (I can't remember offhand if capitalization is
respected or not, though).
I'd name them something other than opls_XXX, though, if you are adding
your own, just to avoid confusion of you share the files wi
> Has problem been solved about molecules drifting in md-vv integrator?
We think so, but some additional tests are running to address any
other lingering integrator issues. The fix is in 5.1 beta, but not in
the 4.6 branch yet (it's a pull request, but not yet merged).
> Also can you please chec
I put in 1-1-48, and although somewhat more efficient (20-30%), it has
a tendency to have floating point errors in single precision. I'll
probably take it out sometime in the next few versions, as the benefit
is not really worth the hassle. If you are off by a lot, it's almost
certainly not going
I would probably do this not using the couple-moltype functionality,
since you want to change TWO molecules. GROMACS thinks they are part
of the same molecule, and thus they should not move too far away (I'm
guessing). Instead, you can do this through the definition of A and B
types (see documenta
At some point, a COM removal step was omitted for md-vv. A change has been
proposed in in gerrit for a while (See https://gerrit.gromacs.org/#/c/4649/
and https://gerrit.gromacs.org/#/c/4467/ ), but one of the developers asked
to have the source of the problem traced to understand better what happ
Expanded ensemble doesn't require any special number of nodes (states are
visited sequentially, not in parallel) so the rules are the same as for any
other free energy simulation.
On Thu, Jun 18, 2015 at 4:46 AM, Andreas Mecklenfeld <
a.mecklenf...@tu-braunschweig.de> wrote:
> Hi,
>
> I'm new to
The 'compressibility' in the pressure coupling is not actually the
compressibility of the fluid; it's another way to write the mass of the
fictitious piston. It affects the inertia of the piston and the period of
the oscillations of the box, but won't affect the magnitude of fluctuations
of the bo
file obtained from NPT as initial configuration for NVT, since I
> wanted to have correct volume). The second way gives that big difference in
> Cv and Cp.
> I run a long simulations in all cases, but why both ways produce such
> difference?
>
>
>
> Best regards
>
>
example for Benzene the
> difference in Cv and Cp is 25%, and here Ii get similar ( my molecule is
> also carbon ring). But still I can answer why two both ways does not give
> same Cv-Cp.
> I will follow your suggestion.
>
>
>
> Best regards
>
>
> On Tue, Ju
0 J/mol.K difference in Cv and Cp.
>
> Mean that fluctuation play big role. Which way of checking I can rely?
>
>
> Best regards
>
>
> On Tue, May 26, 2015 at 3:38 PM, Michael Shirts
> wrote:
>
> > By definition (more fundamental that fluctuation formulas)
> &
ar Michael,
> >>
> >> I use Parrinello-Rahman for barostat and v-rescale for thermostat.
> >>
> >> Sorry, could you explain more the second paragraph please? I did not get
> >> the method. What I checked so far is checking if gromacs correctly gives
&g
Are you running with the Berendsen thermostat or barostat? The gromacs
g_energy functions for heat capacity use the fluctuation formula, and the
fluctuations with both of these algorithms are wrong (as should be printed
in the log file warning message). Make sure you use ensemble-preserving
thermo
state.
On Tue, May 19, 2015 at 11:58 AM, Hannes Loeffler <
hannes.loeff...@stfc.ac.uk> wrote:
> On Tue, 19 May 2015 11:37:11 -0400
> Michael Shirts wrote:
>
> > >
> > > somewhat off-topic but I wonder why in your free energy protocol you
> > > only vary the
>
>
> As Michael said it is better to leave most things untouched and just change
> VDW and LJ - at least this is what I have been reading so far.
>
Well, I didn't quite say this. Changes in the bonded terms do not entirely
cancel. Changes in masses do 100% cancel because of the seperability of
>
> somewhat off-topic but I wonder why in your free energy protocol you
> only vary the vdW and electrostatic lambdas. What about the others?
> Your mutation also transforms bonded terms and masses.
>
Minor point - if you are taking the difference between two mutations (say,
with and without a
For #1, see: http://pubs.acs.org/doi/abs/10.1021/ct400109a
For #2, I think you'll have to figure that out for yourself.
On Wed, Mar 11, 2015 at 5:42 AM, sujithkakkat .
wrote:
> Dear all,
>
> I am simulating a system of small hydrophobic solutes ( CH4) dissolved
> in water. Simulations are pe
Did you look at the energy distributions at each temperature? Overlap
decreases with system size; the larger the system size gets, the closer
temperatures need to be to get any overlap. If you are simulating 30K or
more MARTINI particles, those sort of ratios may not be wrong at all.
On Mon, Nov
I didn't write the g_bar script, so I can't answer all of the questions
about it.
Equation 3 in that paper is a general equation true for any weight function
alpha. It is not Bennett's acceptance ratio. Equation 4 is the one that's
actually Bennett's equations, which chooses an alpha that gives a
Those conserved quantities are indeed included in the reported
information. For thermostats/barostats where there is such a quantity,
it's included in the .edr file.
On Fri, Oct 24, 2014 at 4:23 AM, Paolo Franz wrote:
> Hi,
> I am wondering how to test the conserved quantity in an NPT simulatio
ggested in
> http://web.mit.edu/~ripper/www/research/efficient_md_integrators.pdf
>
> This is why it is important for me to know where each step starts and
> finishes.
>
> 2014-10-15 15:52 GMT+02:00 Michael Shirts :
>
> > Because the 'start' of the vv integrator step
Because the 'start' of the vv integrator step is halfway through the loop.
This is a byproduct of 1) putting leapfrog and velocity verlet in the same
loop and 2) minimizing communication and output. It is not as elegant as
it should be. There are efforts to clean this up, but it's a lot of
reorga
> I guess, if I pick a coupling constant that is just small enough to keep the
energy conserved, I would get a NVT simulation that is as close as a NVE
simulation as possible.
> Is this correct?
Yes, but then at that point the thermostat isn't actually thermostatting.
The Bussi comment is merely
les is
> correct
> > (as in fluctuation-dissipation theorem). If we use thermostats, the
> > thermostats will correct the energy drop/raise caused by numerical error
> > (which acts as a fluctuating source/sink that may not even follow the
> > normal distribution), wit
> The point was that if the energy leaves the system, then the system
is never in equilibrium. Not vice versa, please.
This is incorrect logic. f there are errors in the integrator, then
F=gradient of the potential is not true. If that is not true, energy is not
conserved, and energy can leave or
mulations"
http://pubs.acs.org/doi/abs/10.1021/ct400109a
Best,
~~~~
Michael Shirts
Associate Professor
Department of Chemical Engineering
University of Virginia
michael.shi...@virginia.edu
(434)-243-1821
On Fri, Oct 10, 2014 at 7:40 PM, Antonio Baptista
wrote:
> On Fri, 10 Oct 2014, Mark Abr
9 p.m. GMT tonight (September
2nd) until 11:59 p.m. Pacific Daylight Time tomorrow, Wednesday, September
3rd.
Best,
~~~~~~~~
Michael Shirts
Associate Professor
Department of Chemical Engineering
University of Virginia
--
Gromacs Users mailing list
* Please search the archive at
http://www.g
com> wrote:
> Hey Mark,
> In ref to part 3 of the survey, I am surprised that it does not capture
> Africa.
>
>
> On Tue, Aug 26, 2014 at 1:39 PM, Mark Abraham
> wrote:
>
> > Dear Gromacs users/developers,
> >
> > Gromacs developers Profs. Michael
> However, if I were to linearly decouple the Coulombic interactions, does
it mean I slowly reduce the cut-off for > Coulombic interaction to 0
angstroms?
No, it is reducing the contribution of q_i q_j energy terms in the atoms
being coupled to zero.
On Sun, Jul 13, 2014 at 10:59 PM, Yip Yew Mun
A nonisotropic system will not necessarily end up with exactly the applied
pressure -- things like surface tension and stresses/strains end up getting
mixed in and throwing things off.
On Mon, Jun 23, 2014 at 3:32 PM, Chetan Mahajan
wrote:
> Thanks.
>
> Average was taken over 1000 ps. Error est
Try this talk posted on Alchemistry.org by Gabe Rocklin who has carefully
looked at this problem.
http://www.alchemistry.org/wiki/images/5/50/GabrielRocklin2014-PeriodicityArtifacts.pdf
And the full paper:
http://scitation.aip.org/content/aip/journal/jcp/139/18/10.1063/1.4826261
On Sun, Jun 1
Vmd can convert trajectories, as can other tools. Google is your friend!
Sent from my iPhone
> On May 24, 2014, at 15:13, Albert wrote:
>
> this plug-in looks interesting.
>
> However, as far as I saw from the tutorial that even we run simulation from
> Gromacs, this plugin can only import NA
Look for papers by D. E. Shaw and Gianni di Fabritiis. They have done
this, but it generally takes at least microseconds or milliseconds to
converge.
On Sat, May 24, 2014 at 6:24 AM, Nidhi Katyal wrote:
> Hi all,
>
> I would like to ask if unbiased MD in nanoseconds time scale be used to
> find
There are many ways for free energy calculations to go wrong. Look at
http://www.alchemistry.org for some discussions.
One thing I noticed (I didn't have a chance to look at everything) -
you only talk about one simulation. When doing classical simulations,
then you can't just mutate atoms, becaus
Berendsen has the right average volume, but not the right distribution of
volumes.
http://pubs.acs.org/doi/abs/10.1021/ct300688p
Did you equilibrate berendsen until the volume stabilized and THEN switch
to parrinello-rahman for production? That's the suggested thing to do.
On Fri, May 16, 2014
-estimate in Gromacs, and I'm wondering if it's connected to this. At
> > least if I can rule out a trivial error in my input files, there can be
> > confidence to look deeper.
> >
> > Thanks again for your help.
> > With best regards,
> >
> > James
> >
> >
What are the energies with the same initial configuration? Perhaps LAMMPS
is doing something under the hood like shifting the cutoff which Gromacs
won't do unless you ask. Knowing the step=0 energy will help identify if it
is a difference in dynamics or in the energy function.
Simulating water wit
The error estimate is a rough guide, but is not always 100% accurate when
one is making.
I'd consider looking at:
http://pubs.acs.org/doi/abs/10.1021/ct300688p
With accompanying code here:
https://github.com/shirtsgroup/checkensemble
This might help ascertain a bit more how the ensemble that i
Mean of the three calculations is 4.35, standard error is 0.19. So this
seems statistically consistent. If you want to get more consistent
results, you'll need to run longer until the estimated statistical
uncertainty is lower. Also, the estimated uncertainty is often an
underestimate (BAR often
Putting both velocity Verlet and leapfrog Verlet both in Gromacs turns
out to be non-trivial for the bookkeeping. The easiest way to do this
was split the velocity Verlet updates.
Also, the additional computational cost of two half steps for
velocities is trivial compared to the cost of the force
Setting the compressibility in gromacs actually is just as way adjust the
stiffness of the barostat.
Berendsen thermostat guarantees a Cp that is too low. Try v-rescale or
nose-Hoover.
See http://pubs.acs.org/doi/abs/10.1021/ct300688p
For more info.
Sent from my iPhone
> On Mar 16, 2014, at
Although a recent version have dealt with it, its implementation may
> require some time.
> Anyway, it's fine if you do not care about dynamics.
>
> Francis
>
>
> On Wed, Feb 26, 2014 at 6:39 AM, Michael Shirts wrote:
>
>> Your best bet is probably something like the
Your best bet is probably something like the PLUMED plugin to do
metadynamics. aMD raises up the potential if it is below a given
cutoff; metadynamics puts in bumps in the potential where you have
visited. There are definitely subtle differences, but the general
effects are similar.
On Tue, Feb
Feb 25, 2014 at 3:57 PM, Michael Shirts wrote:
>> No full solutions to this problem -- I'll write a few notes.
>>
>> * With large systems run for relatively small amounts of time, the
>> ensemble could be statistically indistinguishble from the true
>> distribution
No full solutions to this problem -- I'll write a few notes.
* With large systems run for relatively small amounts of time, the
ensemble could be statistically indistinguishble from the true
distribution even if the averages don't line up correctly. See:
http://pubs.acs.org/doi/abs/10.1021/ct3006
cous friction by heat bath without coupling the
> velocity profile to heat bath? Also, why to remove the velocity
> profile of the kinetic energy?
>
> On Sat, Feb 22, 2014 at 9:08 PM, Marcelo Vanean wrote:
>> Michael Shirts, in this article is not discussed about the periodic
>&g
1 - 100 of 120 matches
Mail list logo