Bert de Groot
bgroot at gwdg.de
Mon Aug 9 11:19:38 CEST 2004
Chris Shaw wrote:
> Hi everyone
> Has anyone used make_edi to generate the input for mdrun? I see that a
> question was posted earlier today. There are various options in which
> the trajectory can be driven by eigenvectors, which ones do I use
> linfix, linacc, radfix, radacc? Then there is the steps and cycles.
> Furthermore, i cant get hold of the papers decribed in the information
> bit of make_edi ( i think just because they are old) and i cant find
> anything in the manual either. So, can anyone give me some pointers to
> what is what, and possible setting for the slope, stepsize bits
The papers can partly be downloaded as pdf from our website:
and some are also part of my thesis which is also available online there.
What type of constraint you want to use depends on what you want to do and
which type of system you have. If you have a (set of) particular eigenvector(s)
of interest, then you may want to explore them by linear constraints. This means
that for those eigenvectors you specify a direction (and optionally a stepsize)
and during the simulation the system will be constrained towards that direction.
You can either specify a fixed stepsize (nm per MD step), using "linfix", or you
can give the system more freedom and do a simulation in which you only accept
MD steps if they happen in the specified direction (linacc). Only if the system
moves in the reverse direction will it be constraied (or put back) to its original
Alternatively, you can also specify a range of vectors along which a volume
should be sampled (ie without specifying an individual eigenvector or direction),
a so-called radius expansion. Similar to the linear constrains, you can choose
between a fixed radius increment per step (radfix) or an acceptance-style
constraint (radacc). The 'slope' criterion is only relevant for the "radacc"
constraint, and is used to measure the (spontaneous) increase of the radius
of the sampled hypersphere. If this increase or slope becomes smaller than a
predefined value (this is interpreted as reaching the boundaries of a local
minimum of the potential energy landsacpe) a new expansion "cycle" will be started
with the current position as radius. "steps" therefore always correspond to MD steps,
"cycles" correspond to such expansion cycles (of which also the maximum length
can be specified).
Depending on system type and size you probably want to test some setups and
figure out the optimal settings for your system by trial and error.
Dr. Bert de Groot
Max Planck Institute for Biophysical Chemistry
Computational biomolecular dynamics group
Am Fassberg 11
37077 Goettingen, Germany
tel: +49-551-2012308, fax: +49-551-2012302
email: bgroot at gwdg.de
More information about the gromacs.org_gmx-users