lmin - compute the local minimum of the potential energy
lmin [ --help ] [ --version ] [ --verbose num ]
[ --vdw14 num ] [ --eel14 num ] [ --diel-lin num ]
[ --diel-const num ] [ --cutoff num ] [ --numgrad ]
[ --numgrad ] [ --begin-coords file.crd ]
[ --outfile file.crd ] [ --accuracy num ]
[ --seed num ] file.top
The lmin program computes the local minimum of the potential energy function
corresponding to a given starting coordinates (or, when --begin-coords is not given, from a randomly choosen ones). This program needs one file -
the file.top describing the molecule (potential energy function).
Prints out help message and exits.
Prints out the version of the program and exits.
- --verbose number
Sets the verbosity level. The levels have the following meaning:
Be really quiet. Do not print anything. Well, with this option this program
is really unusable.
Print the final value of the potential energy only
Print the following message:
Minimum is <number>
Where the <number> is the value of energy in the local minimum.
Print the final value of the potential energy as well as the number of
iteration needed to compute it.
Prints detailed report - the potential energy at each iteration is printed
as well as the number of this iteration.
- --vdw14 number
Sets the scaling parameter for van der Waals part of the 1-4 interactions.
The default value is 2.0.
- --eel14 number
Sets the scaling parameter for the electrostatic part of the 1-4
interactions. The default value is 1.2.
- --diel-lin number
Sets the linear part of dielectric funtion. The default value s 0.0 which
means to simulate the force field without the solvent. Set this to 4.0 to
simulate water. This option is used together with the next one.
- --diel-const number
Sets the constant part of dielectric function. The default value is 1.0
which means to simulate the force field without the solvent. Set this to
4.0 to simulate water. This option is used together with the previous one.
- --cutoff number
Sets the cutoff parameter for the 1991 force field. The cutoff means that
the nonbonded interactions for atoms from different residues that have
distance greater than the cutoff are not evaluated. This is used in Amber
1991 force field for reaching some speedup. I don't think it speeds up
Opal, but I have included this for compatibility. The default is not to use
cutoff, i.e. +infinity.
- --begin-coords filename.crd
The coordinates from which the minimization is started. If this option is
not given, lmin generated the starting coordinates randomly.
- --outfile filename.crd
The coordinates of the local minimum are written into this file at the end
of the program.
- --seed number
The random number generator seed. The random number generator is set to use
this seed for generating of random coordinates. This option is usable only
when --begin-coords is not given.
- --accuracy number
The accuracy of the local minimizing procedure. The default is 1.0e-5. This
number controls when the minimizing stops.
The gradient values computed numerically instead of explicitly computed
gradient. This can probably used for testing as it is slower than default
The lmin can be compiled to use two local minimization procedures: The e04dgf from the NAG Fortran library (see the www.nag.co.uk for details) or (by
default) the MINPACK dvmlm function. Minpack is distributed with Opal but has not the same license.
See the vmlm/README for details. The vmlm library calls the various vector and matrix operation from the Basic Linear
Algebra Subprograms (blas). In Opal there is only a reference implementation of blas. Please check
www.netlib.org/blas if there is a platform-specific blas for your hardware. The platform-specific version can be magnitude faster
(altough lmin spends in blas routines only a small fraction of its running time).
Opal itself is an open source(tm) software. It can be freely distributed and used under the terms of GNU General Public License. The minimization procedures distributed with Opal in directories
vmlm and blas are not freely usable. See their license files for more information.
Opal was created as part of master thesis of Jan ``Yenya'' Kasprzak, who is
the current maintainer of Opal. Some parts of Opal (most notably the lmin program and the function for numerical computing of the gradient) as well
as many bug-fixes and suggestions for improvement were contributed by Petr
Probably there are some. Send the bugfixes (prefered form is a
diff -u output) and bug-reports to Jan ``Yenya'' Kasprzak (email@example.com).
A not-exactly-bug: I am looking for a good local minimization routine
written in C so that Opal can be compiled without the Fortran compiler.
opal_singlept(1), opal(1), opal_lmin_t(1)
http://www.nag.co.uk/ - The Numerical Algorithms Group, Ltd.
http://www.netlib.org/blas - The Basic Linear Algebra Subprograms.