Table of Contents
This section will discuss a sequence of molecular dynamics simulations using Ti-MEAM poteital of omega structure. The configuration file, hcpomega-y.xml, and the potential file MEAMPOT.Ti.308a are provided by Prof. Wilkins at Ohio State University.
The main input file of this section is hcpomega-y.evn.xml.
<?xml version="1.0" encoding="utf-8"?>
<simulation>
<keyword> Ti 360 atom hcp-omega interface along Y MEAM </keyword>
<project id="hcpomega-y.evn.1000K.grid.dt=2" series="0">
<application name="ohmms" role="md" version="0.2.3.1"/>
</project>
<random parallel="true" seed = "-1"/>
<process>
<particlesets>
<particleset src="hcpomega-y.xml" />
</particlesets>
<mdsim>
<particlelayout method="grid" scale="1.1"/>
<potential name="MEAM">
<filename>MEAMPOT.Ti.308a</filename>
</potential>
<ensemble name="evn" engine="EVNVelocityVerlet">
<parameter name="temperature" condition="K">1000</parameter>
<parameter name="timestep" condition="fsec">2</parameter>
<parameter name="update" >10</parameter>
</ensemble>
<loop blocks="1" steps="1000">
<record name="ensemble" stride="1" />
<record name="particle" format="ascii" stride="100">
<attrib name="position" datatype="xmolArray"></attrib>
</record>
</loop>
</mdsim>
</process>
</simulation>
In order to perform const energy, volume and number (EVN) simulations, the name of ensemble should be set to evn. The default engine for an EVN simulation is EVNVelocityVerlet. As the name implies, it uses velocity-verlet algorithms for the updates. The value of the engine is the class name of the code.
<ensemble name="evn" engine="EVNVelocityVerlet">
<parameter name="temperature" condition="K">1000</parameter>
<parameter name="timestep" condition="fsec">2</parameter>
<parameter name="update" >10</parameter>
</ensemble>
To use Verlet algorithm, set engine="ENVVerlet"
The nearest-neighbor engine is set by <particlelayout method="grid" scale="1.1"/>. It says to use grid method and use 1.1 of the physical cutoff radius set by a model Hamiltonian to evaluate the list. The choice depends on the problem, e.g., temperature, time step and update of the ensemble.
The parameters of ensemble should be self explanatory but <parameter name="update">10</parameter>. This concerns the algorithm ohmms uses for efficient evaluations of neighbor-list for any potential with a finite interaction radius. The choice depends on the temperature and other parameters of particlelayout. The proper value of update parameter can be determined by running a short simulation: use the maximum number to obtain a good total energy conservation, which is subject to the user.
EVN simulations are are important in testing the code and setting the optimal parameters of other simulations. Especially, choosing the optimal parameters for the particlelayout and update by running few short and small and medium problems is important.
The first thing to look at is the energy conservation Figure. The kinetic energy Eke is shifted by the average total energy Etot to show the energy conservation and the relations between the kinetic and potential energies. When the parameters are set too aggressively, the total energy starts growing and eventually, the simulations become meaningless. The parameters optimized by EVN runs can be used by other ensemble simulations.
