Nebspline.pl produces exts.dat which containing extrema found along the MEP. In my case, the plot of energy vs. reaction coordinate, after visual inspection, contains only one maximum. But exts.dat indicates that there should be four extrema. My question is, how reliable is this output. Spline interpolation can cause non-physical oscillations by their very nature. Are there any relevant publications you can point me towards?
-Takis
nebspline.pl and exts.dat
Moderator: moderators
The NEB cubic spline is better behaved than what is normally referred to as a cubic spline, because the slope at each image is known from the calculated force along the minimum energy path. This means that for each spline segment, there are 4 quantities (energy and force at the endpoints), which uniquely determine the cubic parameters. Also, the cubic parameters in one segment do not affect those of the next. In a normal spline, it is the few constraints and the influence of one segment on the next can cause the wild oscillations you refer to. Neither of these problems exists with the NEB because of the constrained slopes.
That said, there are often extra extrema. The data in the spline.dat is generated from the same parameters from which the extrema are found, so any extra extrema will show up in a plot of the spline.dat data. Sometime the extra extrema are very valuable, and the script can show you a segment in which the forces at the ends of the segment are not consistent with the energy change between the endpoints. This can (typically) indicate an intermediate minima along the band, which may not be apparent in the energy values.
Most often, though, the extra extrema are caused by small fluctuations (around zero) in the force at the endpoints. If the force at the initial point, for example, is slightly negative when projected along the minimum energy pathway, there will be a shallow minimum and an extrema at the start of the path. These kind of extrema, however, are very easy to identify and ignore.
The values of the spline parameters, and how they are derived from the forces and energies can be found in the perl script or in the following publication. http://theory.cm.utexas.edu/henkelman/p ... 00_269.pdf
That said, there are often extra extrema. The data in the spline.dat is generated from the same parameters from which the extrema are found, so any extra extrema will show up in a plot of the spline.dat data. Sometime the extra extrema are very valuable, and the script can show you a segment in which the forces at the ends of the segment are not consistent with the energy change between the endpoints. This can (typically) indicate an intermediate minima along the band, which may not be apparent in the energy values.
Most often, though, the extra extrema are caused by small fluctuations (around zero) in the force at the endpoints. If the force at the initial point, for example, is slightly negative when projected along the minimum energy pathway, there will be a shallow minimum and an extrema at the start of the path. These kind of extrema, however, are very easy to identify and ignore.
The values of the spline parameters, and how they are derived from the forces and energies can be found in the perl script or in the following publication. http://theory.cm.utexas.edu/henkelman/p ... 00_269.pdf