If a sampled vloop distance is far from everything ever encountered during sampling before, ks is near zero (e.g. 2*10^-230) and the energy value can become unexpectedly high.
In the next step, it will drop dramatically again, because ks now does not have such a high influence.
The kernel density estimations from the initial 100 values loaded into the file seems to be rather sharp around certain values.
Bernhard10
commented
8 years ago
If a sampled vloop distance is far from everything ever encountered during sampling before, ks is near zero (e.g. 2*10^-230) and the energy value can become unexpectedly high.
In the next step, it will drop dramatically again, because ks now does not have such a high influence.
The kernel density estimations from the initial 100 values loaded into the file seems to be rather sharp around certain values.