Dynamical Noise Correction
Introduction
In Nagel et al., 2019 it was emphasized, that dimensionality reduction methods, such as PCA or tICA, suffer artefacts on the barriers. Those frames should be identified as noise and assigned dynamically. It should be used together with \(\mathtt{coring}\) which is the more powerful tool.
Execution
clustering noise -s state_file
-c cmin
-b basename
-o output
--cores output_cores
--concat-nframes NFRAMES
--concat-limits limit_file
-v
Parameters
Input Parameters
| Parameter | Description |
|---|---|
| \(\mathtt{\mbox{-}s :}\) | The name(path) of the clusterd state trajectory file. |
| \(\mathtt{\mbox{-}c :}\) | The population threshold (in percent) below which an isolated cluster is assigned as noise. (Default: \(0.1\)). |
| \(\mathtt{\mbox{-}b :}\) | The basename of the input cluster files used to determine isolated clusters. (Default: \(\mathtt{clust}\)). |
| \(\mathtt{\mbox{--}concat\mbox{-}nframes:}\) | The number of frames per (equally sized) sub-trajectories for concatenated trajectory files. |
| \(\mathtt{\mbox{--}concat\mbox{-}limits:}\) | The name(path) to the limit file. It should be a single column file with the length of each trajectory. for a concatenated trajectory of three chunks of sizes 100, 50 and 300 frames: '100 50 300'. |
Output Parameters
| Parameter | Description |
|---|---|
| \(\mathtt{\mbox{-}o}\) | Filename for the output file of the noise corrected microstates. |
| \(\mathtt{\mbox{--}cores}\) | Filename where the microstate trajectory is stored with all noise frames denoted by \(-1\). |
Miscellaneous Parameters
| Parameter | Description |
|---|---|
| \(\mathtt{\mbox{-}v}\) | Verbose mode with some output. |
Detailed Description
On the barriers we are facing a poor sampling. Therefore, also due to projection artefacts, frames can be assigned to the wrong state. To avoid this, we simply reassign those frames dynamically instead of geometrically. This was proposed in Nagel et al., 2019. We use the following definition of noise:
- All geometrically isolated clusters with population lower than \(\mathtt{c_\text{min}}\) are assigned as noise.
- They are assigned dynamically (in the sense of coring) to the previously visited state.
Where \(\mathtt{c_\text{min}}\) is given in percent of the total population. This has an impact of
- \(\approx5\%\) of total frames
- improves the Markovianity in some systems