## Reweighting and enhanced sampling techniques
If some data get generated at distribution $P(x)$, and another set get generated on $P'(x)$, we wanna extract information of $P(x)$ from $P'(x)$ samples. This is the goal of enhanced sampling techniques, and may be achieved by reweighting.
>[!Note]
>By "enhanced sampling techniques", one assigns higher probability to those hard-to-reach regions, to get more samples, and recover real landscape based on the probability profile.
Some enhanced sampling techniques:
- [Metadynamics](https://en.wikipedia.org/wiki/Metadynamics)
- [Umbrella sampling](https://en.wikipedia.org/wiki/Umbrella_sampling)
- [Replica exchange molecular dynamics](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6484850/)
- [Accelerated molecular dynamics](https://ambermd.org/tutorials/advanced/tutorial22/)
>[!Info]
>See similar chapters: [[Data representation]], [[Landmarks]], [[Dimensionality reduction]].