Communication: A multiscale Bayesian inference approach to analyzing subdiffusion in particle trajectories

[jezcope]

Paper title: A multiscale Bayesian inference approach to analyzing subdiffusion in particle trajectories

Description:

Using a Bayesian inference approach to evaluate asymptotic behaviour by sampling finite simulated trajectories.

Participants

• Jez Cope @jezcope

Resources

• paper URL: https://aip.scitation.org/doi/10.1063/1.4965881

• data URL: ftp://ftp.aip.org/epaps/journ_chem_phys/E-JCPSA6-145-039640

• code URL: ftp://ftp.aip.org/epaps/journ_chem_phys/E-JCPSA6-145-039640

• reproduction/replication URL:

[jezcope]

Theoretically, the paper is packaged up using the ActivePapers format, which should allow the whole paper to be easily reproduced. The code, data and paper are packaged into a HDF5 file, which can be extracted using the ActivePapers toolkit.

Sadly I haven't managed to get it working on Windows yet! 😖 Downloading HDFView requires me to create an account and login, which is frustrating since I don't feel I should have to have to do that for open source software.

So I have the code and data but can't access it because the tooling is annoying!

[khinsen]

Paper author's remark:

Sorry for inconvenience! HDFView used to be a one-click download, but the HDF group recently locked up some of their code, apparently to please their sponsors if I remember the story correctly.

A good alternative to HDFView is HDF Compass, written in Python with all source code on GitHub. It used to be an official HDF Group project, but it is not in the product list any more. That also means that binaries are no longer available... The quality of HDF5 tooling is definitely not improving!

On the plus side, you have made a contribution by identifying a new kind of reproducibility obstacle: software dependencies becoming more difficult to access over time.