Update dataset

[gomezzz]

Description

• Added ignoring local notebooks (my_notebook folder for that)
• Reduced dataset radius lower cutoff to 175km
• Added BSTAR , ID to dataset
• Small fix in results analysis NB to load new HDF5 files (Properties killed it)
• Improved radius computation
• Cleaned up input data generation notebook (it was getting large)

For radius in particular we no longer just assume static values for space-track RCS data. I added sampling to make it more diverse.

Basically, I computed mean and std for the different categories RCS SMALL, MEDIUM and LARGE (< 0.1, 0.1 < RCS < 1.0, 1.0 <) and sampled from gaussian like that (minimum radius being 0.5cm,RCS=0.1, RCS=1.0 for the respective categories).

Values for that are

Celestrak RCS distribution that I used for that is

Distribution of the sampled values looks like this

It's rather tame especially on the large ones but I think it's good to be a bit conservative here, what do you think? (the maximum randomly sampled object has a radius of 7.05m)

The sampled radii look like this

(looks so different due to logscale)

@FG-TUM Should we put necessary I/O changes in C++ into same PR or create an issues and have another one?

Related Pull Requests

N / A

Resolved Issues

• [x] fixes #104
• [x] fixes #105
• [x] fixes #97

How Has This Been Tested?

Not yet.

[FG-TUM]

Should we put necessary I/O changes in C++ into same PR or create an issues and have another one?

New PR for the necessary C++ changes, if I make them. Otherwise, you would have to review your own PR. -> #109