Michele Simionato

@mmpagani says that we should use a concave hull. Here are two references: https://gist.github.com/dwyerk/10561690 https://towardsdatascience.com/the-concave-hull-c649795c0f0f

> The fragility and vulnerability functions for liquefaction usually accept only one intensity measure, and not two (i.e. gmv_VS, gmv_LS). Thus, it is necessary for the user to specify whether...

> the geometric mean (ex. numpy.sqrt(gmv_VS^2+gmv_LS^2)) @VSilva here I assume that there was a misprint and that the correct formula is `numpy.sqrt(gmv_VS^2*gmv_LS^2)` otherwise it would not be the geometric mean,...

I am not sure if we can close this given that the TREQ project ended

LGTM

Here are then things to do on the importer side: 1. change the line ```python al = list(readinput.Exposure.read(nrml_file, asset_nodes=True)) ``` with ```python al = asset.Exposure.read([nrml_file]).assets ``` 2. change in _import_asset...

Is this a problem, @ptormene ? If not, we can close the issue.

This will come hot again due to https://github.com/gem/oq-engine/issues/6668: h5py 3.1 returns bytes instead of strings, so it may break the plugin even more.

We already switched to h5py 3.1 without big issues; let's keep the status quo.

The difficult thing here is to make sure that the rupture ID is reproducible, i.e. running twice the same calculation should produce the same IDs, even on different machines. Perhaps...