COSMO-SAC sigma range hard-coded

[thomvet]

I noticed that the sigma profile range appearing in COSMO-SAC models is hard-coded within Clapeyron, see for example here:

Clapeyron.jl/src/models/Activity/COSMOSAC/COSMOSAC10.jl

Line 184 in 8d9ebe9

σ = -0.025:0.001:0.025

The good thing is that [-0.025;0.025] is the commonly used range (with spacings at 0.001 intervals). However, there are relatively simple (and "non-extreme") compounds that fall outside this range; for example tartaric acid (-0.0250415), arginine (zwitterion form, +0.0256457), lysine (zwitterion form, +0.0251358). The QC calculations were performed with the newest version of GAMESS in this case.

If one considers more charged things, such as ionic liquids, I am sure they would fall further outside the hard-coded range.

A potential way to allow flexibility would be to include the sigma-values in the sigma-profile files in the database, i.e., having a second vector with the sigma values and then the pSigma values as before:
https://github.com/ClapeyronThermo/Clapeyron.jl/blob/master/database/Activity/COSMOSAC/COSMOSAC02_like.csv

And then some minor code changes in the model files dealing with the sigma profiles.

I am happy to attempt this, but since the database files would need a different formatting and content, it would likely be a breaking change for users that already use these files? Or maybe it's doable to subtype the existing COSMO-SAC models?

Any thoughts?

[longemen3000]

For this I would recommend marking the sigma profile points as optional, so, if they are not provided, they will default to the current values. In this way, users don't need to update their database when working with sigma profiles with the default spacing.

Adding new data on the databases is not considered a breaking change. If anything, this is more of a bugfix.

[thomvet]

Yes! That's way better. I'll give it a try and come up with a PR, but won't get to it before next week.