Chemspecies update

[tjjarvinen]

This is update to ChemicalSpecies.

• Change info field to name field and make it represent 4 characters
• Add function atom_name to access the new name field if given, else return atomic_symbol
• Added masses for isotopes
• Added symbol X to valid input. Gives weightless particle with atomic number zero

The idea with atom_name is to represent PDB file format defines atom name columns (13-16). While element_symbol represents element columns (77-78)

1. element_symbol always gives the symbol of the element
2. atomic_symbol adds isotope information if given
3. atom_name defaults to atomic_symbol, but overwrites it with atom_name if species has that information.

This works now

julia> ChemicalSpecies(:C)
C

julia> ChemicalSpecies(:C12)
C12

julia> ChemicalSpecies(:C; atom_name=:MyC)
C

julia> ChemicalSpecies(:C13; atom_name=:MyC)
C13

julia> ChemicalSpecies(:X)
X

Comparisons

The basic idea is that if no isotope or name information is given, then that ChemicalSpecies works as a general atom with that atomic number.

atom_name on the other hand is used to distinguish atoms with same type of atoms.

# true
ChemicalSpecies(:C13) == ChemicalSpecies(:C)

# false
ChemicalSpecies(:C12) == ChemicalSpecies(:C13)

# true
ChemicalSpecies(:C; atom_name=:MyC) == ChemicalSpecies(:C)

# true
ChemicalSpecies(:C12; atom_name=:MyC) == ChemicalSpecies(:C12)

# false
ChemicalSpecies(:C; atom_name=:MyC) == ChemicalSpecies(:C12)

# true
ChemicalSpecies(:C12; atom_name=:MyC) == ChemicalSpecies(:C)

# true
ChemicalSpecies(:C; atom_name=:MyC) == ChemicalSpecies(:C12; atom_name=:MyC)

So basically, you can create a potential using ChemicalSpecies(:C). Users can then mark their systems with isotopes or atomic names, for some additional information and the potential that uses "the general carbon" works with them.

masses

julia> mass(ChemicalSpecies(:C))
12.011 u

julia> mass(ChemicalSpecies(:C12))
12.0 u

julia> mass(ChemicalSpecies(:C13))
13.003354835340001 u

julia> mass(ChemicalSpecies(:C; atom_name=:MyC))
12.011 u

julia> mass(ChemicalSpecies(:C13; atom_name=:MyC))
13.003354835340001 u

julia> mass(ChemicalSpecies(:X))
0.0 u

@mfherbst @cortner

edit. changed atomic_name to atom_name

[rashidrafeek]

just a suggestion: atomic_name seems to imply the name is a property of the atom itself instead of an identifier (which seems to be the case here). Wouldn't just atom_name be better which also matches the PDB defintion?

[tjjarvinen]

This is now up for review and merge.

just a suggestion: atomic_name seems to imply the name is a property of the atom itself instead of an identifier (which seems to be the case here). Wouldn't just atom_name be better which also matches the PDB defintion?

I agree with this one and made the change.

This is of course up for a debate, so feel free to state your opinions.

[cortner]

I read through and at least for the use-cases I can envision at the moment I am happy. For me this is ok to merge for now.

There is one small item worth thinking about : I find it odd to have a large table of isotope data in a julia source file rather than a JSON or similar. But this is just a matter of preference and I don't feel very strongly about it.

[tjjarvinen]

Fixed a bug where ChemicalSpecies(:U238) and similar did not work correctly.

There is one small item worth thinking about : I find it odd to have a large table of isotope data in a julia source file rather than a JSON or similar. But this is just a matter of preference and I don't feel very strongly about it.

The reasoning here was that having JSON or similar would add more dependencies. So, having it in julia source allows minimizing dependencies.

[Luthaf]

Sorry for the drive-by comment, but why are you limiting the name to 4 characters? This is the limit for the obsolete PDB format, but for example the mmcif format that replaced it does not have any length limitation: https://mmcif.wwpdb.org/dictionaries/mmcif_pdbx_v40.dic/Items/_chem_comp_atom.atom_id.html

[tjjarvinen]

In order to have a bits type you need to have fixed amount of bits. This leads to a limitation on character length, which for now is 4*8bit char. We could of course increase it later on (there are several ways to do this). But the main design goal was to have it as bits type, which e.g., allows moving atom types to GPU etc.

[Luthaf]

Fair enough!

[rkurchin]

This looks good to me also – should we link this with #115 so it will close when this is merged?

[mfherbst]

Very nice, thanks for the extension.

[mfherbst]

@assert somewhere on n_neutrons and/or elsewhere to avoid this gets triggered without notice ?

[mfherbst]

Use 3-argument get with default

More broadly: is this a good idea, I mean typos and small missmatches lead to a zero here, which could lead to subtle bugs, no? Same with the other lookups

[tjjarvinen]

Yes the subtle bugs are possible here.

We could also have it error, if symbol is not recognized and is not :X. This would probably be the safest option.

[tjjarvinen]

I added a check for symbols and now symbols that are not recognized throw an ArgumentError.

[mfherbst]

Quite a few newlines

[tjjarvinen]

Everything should now be ok with this PR.

[mfherbst]

Lgtm

[mfherbst]

Unless there are objections I'll merge this tomorrow.