Species Type: Clarifications on Molecule Syntax

[s9105947]

The current version of the species type extension demands for molecule syntax:
"Use standard chemical notation, e.g.: H2O."

What is "standard chemical notation" specifically?

Notable cases:

• Molecules with charge, e.g. OH-
• complexes, e.g. [NiCN]+
• organic compounds, e.g. C2O5OH, CH3COOH, maybe even CH3-COOH, CH3-CH2-CH3
• pedantic there are more complex formulae which are "standard" too, but can't easily be translated to text, e.g. structural formulae
• Are parenthesis allowed? e.g. SO2(OH)2 (admittedly non-standard)
  • Note that parenthesis need to be balanced, which increases parsing complexity: With paranthesis, this can't be parsed by regex.
  • Semantics: Are (H2)O and H2O considered equal?

This is a bottomless hole, so I'd like to suggest as a minimal version:

• ("An implementation must support at least the following notation. Additionally, more specific notation may be supported:")
• A molecule is given as a sum formula, e.g. semantically only the amount per atom is extracted.
• Parenthesis, brackets, (and braces) of any form are forbidden.
• If an atom occurs twice its quantities are added. ("An implementation may derive additional information from this, e.g. anorganic groups like OOH.")
• Only the total charge of a molecule can be specified by appending "^" *DIGIT ("+" / "-"), e.g. HCO^-, SO4^2-
  • If the digit is omitted it is assumed to be one, i.e. HCO^1- and HCO^- are equivalent.
• Note: A total quantity of a molecule can not be given, i.e. 2H^+ is invalid.

Note: Depending on if an isotope is also an atom maybe use "atom or isotope" repectively.
Note: Potentially replace the charge separator character.

[ax3l]

The answer for this is similar to #260 (comment):

We don't have a use case with a committed chemistry code yet and leave this open to the first one(s) that would like to adopt openPMD. I like your suggestion, but would not standardize it until we have a concrete use case/adopter.

Note for the future reader: @s9105947 works on PIConGPU, which is an electromagnetic particle-in-cel code in laser-plasma physics (keV to MeV range physics, while molecules are eV physics and not covered by em. PIC.)

[s9105947]

In this case I would suggest to remove the paragraph about molecules alltogether:

"The base standard does not encode molecules. However, an implementation MAY support molecules with an implementation-defined syntax."

In my opinion the current "use standard chemical notation" is not sufficiently clear.