ohno/Antique.jl

Programming paradigm & style

Closed this issue · 3 comments

ohno commented

This is a continuation of previous discussions #3 .

ohno commented

@hyrodium

Thank you for your comments and discussion. I am planning to rewrite codes as multi-dispatch style before adding new models. Please see a notebook on gist. I have considered multi-dispatch style with reference to your past comment.

Multi-dispatch coding allows to adapt Symbolics.jl, Measurements.jl, etc. I am not going to specify a type of the struct to adapt these packages.

Although the internal implementation a little loses its similarity to formulae, the usage is similar to formulas. For example, the energies of $\mathrm{H}$ and $\mathrm{He}^+$ are occasionally wrriten as $E(\mathrm{H})$ and $E(\mathrm{He}^+)$. In the multi-dispatch version, we can write it like this:

H = HydrogenAtom(Z=1)
E(H)
He⁺ = HydrogenAtom(Z=2)
E(He⁺)

There are some undecided matters.

  • Should multi-dispatch be applied to Hermitian polynomials H(x) as well?
  • Should codes be into a module?
    • Modules are useful for document build.
    • The module name and the type name conflict. What should I name the module?
    • Not necessarily for multi-dispatch Hermitian polynomials.
  • etc...

The multi-dispatch version is scheduled to be released as v0.2.0. I will continue to support antique() as a deprecated function and display an alert.

  • 👀1
ohno commented

@hyrodium
I rewrote all of codes, documents and tests for multi-dispatch style. There was a small problem but it passed tests. These will be merged and released as v.0.2.0. Thank you again for your advices.

hyrodium commented

Great job! Looking forward to seeing you in JuliaTokyo 😄

  • 😄1