Pure Julia Atoms

[cortner]

CC @jameskermode

I thought, we'd collect some ideas and action points to discuss how to and the consequences of a pure Julia Atoms object.

TODO

• pure Julia neighbourlist (EDIT)

[cortner]

Q1: store positions, momenta as fields, or in a dictionary Dict{Any, Vector{JVecF}} for storing only arrays of 3-vectors, so they can be extracted without type instability.

Which other arrays need to be stored?

• masses
• atomic numbers
• species symbols

[cortner]

Storing the cell as a matrix creates ambiguities. Are the columns or the rows the cell vectors? I would like to switch storing the cell as a list of vectors,

# either 
Tuple{JVec, JVec, JVec}
# or 
Vector{JVec}

Any concerns?

[cortner]

I won't do this, to keep compatibility with ASE. But instead introduced cell_vecs to get the vectors instead of the matrix.

[gabor1]

can I plead for full words as names.. cell_vectors is just fine!

…

-- Gábor

On 8 March 2018 at 09:42:59, Christoph Ortner ***@***.***) wrote: I won't do this, to keep compatibility with ASE. But instead introduced `cell_vecs` to get the vectors instead of the matrix. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: #63 (comment)

[jameskermode]

Normally I would agree, but the function that goes from matrix to vector is already called vecs()

[cortner]

Normally I would agree, but the function that goes from matrix to vector is already called vecs()

that was the reason, but I'll think about it.

[gabor1]

That's fine then. Vecs breaks up the matrix into a bunch of vectors?

…

-- Gábor

On 9 Mar 2018, at 22:31, Christoph Ortner ***@***.***> wrote: Normally I would agree, but the function that goes from matrix to vector is already called vecs() that was the reason, but I'll think about it. — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

[cortner]

C^T = [ c1 c2 c3 ] where ci are of type SVector{T}