jarvist / quante.jl

Quantum-chemistry methods in Julia. Based on Rick Muller's PyQuante2

Quante.jl

An open-source implementation of quantum-chemistry methods in the Julia programming language. These codes are based on Rick Muller's pyquante2 (https://github.com/rpmuller/pyquante2/).

Julia is a high-level dynamic programming language that is capable of fast execution. In particular it can generate performant code targetting SIMD, parallelisation and GPU execution, from high level constructions.

ToDo

• Debug H2O / LiF RHF energy error
• Add CCSD 'explicit forloops' - see Psi4Numpy https://github.com/psi4/psi4numpy/blob/master/Coupled-Cluster/CCSD.py
• Add CCSD 'Tensor contractions' - via https://github.com/Jutho/TensorOperations.jl
• Play with C_∞ / D_∞ symmetry for diatomic CCSD interaction energies

References

• Attila Szabo and Neil S. Ostlund - Modern Quantum Chemistry (https://www.amazon.co.uk/Modern-Quantum-Chemistry-Introduction-Electronic/dp/0486691861)
• David B. Cook - Handbook of Computational Quantum Chemistry (https://www.amazon.co.uk/Handbook-Computational-Quantum-Chemistry-Dover/dp/0486443078)
• PyQuante2 (https://github.com/rpmuller/pyquante2)

quante.jl's Issues

Hartree-Fock as function minimization

Rather than jump straight to CI, maybe it's more interesting to explore lower-hanging fruit: The possibility of framing the Hartree-Fock approximation not as a matrix eigenvalue problem, but as a constrained minimization problem. Then we could potentially use autodifferentiation to speed up the optimization.

'Warning: Package Quante does not have Test in its dependencies' when using with julia 1.6

When using with Julia 1.6:

julia> using Quante
[ Info: Precompiling Quante [cffb2240-8e2b-11e9-2857-d3cff796b0cb]
Warning: Package Quante does not have Test in its dependencies:
.......

the issue is too small for a pull request ...

@distributed does not work for multi-core 1 and 2 electron integrals

The new @distributed parallel macro does not seem to actually work when running on multiple cores.
The error appears that it may just be a matter of understanding the new Package manager sufficiently so that the correct environment is instantiated on each process.

[Fri28Sep-16:45]jarvist@nmscde005688:~/REPOS/Quante.jl/test/ 
> julia-1.0.0 -p auto runtests.jl 
ERROR: LoadError: On worker 2:
ArgumentError: Package Quante [top-level] is required but does not seem to be installed:
 - Run `Pkg.instantiate()` to install all recorded dependencies.

_require at ./loading.jl:923
require at ./loading.jl:852
#2 at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.0/Distributed/src/Distributed.jl:77
#116 at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.0/Distributed/src/process_messages.jl:276
run_work_thunk at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.0/Distributed/src/process_messages.jl:56
run_work_thunk at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.0/Distributed/src/process_messages.jl:65
#102 at ./task.jl:259

It may just be a matter of installing the package properly via the package manager to begin with!

First steps towards adding full configuration interaction

Required for CI:-

• Transform 2e integrals from atomic basis to molecular orbitals (N^5 process is done naively).
• Matrix elements between Slater determinants. Either via Slater-Condon rules (see Szabo and Ostlund), or the full Slater-Condon-Harris rules
• Solve the new secular / matrix equation. (Nb: for any serious calculation, need to exploit sparsity + possibly locality, to avoid the N! scaling)

References:-
c.f. (old) pyquante 'CI.py' - this is a pretty good reference for the singles at least http://pyquante.sourceforge.net/
Chapter 4 in Szabo and Ostlund
Sherrill notes http://vergil.chemistry.gatech.edu/notes/ci.pdf

Travis CI

Might be handy to have, in addition to appveyor

Add `Project.toml` and `Manifest.toml`

It seems like specifying dependencies in a REQUIRE file is outdated, with the modern approach being to have two files, Project.toml and Manifest.toml (see https://docs.julialang.org/en/v1/stdlib/Pkg/).

Things seem to work for now, but this should probably be changed sooner or later -- there might also be other changes that need to be made that I'm not aware of from a quick reading of the Pkg documentation.

CCSD (+ others) via Tensor Contraction

Intent is to use
-> https://github.com/Jutho/TensorOperations.jl

An interesting approximate algorithm here:
https://arxiv.org/abs/1708.02674

The granddaddy is the TCE, as used in NWCHEM:
http://www.csc.lsu.edu/~gb/TCE/
https://doi.org/10.1080/00268970500275780

• it's a DSL written for tensor operations - so something that should be relatively easy to implement in Julia, and then quite fast with the . dot splatting.

CCSD via explicit for-loops

see Psi4Numpy https://github.com/psi4/psi4numpy/blob/master/Coupled-Cluster/CCSD.py

port to julia 1.0?

Hi!

I had recently embarked on the task of porting pyquante.jl to julia 1.0, and I had it working, except for running into the energy error bug which it looks like you managed to solve... I like what you've done with the package structure, so would you accept a PR that implements the changes needed to get the code running on Julia 1.0?