Hartree-Fock-in-CPP

=============================================================================

Three years ago, there was a Hartree Fock program written in C (https://github.com/Walter-Feng/Hartree-Fock). During that time, I only learned C, so I needed to write all the structures myself - tons of pointers, stupid recursion, and so on. The code was super inefficient, buggy, maybe with some memory leaks.

Things changed. I learned C++ and software engineering practice, read some good code, wrote a lot more code in different languages. Externally, C is no longer popular among undergraduates. Therefore, reviewing the original C-style Hartree-Fock code, I thought instantly "hey this code need a complete reconstruction", so making it a lot easier for newcomers to understand the math, run the code, and get a taste of what good code is.

"Well... but you could have used Python which is more friendly"

No. I refuse to use python for this project. Learn some static programming!

=============================================================================

Back to the topic, this is a Hartree-Fock program that aims to show how Hartree-Fock really works in every process, but in full C++ standard. (Hopefully) it is well documented with good variable names so that you can read the code without much effort and understand what is going on. A good tex document that illustrates the mathematics and C++ features is (probably) on the way.

As a comparison, the features in this program are:

1. The programming style is completely in C++ - OOP, templates, but no more pointers.

2. Tons of dependencies, including Boost (you need to install yourself)

   Armadillo (same as above)

   sunqm/libcint (but only using its rys quadrature roots generator)

   catchorg/Catch2

   Taywee/args

   fmtlib/fmt

   nlohmann/json

3. It is (in principle) able to use all the bases that have been published in Basis Set Exchange (https://www.basissetexchange.org).

4. No special trick is used generally, but some C++ tricks may need to be learned.

5. Templates may prevent understanding the code via some debug techniques.

And something good:

1. GOOOOOOD PROGRAMMING STYLE

2. The output is mostly handled by printer functions, so easier to mess around.

To clone the repository, don't forget to add "--recursive" to include all the dependencies,

git clone --recursive https://github.com/Walter-Feng/Hartree-Fock-in-CPP.git

CMake is used to generate the executable:

Hartree-Fock-in-CPP-repo-directory: $ mkdir build 
Hartree-Fock-in-CPP-repo-directory/build: $ cmake ..
Hartree-Fock-in-CPP-repo-directory/build: $ make

Type

./hfincpp -h

to unlock all options.

To run the example:

Hartree-Fock-in-CPP-repo-directory/build: $ ./hfincpp ../example/H2_6_31g.json

The output should be something like:

============================================================
|Iter|        Time / s |   Energy / a.u. |     Energy Diff |
============================================================
     0                 0       -0.90365255                 0
     1        2.0781e-05       -0.91596619      -0.012313641
     2        3.5345e-05       -0.91626362    -0.00029743166
     3        4.6462e-05       -0.91627108    -7.4536724e-06
     4        5.6896e-05       -0.91627126    -1.8785583e-07
============================================================

Total time elapsed: 0.032478466 s

Json file format is used for input, and due to lack of default parameters, any missing keywords will generate a non-trackable error. So for now it is advised to only change the values in example json input file.