General purpose HNC code

The current release is version 1.13.

Version 1.13 allows dissimilar and non-additive sphere diameters in the RPM and hard sphere potentials.
Version 1.12 adds direct calculation of HNC free energy (see documentation).
Version 1.11 fixes another bug: missing mean-field contribution to thermodynamics for off-SYM RPM potential.
Version 1.10 fixes a bug: missing contact virial pressure contribution in the HNC free energy.
Version 1.9 brings the source code up to modern FORTRAN standards and corrects a problem with EXP in v1.8.
Version 1.8 includes the MSA closure and various potentials for Slater smeared charges.
Version 1.7 handles an arbitrary number of components (rather than a maximum of three, as for previous versions).

Features

• FORTRAN 90 based, with example python driver scripts;
• HNC, MSA, RPA, and EXP closures;
• fast Ng solver, and Ng decomposition for long range potentials;
• multicomponent (arbitrary number of components);
• hard core (RPM-like) and soft core (DPD-like) potentials;
• full structural thermodynamics;
• fully open source.

See PDF documentation for details.

A related pure python code with a more limited feature set is available in a separate GitHub project. It might also be helpful to browse the documentation for this project too.

Citation

You are of course expected to use and modify this code for your own projects! If you end up publishing results based on this code, or a derived version, please cite :

Screening properties of Gaussian electrolyte models, with application to dissipative particle dynamics, P. B. Warren, A. Vlasov, L. Anton and A. J. Masters, J. Chem. Phys. 138, 204907 (2013).

If you have difficulty getting hold of this article, contact me and I will send a reprint. Alternatively, a preprint version can be found at arXiv:1303.0891.

Here is a BibTeX entry for the same :

@article{WVA+13,
  author         = {Warren, P. B. and Vlasov, A. and Anton, L. and Masters, A. J.},
  title          = {Screening properties of {G}aussian electrolyte models,
                    with application to dissipative particle dynamics},
  journal        = {J. Chem. Phys.},
  volume         = {138},
  pages          = {204907},
  year           = 2013
}

Installation Notes

The main code is written as a FORTRAN 90 module, and can be compiled using gfortran from the GNU compiler collection. A basic Makefile is provided. The code is designed to be run from python using the f2py interface from SciPy, though some examples of 'vanilla' FORTRAN 90 driver codes are included. The default target in the Makefile builds the required files for the f2py interface.

The LAPACK linear algebra library and FFTW fast Fourier transform library are both required.

The compiler tools including f2py, and pre-built versions of the lapack and fftw libraries, are available for most modern GNU/Linux distributions.

Roadmap

Whilst the code is very functional there still remain legacy issues from its origins in a monolithic FORTRAN 90 code:

• aspects of the initialisation (potential model definitions) and post-processing (thermodynamic calculations) do not necessarily have to be done in FORTRAN;

• many parameters are hard-wired into the code, such as in the potential functions, with in some cases overloaded definitions;

• adding new potentials, or changing / verifying the current ones is clunky and requires access to the underlying FORTRAN code;

• the solver grid is allocated at startup, and cannot currently be re-sized, nor can multiple grids be used in the same application;

• certain features of the current code, such as reporting the solver status, involve string manipulations which are certainly better handled directly in python;

• overall, the current interface doesn't take full advantage of what might be considered modern python functionality and design patterns.

A longer term aim therefore is to split out the central Ornstein-Zernike solver with the closure approximations into a streamlined standalone suite of FORTRAN 90 kernel functions. This would particularly preserve the technically complex and bespoke Ng accelerated convergence scheme. The initialisation of specific potential models (DPD, RPM, etc) would then be implemented using NumPy in a separate python module. This would make the potential model definitions and the associated parameter specifications more accessible, user-friendly, and better in accord with pythonic idioms.

Initially, the downstream thermodynamic calculations could be left in FORTRAN but in time these too could be moved into python, using for example the trapz function in NumPy to handle the integrations, along the lines of the related pure python Ornstein-Zernike-HNC solver here.

Thus, functionally, the code would split into two:

• a sunlighthnc-core module, wrapping the core FORTRAN 90 functions;
• a sunlighthnc module, providing the application user interface.

If-and-when this comes to pass, this will be version 2.0 of the code !!

Copying

SunlightHNC is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

SunlightHNC is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.

Copyright

SunlightDPD is based on an original code copyright © 2007 Lucian Anton, with modifications copyright © 2008, 2009 Andrey Vlasov, and additional modifications copyright © 2009-2018 Unilever UK Central Resources Ltd (Registered in London number 29140; Registered Office: Unilever House, 100 Victoria Embankment, London EC4Y 0DY, UK). Later modifications copyright © 2020-2023 Patrick B Warren (STFC).

Contact

Send email to patrick.warren{at}stfc.ac.uk