Hello Devs, I have a version of the code with Lees-Edwards boundary conditions implemented, on my github page:
https://github.com/tojb/espresso
I'm hoping that the forthcoming coding day will be a good opportunity for people to have a look at it.
My approach has been to add #ifdef LEES_EDWARDS a lot to the domain decomposition code, and then to break out new functions to a le_domain_decomposition.c file whenever I found that I was making the main domain_decomposition.c too messy.
I expect that when Anwar gets back from his holiday (next Monday) he will start using the code for serious production runs, so that I'm afraid is when the bugs will probably start to show up in numbers.
Current status:
- passes the lees_edwards.tcl test suite and doesn't break any of the other features (that are compiled-in by default)
- there is a new paragraph on Lees-Edwards in the docs.
Scope for optimisation:
- forces a Verlet rebuild at every timestep
- makes a certain amount of redundant comms
So: I feel like this is ready for scrutiny, although probably it needs another tidy and some soak testing before an actual release, and am announcing for the benefit of interested parties.
Myself and Anwar have a growing collection of mods to Espresso which we are hoping to mainstream, mainly to save ourselves the trouble of patching them into new releases every time.
Josh