13:12:31From  Henke :
1. Multi-core simulation.
2. GPU / CUDA.
3. Supermachine at Ångström.

13:23:19	 From  Henke : Start by doing as in the paper by Kaxiras & Neofotistos.

13:28:35	 From  Henke : The project is about the simulation(s),
 not about the theoretical, analytical, mathematical solutions.

13:28:38	 From  Yuan Yao : The project is very like a programming contest 
problem held by ACM or IEEE. Usually the first step is being familiar with the 
model, that is, the paper. Speaking of implementation, the following steps are 
helpful: 1, problem abstraction (how to build the model in programming 
language); 2, data structure/class definition (what data structure you use for 
the problem); 3, develop the mechanisms/dynamics. 4, run the built module. 

13:30:02	 From  Yuan Yao : For each step, a brain-storming with teammates is 
necessary (and more importantly it is fun!) Do things in paper first, and then 
move to codeing.  

13:35:27	 From  Yuan Yao : compressed column format

13:36:58	 From  Henke : Start with "agent-based" simulation.
For example something like:
https://github.com/topics/agent-based-simulation  ?
https://github.com/topics/agent-based-modeling    ?

13:37:57	 From  Henke : Look at clustering, sampling.

13:39:29	 From  Henke : https://en.wikipedia.org/wiki/COVID-19_apps (?)

13:44:09	 From  Henke : Distance of infection?
 - Fixed distance.
 - Some model for dynamic distance.
(Compare with covid apps in smartphones.)

13:46:18	 From  Henke : 1. Multi-core simulation: skymarc.
2. GPU / CUDA.
3. Message passing on Rackham (supermachine at Ångström).

13:47:28	 From  Henke : Arrays or graphs? (CRF / CCF)
 - Data structure differentiations.