eafit-apolo / 2dpartint Goto Github PK

It is currently using the 70% of the simulation time. We have to improve this.

• New visualization program?: Try to find a visualization program that gives us the possibility to implement a new way to write to files in a more performant fashion.

• New file format?: Should we use a more flexible file format? One that allows us to write in binary? One that keep all the step points in time in the same file, opposed to the dozens CSV that we currently have?

• On a new thread?: Would it be better implement a thread with the sole purpose of writing to the files? Keep in mind that this might not be an option now, because we do not know how we are going to parallelized the code, so we have to be sure that this thread will not interfere with the ones we will have when we implement our parallelization.

We want to test a simulation as a hole. Given a number of particles n in a given time t, we want to know where these particles will be in t+X, where n and X are small numbers. Use Cundall's paper, which is in files in Teams.

Reading Cundall paper, I realised there is a step we are not doing. We are not calculating the resultant momentum when two particles are colliding.

From Cundall, 1979:

Would it be better implement a thread with the sole purpose of writing to the files? Keep in mind that this might not be an option now, because we do not know how we are going to parallelized the code, so we have to be sure that this thread will not interfere with the ones we will have when we implement our parallelization.

Check the existing branches, and the existing pull request. Take the appropiate actions (merge, remove, save for later, etc).

Searching for a new data structure...

Should we use a more flexible file format?

Find out what it does, and if it is save to uncomment it. If not, rewrite it so that it works.

Try to find a visualization program that gives us the possibility to implement a new way to write to files in a more performant fashion.