Programming
Programming is one of the most important skills in exploration
geophysics. My goal as a programmer is not ultimate efficiency,
but accuracy, clarity and reusability. I've programmed to solve:
- Research-size problems in Fortran, C and C++
- Toy-size problems in Java, Matlab
High performance computing
We deal with large scale parallel computation all the time. I
achieve multilevel parallelism using two different approaches:
- OpenMP + MPI
- OpenMP + scripts
I found the latter one simpler and easier to debug. It is also
more robust against system failures.
Reproducibility
At SEP, we take reproducibility seriously. Personally, I benefit a
lot from the reproducibility requirements. Often times, the need of
reproducing a result comes from the researcher himself/herself. That
is why I am always one of the first students who finish their own
reproducibility tests.