I just wanted to say hi and share what made me change my mind about my current setting.
I am an applied mathematician working in scientific computing and geospatial data analysis. On the scientific computing side, I will be definitively switching from Python+Fortran to Julia for my next project. As for geospatial data analysis, I believe I will be sticking to pure Python for the moment, as I’m only using it as a wrapper to efficient C libraries.
First, I wanted to say that I didn’t believe in the publicized performance tests until I run my own. I thought, maybe they are cherry-picking stuff, running toy problems, etc. So I went to the most significant (and yet simple) operation that sucks up most computational resources in my area, which is the evaluation of lots and lots of complex exponential functions.
I still can’t believe that Julia is about 30% faster than gfortran in my test, with multithreading and full optimization flags.
I wrote a blog post with details of my experience. I’m also new to blogging so I would really appreciate any feedback. Here you can find the link:
I look forward to probably blogging more about Julia, as the writing stimulates me to learn more and document technical details that don’t end up in papers or the like. I will also be checking the coverage of my most frequent topics in terms of Julia packages.