Not Invented Here: Porting Scientific Software to Python
summary
Shows a concrete migration story: a MATLAB conformal‑mapping port stalls after five files then a one‑character conjugate‑transpose bug, so the practical choices become clear. Demonstrations include calling Graphviz via subprocess, loading C functions with ctypes, embedding C++ with scipy.weave to use the Seagal Minkowski‑sum library, and running MATLAB from IPython with matlab magic. The talk names tradeoffs (compile time, dependency pain, profiling needs) and when each method actually works or hits a wall.
AI-written summary · about
chapters (7)
- 0:00Start
- 0:10NOT INVENTED HERE PORTING SCIENTIFIC SOFTWARE TO PYTHON
- 0:53WHAT MAKES SCIENTIFIC SOFTWARE DIFFERENT?
- 4:01NOT INVENTED HERE SYNDROME
- 5:18TWO REASONS TO LEARN ABOUT INTEROPERABILITY
- 5:34High Performance Python
- 6:44A WHIRLWIND TOUR OF INTEROPERABILITY TOOLS
related talks
32mNot Invented Here: Porting Scientific Software to Python
27mEngineers to Data-Scientists with Python
32mFrom Fortran to Python: A Conversation Across Generations of Scientific Computing
29mC++ ♥ Python: from modules to hybrid applications
41mPorting a Project from Python 2 to 3
30mEmpowered by Python - A success story
25mHow to bring your C++ project to Python land
23mPython(x,y): diving into scientific Python