Exercise G#

Nonlinear problems: profiling, scalability, and work-precision analysis for the KdV solver.

Available scripts

Profiling#

compute_profiling_functions.py – Function-level profiling using cProfile. Saves to data/A2/ex_g/cprofile_functions.parquet.
compute_profiling_lines.py – Line-by-line profiling using line_profiler. Requires @profile decorators in src/spectral/tdp.py.

Usage: uv run kernprof -l Exercises/exercise_g/compute_profiling_lines.py then uv run python Exercises/exercise_g/compute_profiling_lines.py --parse
plot_profiling_functions.py – Function-level profiling visualizations. Saves to figures/A2/ex_g/function_profiling.pdf.
plot_profiling_lines.py – Line-by-line profiling visualizations. Saves to figures/A2/ex_g/line_profiling.pdf.

compute_scalability.py – Measures strong/weak scaling behavior. Saves to data/A2/ex_g/scalability_timing.parquet.
compute_work-precision.py – Sweeps stable time-step fractions for RK3/RK4 and records work vs accuracy metrics. Saves to data/A2/ex_g/work_precision.parquet.
plot_scalability.py – Creates scalability analysis plots. Saves to figures/A2/ex_g/scalability_analysis.pdf.
plot_work-precision.py – Creates work-precision diagrams. Saves to figures/A2/ex_g/work_precision.pdf.

Note: Run the corresponding compute_*.py or profile_*.py scripts before plotting.