How foo/online-knapsack works
Project type and purpose
• This is an academic research‐code repository accompanying the ICALP 2020 paper “An optimal algorithm for online multiple knapsack.”
• It contains:
• Three LaTeX targets (ICALP submission, Arxiv version, journal version) under top‐level folders “0-ICALP”, “1-Arxiv”, “2-journal”.
• A “code” folder with multiple experiment modules (C++ implementations, Python scripts, plotting pipelines).
• Utilities, data files, and Makefiles to drive builds, runs, data‐generation, and plot generation.
File structure
• Top level:
• Makefile & README.md – orchestrate full build (PDF + experiments + plots)
• 0-ICALP, 1-Arxiv, 2-journal – LaTeX sources and assets for each paper variant
• code/ – all experiment source code submodules
• utilities.py & code/utilities/ – shared helper functions
• code/ submodules (each with its own Makefile or script pipeline):
• case-check, fighting-two-arg, eval, estimations – C++ executables to run algorithms and collect raw results
• knapsack-plot – Python + gnuplot scripts to transform data.dat → .eps/.pdf
• plots – C++ + gnuplot templates to generate the “final” set of 09 figures
• szysio – various Python helper scripts (data processing, bounds calculators)
• trash – legacy or exploratory code (safe to ignore or annotate as “archived experiments”)
• Data flow: C++/Python code → raw data (.dat) → gnuplot/python → plot files (.eps/.pdf) → LaTeX inclusion → final paper PDFs.
Key insights from README
• Describes the core algorithmic contribution (online multiple knapsack).
Detail
• Shared utilities as a dependency across multiple experiment modules.