Mind the Gap: General-Purpose Programming Languages Impede Scientific Model(…)
summary
Position: general-purpose languages make domain meaning disappear, hampering model reuse and verification; evidence comes from re-architecting a neural‑network subgrid parameterization from SAM into CAM/CESM (required deep refactoring and lots of unit/semantic detective work). Prototype language Fortel shows statically checked, descriptor‑indexed floats that catch unit mismatches (meters vs miles) and layered quantity distinctions; tooling example Camfort reads Fortran comments to run dimensional checks; a single-source DSL generates constraints used by UI sliders, a genetic optimizer and LaTeX, and proves practical in those projects.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:02Programming languages in science
- 3:30Semantic gap in scientific code
- 8:25General-purpose languages hinder science
- 10:53Naming challenges in scientific code
- 15:53Single source of truth
- 20:47Better languages for science
- 22:53Demo: Fortel
- 27:23Closing remarks & Q&A
related talks
13mA Family of Languages for Trustworthy Agent-Based Simulation
17mFour Languages for Portability
47mThe linguist programmer
05mModel Driven Software: Putting the Cart Before the Egg
29mDeriving Integrated Multi-Viewpoint Modeling Languages from Heterogeneous Modeling...
29mApplying Self-Extension Mechanism to DSLs for Establishing Model Libraries
01h 07mThere Is Only One Time in Software (Language) Engineering!
18mOn the Expressive Power of Languages for Static Variability