Skip to content

Examples

Choose a calculation by the question you want to answer. Each worked example connects physical assumptions, code, and saved results, then explains what the result supports and how to investigate further. If the topic name is unfamiliar, use the topic explorer to move between a broad idea, its prerequisites, and related questions.

Follow the first CPU simulation, then work through fundamentals. In particular, size distributions explains why count, concentration, and distribution density must be distinguished before comparing any plots. Gas and environment connects stored vapor mass with saturation.

Question Start here What to compare
Why do particles gain or lose material? Condensation and other processes Gas plus particle inventories, size-dependent growth, and timestep sensitivity.
Can number decrease while material is conserved? Coagulation representations Number, each material inventory, and the meaning of the distribution ordinate.
Where do particles enter or leave a chamber? Nucleation and wall loss Source and sink budgets, size-dependent survival, and stochastic spread.
How does composition alter equilibrium? Thermodynamics Activity, curvature, phase behavior, and the assumptions behind an equilibrium endpoint.
How does a size population interact with light? Sphere optics Single-particle efficiencies, population-weighted coefficients, and optical depth.
What changes when processes act together? Scientific simulations Controlled comparisons, external forcing, and limitations that a final plot can hide.
How can many independent histories be practical? GPU studies Per-box response, ensemble variability, numerical checks, and complete-workflow cost.

Start with one mechanism and a quantity you can check independently. Combine processes only after you can explain their separate effects. A plausible graph or conserved total does not establish numerical convergence or agreement with measurements. Some retained simulations deliberately expose model or input failures; their explanations identify what must change before research use.

For a new study, state the proposed comparison, hold appropriate inputs fixed, and choose a diagnostic that could distinguish the competing explanations. Use Theory for the physical basis and the Guide for supported interfaces. The agent plugins can help carry this structure into simulation design, execution, and analysis.

Published pages render saved outputs from maintained Python sources. Download the Python file or the helper ZIP when a study uses shared code or data. Follow the page’s environment and execution requirements; CUDA publication runs require CUDA, while the scaling report plots existing measurements on the CPU.

The GPU index distinguishes prepared scientific studies from small direct-kernel and resident-interface demonstrations. Longer simulation workloads retain their full inputs and numerical checks; begin with a small process example before adapting them. See installation for setup and example authoring for the source and saved-result workflow.

Explore this topic: Aerosol science and modeling