JevMade Sign in
← Back to guides

JevMade field notes / Python scientific-computing tutorial

Turn a gas-flow description into checked solver inputs

Adam Karchani's Python tutorial uses Jev to match a written gas-flow request to bounded inputs for a small solver, with Claude helping write and use the tool.

Original by Adam Karchani / cfdgasmanIntegrations

Listen to this guide

JevMade’s plain-English explanation

0:00 /

Our summary

Adam Karchani presents a tool for people studying gas moving through tiny gaps. A user describes the gas and the channel in ordinary words. The tool turns that description into inputs for a calculation, or asks a question or refuses when the request does not fit its scope.

Python finds numbers with units and gives each a tag. Jev's Choice questions identify the flow, geometry and gas, then choose which tagged quantity means the gap, pressure or another input. Code converts units, applies decision rules and solves the equations. Claude helps build the program and manage the conversation.

The solver covers small disturbances between parallel plates, with fully diffuse walls, not general fluid problems. The author reports a sixty-request comparison that was not reproduced here. The README uses a confidence cutoff of 0.45 while the code defaults to 0.6; missing temperature is assumed, and new hosted requests transmit text.

Key takeaways

  1. Ask Jev which tagged quantity has a particular meaning; keep numerical extraction, unit conversion and equations in code.
  2. Offer missing or unsupported answers, and check which required values are assumed rather than asking for every missing detail.
  3. Resolve the documented cutoff mismatch before relying on the benchmark, and check that your physical problem fits the limited solver.

The archived source and tests were read, never installed or run. The MIT license credits Adam Karchani, whose account is cfdgasman. No publication date was established. Claims about the solver checks and 60 author-written requests remain author reports. The explanation uses a confidence cutoff of 0.45, but the code defaults to 0.6. That code assumes room temperature, 293.15 kelvin, when it is missing and may treat a lone pressure as mean pressure. A large pressure drop triggers a warning, not a refusal. Jev receives the original numbers in tagged text; code does the arithmetic.

cfdgasman/jev-cfd-copilot on GitHub

Read the original guide Opens the author’s site in a new tab.

Keep this for later

Sign in to bookmark experiments, guides and videos, and keep notes only you can see.

Continue to sign in

We’ll bring you back to this listing.