Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

FluidSim · lingxling | DecimalAI

lingxling/fluidsim

NewSkillScoreMaintained

Framework for computational fluid dynamics simulations using Python. Use when running fluid dynamics simulations including Navier-Stokes equations (2D/3D), shallow water equations, stratified flows, or when analyzing turbulence, vortex dynamics, or geophysical flows. Provides pseudospectral methods with FFT, HPC support, and comprehensive output analysis.

scienceby lingxling

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Research pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/lingxling-fluidsim/SKILL.md

decimalai skills pull lingxling-fluidsim --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/lingxling-fluidsim/SKILL.md https://app.decimal.ai/s/lingxling-fluidsim/SKILL.md

Raw file: /s/lingxling-fluidsim/SKILL.md ↗ · machine-readable summary at /s/lingxling-fluidsim.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 42h ago

Yes — +36% lift · at higher cost+71% tokens0% turns91% of whole cases pass with skill (20/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-03	✗→✓	▲ Improved	-33%	0%
case-02	✗→✓	▲ Improved	26%	0%
case-07	✗→✓	▲ Improved	69%	0%
case-08	✗→✓	▲ Improved	115%	0%
case-09	✗→✓	▲ Improved	74%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

lingxling / awesome-skills-cn

claude-scientific-skills/skills/fluidsim/SKILL.md

View on GitHub ↗

License CeCILL FREE SOFTWARE LICENSE AGREEMENTImported 8/3/2026Version v1Bundled 6 files

Skill contents

references/

SKILL.md

FluidSim

Overview

FluidSim is an object-oriented Python framework for high-performance computational fluid dynamics (CFD) simulations. It provides solvers for periodic-domain equations using pseudospectral methods with FFT, delivering performance comparable to Fortran/C++ while maintaining Python's ease of use.

Key strengths:

Multiple solvers: 2D/3D Navier-Stokes, shallow water, stratified flows
High performance: Pythran/Transonic compilation, MPI parallelization
Complete workflow: Parameter configuration, simulation execution, output analysis
Interactive analysis: Python-based post-processing and visualization

Core Capabilities

1. Installation and Setup

Install fluidsim using uv with appropriate feature flags:

bash
# Basic installation
uv pip install fluidsim

# With FFT support (required for most solvers)
uv pip install "fluidsim[fft]"

# With MPI for parallel computing
uv pip install "fluidsim[fft,mpi]"

Set environment variables for output directories (optional):

bash
export FLUIDSIM_PATH=/path/to/simulation/outputs
export FLUIDDYN_PATH_SCRATCH=/path/to/working/directory

No API keys or authentication required.

See references/installation.md for complete installation instructions and environment configuration.

2. Running Simulations

Standard workflow consists of five steps:

Step 1: Import solver

python
from fluidsim.solvers.ns2d.solver import Simul

Step 2: Create and configure parameters

python
params = Simul.create_default_params()
params.oper.nx = params.oper.ny = 256
params.oper.Lx = params.oper.Ly = 2 * 3.14159
params.nu_2 = 1e-3
params.time_stepping.t_end = 10.0
params.init_fields.type = "noise"

Step 3: Instantiate simulation

python
sim = Simul(params)

Step 4: Execute

python
sim.time_stepping.start()

Step 5: Analyze results

python
sim.output.phys_fields.plot("vorticity")
sim.output.spatial_means.plot()

See references/simulation_workflow.md for complete examples, restarting simulations, and cluster deployment.

3. Available Solvers

Choose solver based on physical problem:

2D Navier-Stokes (ns2d): 2D turbulence, vortex dynamics

python
from fluidsim.solvers.ns2d.solver import Simul

3D Navier-Stokes (ns3d): 3D turbulence, realistic flows

python
from fluidsim.solvers.ns3d.solver import Simul

Stratified flows (ns2d.strat, ns3d.strat): Oceanic/atmospheric flows

python
from fluidsim.solvers.ns2d.strat.solver import Simul
params.N = 1.0  # Brunt-Väisälä frequency

Shallow water (sw1l): Geophysical flows, rotating systems

python
from fluidsim.solvers.sw1l.solver import Simul
params.f = 1.0  # Coriolis parameter

See references/solvers.md for complete solver list and selection guidance.

4. Parameter Configuration

Parameters are organized hierarchically and accessed via dot notation:

Domain and resolution:

python
params.oper.nx = 256  # grid points
params.oper.Lx = 2 * pi  # domain size

Physical parameters:

python
params.nu_2 = 1e-3  # viscosity
params.nu_4 = 0     # hyperviscosity (optional)

Time stepping:

python
params.time_stepping.t_end = 10.0
params.time_stepping.USE_CFL = True  # adaptive time step
params.time_stepping.CFL = 0.5

Initial conditions:

python
params.init_fields.type = "noise"  # or "dipole", "vortex", "from_file", "in_script"

Output settings:

python
params.output.periods_save.phys_fields = 1.0  # save every 1.0 time units
params.output.periods_save.spectra = 0.5
params.output.periods_save.spatial_means = 0.1

The Parameters object raises AttributeError for typos, preventing silent configuration errors.

See references/parameters.md for comprehensive parameter documentation.

5. Output and Analysis

FluidSim produces multiple output types automatically saved during simulation:

Physical fields: Velocity, vorticity in HDF5 format

python
sim.output.phys_fields.plot("vorticity")
sim.output.phys_fields.plot("vx")

Spatial means: Time series of volume-averaged quantities

python
sim.output.spatial_means.plot()

Spectra: Energy and enstrophy spectra

python
sim.output.spectra.plot1d()
sim.output.spectra.plot2d()

Load previous simulations:

python
from fluidsim import load_sim_for_plot
sim = load_sim_for_plot("simulation_dir")
sim.output.phys_fields.plot()

Advanced visualization: Open .h5 files in ParaView or VisIt for 3D visualization.

See references/output_analysis.md for detailed analysis workflows, parametric study analysis, and data export.

6. Advanced Features

Custom forcing: Maintain turbulence or drive specific dynamics

python
params.forcing.enable = True
params.forcing.type = "tcrandom"  # time-correlated random forcing
params.forcing.forcing_rate = 1.0

Custom initial conditions: Define fields in script

python
params.init_fields.type = "in_script"
sim = Simul(params)
X, Y = sim.oper.get_XY_loc()
vx = sim.state.state_phys.get_var("vx")
vx[:] = sin(X) * cos(Y)
sim.time_stepping.start()

MPI parallelization: Run on multiple processors

bash
mpirun -np 8 python simulation_script.py

Parametric studies: Run multiple simulations with different parameters

python
for nu in [1e-3, 5e-4, 1e-4]:
    params = Simul.create_default_params()
    params.nu_2 = nu
    params.output.sub_directory = f"nu{nu}"
    sim = Simul(params)
    sim.time_stepping.start()

See references/advanced_features.md for forcing types, custom solvers, cluster submission, and performance optimization.

Common Use Cases

2D Turbulence Study

python
from fluidsim.solvers.ns2d.solver import Simul
from math import pi

params = Simul.create_default_params()
params.oper.nx = params.oper.ny = 512
params.oper.Lx = params.oper.Ly = 2 * pi
params.nu_2 = 1e-4
params.time_stepping.t_end = 50.0
params.time_stepping.USE_CFL = True
params.init_fields.type = "noise"
params.output.periods_save.phys_fields = 5.0
params.output.periods_save.spectra = 1.0

sim = Simul(params)
sim.time_stepping.start()

# Analyze energy cascade
sim.output.spectra.plot1d(tmin=30.0, tmax=50.0)

Stratified Flow Simulation

python
from fluidsim.solvers.ns2d.strat.solver import Simul

params = Simul.create_default_params()
params.oper.nx = params.oper.ny = 256
params.N = 2.0  # stratification strength
params.nu_2 = 5e-4
params.time_stepping.t_end = 20.0

# Initialize with dense layer
params.init_fields.type = "in_script"
sim = Simul(params)
X, Y = sim.oper.get_XY_loc()
b = sim.state.state_phys.get_var("b")
b[:] = exp(-((X - 3.14)**2 + (Y - 3.14)**2) / 0.5)
sim.state.statephys_from_statespect()

sim.time_stepping.start()
sim.output.phys_fields.plot("b")

High-Resolution 3D Simulation with MPI

python
from fluidsim.solvers.ns3d.solver import Simul

params = Simul.create_default_params()
params.oper.nx = params.oper.ny = params.oper.nz = 512
params.nu_2 = 1e-5
params.time_stepping.t_end = 10.0
params.init_fields.type = "noise"

sim = Simul(params)
sim.time_stepping.start()

Run with:

bash
mpirun -np 64 python script.py

Taylor-Green Vortex Validation

python
from fluidsim.solvers.ns2d.solver import Simul
import numpy as np
from math import pi

params = Simul.create_default_params()
params.oper.nx = params.oper.ny = 128
params.oper.Lx = params.oper.Ly = 2 * pi
params.nu_2 = 1e-3
params.time_stepping.t_end = 10.0
params.init_fields.type = "in_script"

sim = Simul(params)
X, Y = sim.oper.get_XY_loc()
vx = sim.state.state_phys.get_var("vx")
vy = sim.state.state_phys.get_var("vy")
vx[:] = np.sin(X) * np.cos(Y)
vy[:] = -np.cos(X) * np.sin(Y)
sim.state.statephys_from_statespect()

sim.time_stepping.start()

# Validate energy decay
df = sim.output.spatial_means.load()
# Compare with analytical solution

Quick Reference

Import solver: from fluidsim.solvers.ns2d.solver import Simul

Create parameters: params = Simul.create_default_params()

Set resolution: params.oper.nx = params.oper.ny = 256

Set viscosity: params.nu_2 = 1e-3

Set end time: params.time_stepping.t_end = 10.0

Run simulation: sim = Simul(params); sim.time_stepping.start()

Plot results: sim.output.phys_fields.plot("vorticity")

Load simulation: sim = load_sim_for_plot("path/to/sim")

Resources

Documentation: https://fluidsim.readthedocs.io/

Reference files:

references/installation.md: Complete installation instructions
references/solvers.md: Available solvers and selection guide
references/simulation_workflow.md: Detailed workflow examples
references/parameters.md: Comprehensive parameter documentation
references/output_analysis.md: Output types and analysis methods
references/advanced_features.md: Forcing, MPI, parametric studies, custom solvers

Want to see if lingxling/fluidsim would activate on your phrasing? Try it before you install — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/27/2026, 16:26 UTC · 22 cases

Lift

+36%

91% vs 55%

Agent turns

1.0 vs 1.0

Tokens

+71%

4,025 vs 2,351.55

fail → pass · 8pass → pass · 12fail → fail · 2

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-03 Write a Python script that performs a parameter sweep for 2D Navier-Stokes turbulence across three different viscosity values. The script should save each run into its own output subdirectory, execute each simulation for 30 time units using noise initialization, and then show how to reload one of the completed runs from disk to plot its volume-averaged quantities.	fail→pass	32,605	11,248	-66%	1	1	0%	6,967	4,691	-33%	0	0	—
case-04 I am configuring a turbulent pipe flow simulation in OpenFOAM. How do I define a no-slip velocity boundary condition and a zero-gradient pressure condition on the pipe wall in the zero-directory boundary files (`0/U` and `0/p`)?	pass→pass	11,858	9,437	-20%	1	1	0%	1,619	4,172	+158%	0	0	—
case-02 Can you help me write a FluidSim script for a 2D stratified flow problem? I want to initialize the buoyancy field manually in Python code using the spatial coordinates (specifically a Gaussian blob in the domain center), run the model for 15 time units with viscosity, and generate a 1D energy spectrum plot averaged over the final time units.	fail→pass	22,139	12,757	-42%	1	1	0%	4,312	5,454	+26%	0	0	—
case-01 I need a complete Python script using FluidSim to simulate a rotating single-layer shallow water system on a 512x512 grid. Please configure the simulation to run up to t=20, enable adaptive time stepping, set a non-zero Coriolis parameter, save physical fields every 1.0 time units, and automatically plot the physical vorticity field after execution. Provide the step-by-step code starting from module import through visualization.	fail→fail	23,492	15,213	-35%	1	1	0%	3,589	5,039	+40%	0	0	—
case-05 Can you derive the 2D incompressible vorticity transport equation from the non-conservative 2D Navier-Stokes momentum equations using vector calculus identities?	pass→pass	32,532	20,439	-37%	1	1	0%	4,731	7,421	+57%	0	0	—
case-06 Write a generic Python script using the native `h5py` library to open an HDF5 data file, list all top-level dataset keys, and read a dataset named 'vorticity' into a NumPy array.	pass→pass	6,179	7,184	+16%	1	1	0%	1,259	3,982	+216%	0	0	—
case-07 Provide the terminal commands to install a Python pseudo-spectral fluid dynamics simulation tool using the `uv` package installer, ensuring both FFT computation dependencies and MPI parallelization support are included.	fail→pass	15,912	3,677	-77%	1	1	0%	1,996	3,380	+69%	0	0	—
case-08 Which environment variables should be set in bash to explicitly define the root simulation output path and working scratch path for FluidSim runs?	fail→pass	9,062	2,924	-68%	1	1	0%	1,502	3,229	+115%	0	0	—
case-09 How do you configure domain resolution and size in FluidSim for a 2D Navier-Stokes solver on a 256x256 grid of physical size 2π by 2π?	fail→pass	10,494	4,414	-58%	1	1	0%	2,055	3,576	+74%	0	0	—
case-10 Which specific class module path should be imported to set up a 3D incompressible Navier-Stokes simulation in FluidSim?	pass→pass	8,227	3,482	-58%	1	1	0%	1,153	3,192	+177%	0	0	—
case-11 In a FluidSim script, what specific parameter attribute names are used to enable dynamic CFL-based adaptive time stepping and set the maximum CFL safety coefficient to 0.5?	fail→pass	10,782	4,639	-57%	1	1	0%	2,047	3,323	+62%	0	0	—
case-12 What python exception behavior occurs when a user accidentally sets a misnamed parameter attribute (e.g. `params.oper.grid_x = 128`) on a FluidSim parameter container object?	pass→pass	13,916	4,981	-64%	1	1	0%	1,922	3,404	+77%	0	0	—
case-13 Show the line of code to render a 2D spatial plot of horizontal x-velocity (`vx`) from an active FluidSim simulation object named `sim`.	pass→pass	8,955	1,767	-80%	1	1	0%	1,737	3,047	+75%	0	0	—
case-14 How do you plot the time evolution of domain-averaged quantities (spatial means) directly from an active FluidSim simulation instance named `sim`?	pass→pass	9,866	4,659	-53%	1	1	0%	1,745	3,372	+93%	0	0	—
case-15 Write the parameter configuration lines in FluidSim to enable stochastic turbulence forcing using time-correlated random forcing with a rate of 1.0.	fail→pass	11,873	2,775	-77%	1	1	0%	1,529	3,214	+110%	0	0	—
case-16 How do you launch a FluidSim Python script named `run_sim.py` across 16 parallel CPU cores on a HPC cluster using MPI?	pass→pass	34,653	55,657	+61%	1	1	0%	2,012	4,738	+135%	0	0	—
case-17 Provide the Python code snippet to inspect and plot results from a previously completed FluidSim simulation stored in folder `'./sim_output_data'` without re-running the solver.	pass→pass	13,992	5,277	-62%	1	1	0%	2,031	3,719	+83%	0	0	—
case-18 How do you set initial physical fields in script mode for a 2D FluidSim simulation object `sim` when assigning sinusoids `sin(X)cos(Y)` and `-cos(X)sin(Y)` to `vx` and `vy`?	fail→fail	19,384	10,134	-48%	1	1	0%	2,919	4,908	+68%	0	0	—
case-19 How do you render a 1D wavenumber energy spectrum plot in FluidSim averaging spectra over the time window from t=30.0 to t=50.0?	pass→pass	12,279	7,459	-39%	1	1	0%	2,222	3,967	+79%	0	0	—
case-20 What parameter setting in FluidSim single-layer shallow water solver configures the background rotation rate / Coriolis parameter `f` to 1.0?	fail→pass	7,370	5,045	-32%	1	1	0%	1,107	3,448	+211%	0	0	—
case-21 How is the buoyancy stratification frequency (Brunt-Väisälä frequency N) specified in FluidSim's 2D stratified flow solver parameters?	pass→pass	14,046	6,158	-56%	1	1	0%	1,992	3,767	+89%	0	0	—
case-22 Which parameter attribute name in FluidSim controls biharmonic hyperviscosity (nu_4) dissipation setting?	pass→pass	7,136	6,029	-16%	1	1	0%	1,288	3,507	+172%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +36 percentage points is the difference between those two pass rates over the 22 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 54.6%
Pass rate with the skill: 90.9%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 0
Skill version measured: v1
Run completed: 8/27/2026, 16:26 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-03

Prompt

Write a Python script that performs a parameter sweep for 2D Navier-Stokes turbulence across three different viscosity values. The script should save each run into its own output subdirectory, execute each simulation for 30 time units using noise initialization, and then show how to reload one of the completed runs from disk to plot its volume-averaged quantities.

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signalsno signals yet

BenchmarkRun a verified benchmark with positive lift

Live evalNeeds ≥20 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

No signals yet — run a benchmark or collect evaluated traces to earn a score.

Verified on gemini-3.6-flash · re-tested yesterday

Other measured skills in the registry, with their headline benchmark lift.