oneshotlm
Compare all models on this prompt →

Double pendulum · glm-5

Simulate a chaotic double pendulum on a canvas using the correct equations of motion, drawing the arms and a colorful fading trail of the tip.

⛨ sandboxed Open in new tab ↗
transcript · 4 messages
user

Simulate a chaotic double pendulum on a canvas using the correct equations of motion, drawing the arms and a colorful fading trail of the tip.

assistant
thinking
The user wants me to create a double pendulum simulation. This is a classic chaotic system that requires solving the equations of motion numerically. Let me think about what I need to include:

1. **Double Pendulum Physics**: The equations of motion for a double pendulum are derived from the Lagrangian. The system has two angles θ1 and θ2, and their derivatives.

2. **Equations of Motion**: The correct equations for a double pendulum are:

For a double pendulum with:
- m1, m2 = masses
- L1, L2 = lengths
- θ1, θ2 = angles from vertical
- ω1 = dθ1/dt, ω2 = dθ2/dt

The angular accelerations are:

θ1'' = [-g(2m1+m2)sin(θ1) - m2*g*sin(θ1-2θ2) - 2sin(θ1-θ2)*m2*(ω2²*L2 + ω1²*L1*cos(θ1-θ2))] / [L1*(2m1+m2-m2*cos(2θ1-2θ2))]

θ2'' = [2sin(θ1-θ2)*(ω1²*L1*(m1+m2) + g*(m1+m2)*cos(θ1) + ω2²*L2*m2*cos(θ1-θ2))] / [L2*(2m1+m2-m2*cos(2θ1-2θ2))]

3. **Numerical Integration**: I'll use Runge-Kutta 4th order method for accurate simulation.

4. **Visualization**:
   - Draw the two arms of the pendulum
   - Track the position of the second mass (tip)
   - Draw a fading trail with colors

5. **Canvas**: Use HTML5 canvas for rendering

Let me write this as a self-contained HTML file with inline CSS and JavaScript.

I'll create a double pendulum simulation with proper physics, canvas rendering, and a colorful fading trail. Let me write this as a single self-contained HTML file.

write /workspace/index.html (10213 bytes)
toolResult

Successfully wrote 10213 bytes to /workspace/index.html

assistant
thinking
I've created a complete double pendulum simulation. Let me summarize what I built:

1. **Physics**: Implemented the correct equations of motion derived from Lagrangian mechanics for a double pendulum. Used 4th-order Runge-Kutta integration for numerical stability.

2. **Visualization**: 
   - Canvas-based rendering with a dark theme
   - Two pendulum arms drawn as white lines
   - Masses at each joint (gray first mass, colored tip)
   - Colorful fading trail following the tip using HSL color cycling

3. **Features**:
   - Randomize button for new chaotic initial conditions
   - Clear trail button
   - Motion blur effect via partial canvas clearing
   - Glow effect on the tip mass

The simulation is fully self-contained in one HTML file with no external dependencies.

I've created a complete chaotic double pendulum simulation. Here's what's included: ## Physics Implementation - **Correct equations of motion** derived from Lagrangian mechanics for a double pendulum - **4th-order Runge-Kutta integration** for numerical stability and accuracy - Realistic parameters (masses, lengths, gravity) ## Visual Features - **Two pendulum arms** drawn as white lines from a pivot point - **Masses** rendered with gradient shading (gray for first mass, colorful tip) - **Rainbow fading trail** that follows the tip using HSL color cycling - **Motion blur effect** via partial canvas clearing - **Glow effect** around the tip mass ## Interactivity - **"Randomize" button** - generates new chaotic initial conditions - **"Clear Trail" button** - clears the trail while continuing simulation The simulation demonstrates chaos beautifully - even tiny changes in initial conditions produce wildly different trajectories. Click "Randomize" a few times to see how different starting angles create completely unique patterns!

TAKE IT TO CODE

A single self-contained HTML file — copy it and open in any browser, or reuse the prompt with your own pi-agent run.

RUN

modelglm-5
vendorZ.ai · GLM
harnesspi-agent
generated2026-07-22

EVALUATION

4/5 working

Smooth chaotic double pendulum with vivid rainbow trail, animates continuously; no distinct input controls beyond buttons.

Vision grade · anthropic/claude-sonnet-5

MEASUREMENTS

Raw readings from the capture, not a verdict — the grade above is the only judgement. Percentages are how much of the screen changed; the measure under-reports sparse motion on a dark background.

idle motion2.8%
on click3.5%
on drag1.7%
on wheel1.0%
on enter + space0.8%
on arrow keys0.9%
on w/a/s/d5.5%
frame spread21.1 / 255
console errors0
js errors none

METRICS

runtime45.9s
tokens in3.2k
tokens out3.5k
cached2.9k
cost / run$0.0089