Tune learning rate, momentum, and stopping rules easily. Model stable optimization for polynomial objectives accurately. Visualize every update before choosing better training parameters today.
| Scenario | Polynomial | Start x | Learning Rate | Momentum | Tolerance | Expected Trend |
|---|---|---|---|---|---|---|
| Default convex curve | x² - 6x + 13 | 8 | 0.10 | 0.20 | 0.0001 | Moves toward x = 3 |
| Sharper quartic bowl | 0.05x⁴ + x² - 4x + 7 | 5 | 0.03 | 0.35 | 0.0001 | Requires smaller updates |
| Wide quadratic path | 0.4x² - 2x + 9 | -6 | 0.18 | 0.10 | 0.0010 | Converges with larger steps |
The calculator optimizes a polynomial objective:
f(x) = c₄x⁴ + c₃x³ + c₂x² + c₁x + c₀
The gradient is:
f′(x) = 4c₄x³ + 3c₃x² + 2c₂x + c₁
The effective learning rate uses decay:
αᵢ = α / (1 + decay × (i - 1))
The momentum update is:
vᵢ = βvᵢ₋₁ - αᵢgᵢ
The new position is:
xᵢ₊₁ = xᵢ + vᵢ
For maximization, the gradient direction is inverted before the update. Gradient clipping limits unstable steps when the derivative becomes too large.
Important Note: All the Calculators listed in this site are for educational purpose only and we do not guarentee the accuracy of results. Please do consult with other sources as well.