How to work it 01 Choose the mode with the toggle row, or load a preset: Logistic growth, Newton's cooling, Pendulum, Damped oscillator, Van der Pol, Lotka–Volterra or Saddle. 02 Enter the expressions. A new parameter such as a in the Van der Pol preset appears immediately as a slider with a numeric box. 03 Set the view rectangle — xMin, xMax, yMin, yMax — which is both the drawing window and the region searched for equilibria. 04 Click the canvas to seed trajectories (up to twelve, forward and backward); Clear removes them. Presets seed one for you. 05 Read the cards and badges. Up to eight equilibrium cards list the refined point and its kind; a diverged or truncated trajectory raises a warning above the canvas.
Worked readouts The cards below are the exact displayed values.
System
Cards
Predator–prey preset: x′ = x*(1.1 - 0.4y), y′ = y (0.4*x - 0.4)
saddle at (0, 0) ; center at (1, 2.75)
Van der Pol preset: x′ = y, y′ = a*(1 - x^2)*y - x with a = 1
unstable at (0, 0)
From the preset seed (1, 1), the Van der Pol forward trajectory stays inside x ∈ [-2.009, 2.009] and y ∈ [-2.678, 2.678], and no warning badge appears.
How equilibria are classified Newton refines candidate roots inside the view and merges near-duplicates; the Jacobian at each accepted root decides the label. Negative determinant: saddle. Near-zero trace with positive determinant: centre. Negative trace: stable. Positive trace: unstable. The test is linear and local, which is why the cards and the trajectories tell complementary stories.
Assumptions and limits
Everything depends on the view rectangle. Equilibria outside the bounds are not searched, the grid scan can step past very closely spaced roots, and the coordinates shown are numerical roots — expect small residuals, not exact zeros.
Trajectories are adaptive but explicit. Dormand–Prince RK45 with automatic step control reports divergence or step-limit truncation instead of drawing a wrong curve; very stiff systems may hit those guards. What it does not do: no closed-form solution, no parameter sweep, no bifurcation tracking. For symbolic work use Math Tools ; for linear loop analysis use the control panel .
"Unstable" describes the point, not the system Van der Pol's card reads unstable at (0, 0) , yet the plotted trajectory does not escape — it winds onto a closed limit cycle and stays inside x ∈ [-2.009, 2.009] and y ∈ [-2.678, 2.678]. Both are true: the origin is a spiral source, while the nonlinear damping far from it removes energy, so trajectories are bounded but never settle. Read the classification together with the vector field and the seeded trajectories.
Where it fits Teaching qualitative dynamics The predator–prey preset shows the saddle at (0, 0) and the centre at (1, 2.75) in one picture; clicking closed orbits around the centre demonstrates the difference from a stable point without a formula.
Spot-checking a model before simulation Load Van der Pol and confirm the oscillation is self-sustaining rather than divergent, then try the damped oscillator preset for a stable equilibrium. The filter panel covers frequency-domain questions the phase portrait cannot.
Privacy Expressions, parameter values and trajectories are computed inside your browser; nothing you enter is uploaded.
References
Wikipedia, Phase plane , en.wikipedia.org (访问日期:2026-10-01)— trajectories, equilibria and their classification.
Wikipedia, Lotka–Volterra equations , en.wikipedia.org (访问日期:2026-10-01)— the predator–prey pair and its centre.
Wikipedia, Van der Pol oscillator , en.wikipedia.org (访问日期:2026-10-01)— unstable origin, nonlinear damping and the limit cycle.
Wikipedia, Dormand–Prince method , en.wikipedia.org (访问日期:2026-10-01)— the embedded RK45 scheme and adaptive step size.
Calculators in this hub
Hand-picked tools, one click away. The mini versions compute live and carry your values into the full calculator.
Sources & review
Reviewed by CalcX Editorial Team
Updated 2026-10-01