Slopes of Parametric Curves
A parametric curve (x(t), y(t)) rarely passes the vertical line test, it can loop, cross itself, and double back, so it usually isn't the graph of any function y = f(x). Yet at (almost) every point it still has a perfectly good tangent direction, and the chain rule delivers its slope without ever solving for y in terms of x:
dy/dx = (dy/dt) / (dx/dt) = y'(t) / x'(t)
The logic: y changes at rate y'(t) per unit of parameter and x at rate x'(t); their ratio is the change in y per unit change in x, the slope. Thinking kinematically, (x', y') is the velocity vector, and dy/dx is just the direction that vector points, expressed as rise-over-run. The tangent line at the point is the line through it carrying exactly that direction.
How to Use This Calculator
Enter x(t) and y(t), a t-range for plotting, and optionally a specific t at which to evaluate. The calculator differentiates both coordinates symbolically, forms and simplifies the ratio y'/x' as a function of t, and evaluates it at your chosen parameter value, flagging the special cases: x' = 0 with y' ≠ 0 means a vertical tangent (infinite slope, perfectly good geometry), while both derivatives vanishing marks a singular point where the formula is silent (cusps live here). The evaluation point is marked on the plotted curve so slope and picture can be compared by eye.
Worked Example
The curve x = t², y = t³ − 3t crosses itself at (3, 0), reached at both t = √3 and t = −√3. Derivatives: x' = 2t, y' = 3t² − 3, so
dy/dx = (3t² − 3)/(2t)
At t = √3: slope = 6/(2√3) = √3 ≈ 1.732. At t = −√3: slope = −√3. One point, two tangent lines each pass through the crossing carries its own direction, and the parameter keeps them distinguished where a y = f(x) description would collapse in contradiction. That's the parametric viewpoint's superpower in one example, bookkeeping by time of arrival instead of by location.
Common Mistakes to Avoid
- Inverting the ratio. dy/dx = y'/x', y-rate on top. The flipped ratio is dx/dy, the reciprocal slope, a different quantity that happens to be right only when the slope is ±1.
- Calling x' = 0 an error. It's a vertical tangent (if y' ≠ 0), real geometry, as at the circle's leftmost and rightmost points. Only 0/0 is genuinely indeterminate, and those cusp points need limit analysis.
- Differentiating y with respect to x directly. There's no formula y(x) to differentiate, both coordinates must go through t. The whole method exists to avoid eliminating the parameter.
- Mishandling the second derivative. d²y/dx² is not y''/x''. Correctly: differentiate the slope function with respect to t, then divide by x' again, d²y/dx² = (d/dt[y'/x'])/x'. The extra division is the standard trap.
Real-World Applications
Trajectories are parametric by nature, position depends on time, so tangent directions computed this way are headings. A projectile's instantaneous flight angle is arctan(y'/x'); a robot arm's end-effector approach direction, a CNC cutter's instantaneous feed direction, and a car's heading along a GPS track are all y'/x' evaluated on a parametric path. Animation and game engines orient sprites and cameras along curves by exactly this ratio, the "look along the path" feature is a parametric derivative.
In design and manufacturing, CAD curves are Bézier and spline parameterizations; joining two curve segments smoothly requires matching their parametric derivatives at the seam (tangent continuity), the condition behind every "smooth" surface on a car body or product shell. Road and rail engineering matches tangent directions where straightaways meet curves, and matches the rate tangents turn (via the second derivative) to eliminate steering jerk, which is why transition spirals exist between straight track and circular curves.
Frequently Asked Questions
Why does dividing the two derivatives give the slope?
It's the chain rule rearranged: dy/dt = (dy/dx)·(dx/dt), so dividing by dx/dt isolates dy/dx. Equivalently, in a tiny time dt the point moves (x'dt, y'dt), and rise over run is y'dt / x'dt, the dt cancels.
What happens at a cusp, where x' = y' = 0?
The velocity vector vanishes, the moving point momentarily stops, and the slope formula reads 0/0. The tangent direction, if one exists, comes from the limit of y'/x' as t approaches the cusp (L'Hôpital or series). The cycloid's ground-touching points are the classic case: the limit gives vertical tangents.
How do I find horizontal and vertical tangents of a parametric curve?
Horizontal: solve y'(t) = 0 (with x' ≠ 0). Vertical: solve x'(t) = 0 (with y' ≠ 0). For the example cubic, y' = 0 at t = ±1 gives the loop's flat top and bottom; x' = 0 at t = 0 gives the vertical tangent at its leftmost point.
Can I always eliminate t and use ordinary differentiation instead?
Sometimes (the parabola chip: t = x^(1/2)… already awkward), but often elimination is impossible or destroys information, the self-crossing example would need a multivalued "function." The parametric formula works uniformly, which is why it's the standard tool.
Is this the same slope a polar curve formula computes?
Yes, a polar curve is the parameterization x = r(θ)cos θ, y = r(θ)sin θ, and pushing it through y'/x' produces the polar slope formula in the Polar Derivative Calculator. Parametric differentiation is the parent method; polar is a pre-packaged special case.
Does the direction of traversal change the slope?
No, reversing the parameterization negates both y' and x', and the ratio's signs cancel. The tangent line is a geometric property of the curve at the point; which way the point happens to be traveling along it is extra kinematic information the slope doesn't record.
What does concavity mean for a parametric curve?
The same as ever, the sign of d²y/dx², but computed with the two-stage recipe: differentiate the slope function with respect to t, then divide by x'(t). Where the result is positive the path bends upward as x advances; the loop of the example cubic switches concavity exactly where that expression changes sign.