The Second Derivative of a Parametric Curve
For a curve given parametrically by x(t) and y(t), the first derivative is the quotient dy/dx = y′(t)/x′(t). The second derivative requires more care than squaring anything: it is the t-derivative of dy/dx, divided again by x′(t):
d²y/dx² = [x′(t)y″(t) − y′(t)x″(t)] / [x′(t)]³
The cube in the denominator surprises everyone the first time, and it is not a typo: one factor of x′ comes from the quotient rule inside, and one more from converting the outer d/dt into d/dx. This calculator computes all four derivatives x′, y′, x″, y″ symbolically, assembles the formula with every number visible, reports concavity at your point, and plots the curve with the point marked.
How to Use the Parametric Second Derivative Calculator
Enter x(t), y(t), and a parameter value t₀. The steps show the first derivatives and dy/dx, then the second derivatives, the assembled formula, and the value with its concavity reading: positive means the curve bends upward (∪) at that point, negative downward (∩). The tool refuses points where x′(t) = 0, there the curve is momentarily vertical or stationary, dy/dx blows up, and concavity in the y-versus-x sense is not defined. For just the slope, the parametric derivative calculator is the lighter tool; for the bending radius, the curvature calculator continues the story.
Worked Example
Sanity-check with the parabola in disguise: x = t, y = t². Then x′ = 1, y′ = 2t, x″ = 0, y″ = 2, so
d²y/dx² = (1·2 − 2t·0) / 1³ = 2
exactly the second derivative of y = x², as it must be when the parametrization is trivial. The default curve x = t², y = t³ − 3t is more interesting: at t = 1.5, the formula gives d²y/dx² = (x′y″ − y′x″)/x′³ = (3·9 − 3.75·2)/27 = 0.7222, concave up on that branch, while the mirror branch at t = −1.5 bends the other way. One curve, opposite concavities, correctly separated by the parameter.
Where the Formula Comes From
Write s(t) = dy/dx = y′/x′, a function of t. The chain rule converts x-derivatives to t-derivatives: d²y/dx² = ds/dx = (ds/dt)/(dx/dt). The quotient rule expands ds/dt = (x′y″ − y′x″)/x′², and dividing once more by x′ produces the cube. The derivation explains two standing warnings. First, d²y/dx² ≠ y″(t)/x″(t): that plausible-looking shortcut skips the quotient rule entirely and is wrong except by coincidence. Second, the sign of the result can flip when x′ changes sign even though the numerator does not, because of the odd power in the denominator, which is how a curve traversed leftward reports its concavity consistently in the geometric y-versus-x sense.
Common Mistakes to Avoid
- Computing y″(t)/x″(t). The single most common error. The correct route differentiates the quotient y′/x′ and divides by x′ again; the cycloid chip gives different answers by the two routes, only one of which matches the curve's true bending.
- Squaring instead of cubing the denominator. The quotient rule contributes x′² and the outer conversion one more x′. Dropping the third power scales answers wrongly wherever |x′| ≠ 1.
- Evaluating at points with x′(t) = 0. The curve is vertical (or has a cusp) there; dy/dx and its derivative do not exist in the y = f(x) sense. Parametrize concavity through curvature instead at such points.
- Confusing parameter concavity with curve concavity. y″(t) > 0 says y accelerates in t, not that the curve is concave up in space. Only the full formula speaks about the curve's shape.
- Forgetting that concavity can differ between branches. A self-intersecting parametric curve passes the same (x, y) point with different t values and possibly different concavities; always attach the answer to t, not to the location.
Real-World Applications
Parametric curves are how motion and design actually describe shapes, and second derivatives are their quality control. In road and rail design, alignments are parametric, and d²y/dx² feeds the curvature that sets speed limits and superelevation. Computer-aided design lives on parametric splines: joining two Bézier segments smoothly requires matching not just tangents (first derivatives) but curvature, second-order data, at the joint, or the reflection line on a car body will kink visibly. Animation and robotics plan trajectories parametrically in time; the formula converts easing curves into spatial bending so paths look natural rather than mechanical. Projectile and orbit analysis reads concavity to distinguish the rising and falling character of a trajectory's arc. Wherever a shape is drawn by a parameter rather than a formula y = f(x), this quotient-of-quotients is the honest second derivative. The parametric arc length calculator and unit tangent vector calculator complete the differential-geometry toolkit for the same curves.
Frequently Asked Questions
Why isn't d²y/dx² just y″(t)/x″(t)?
Because the second derivative is the x-derivative of dy/dx, and dy/dx = y′/x′ is a quotient whose t-derivative needs the quotient rule. The naive ratio ignores how x′ itself changes with t; it agrees with the true value only in special cases such as x = t.
Where does the cube in the denominator come from?
Two sources multiply: differentiating y′/x′ contributes x′² from the quotient rule, and converting d/dt to d/dx divides by one more x′. Three factors total. On unit-speed parametrizations in x (x′ = 1) the cube is invisible, which is why the formula surprises when first met.
What does the sign of d²y/dx² tell me?
The curve's concavity at that parameter value, read in the ordinary y-versus-x sense: positive bends upward like a cup, negative downward like a cap. Zero flags a possible inflection of the traced curve, subject to the sign actually changing across the point.
What happens where x′(t) = 0?
The tangent is vertical (or the point is a cusp if y′ vanishes too), dy/dx is infinite, and the second derivative in this form does not exist. The geometry is still perfectly finite, and the curvature formula, which never divides by x′, is the right instrument there.
Can concavity differ at the same point of the curve?
Yes, when the curve visits the point more than once. Each visit has its own t, tangent, and bending; the default cubic curve crosses itself with one branch concave up and the other concave down. Answers must be indexed by parameter value, not position.
How does this relate to curvature?
Curvature packages the same second-order data without the vertical-tangent blind spot: κ = |x′y″ − y′x″|/(x′² + y′²)^(3/2). The numerator is identical; the denominator uses speed instead of x′ alone, making κ well-defined at every regular point regardless of orientation.
Does reversing the parameter direction change the answer?
No: replacing t by −t flips the signs of x′ and y′ and keeps x″, y″, and careful bookkeeping shows numerator and denominator flip together, leaving d²y/dx² unchanged. Concavity is a property of the traced shape, not of the travel direction, and the formula respects that.
How do I find inflection points of a parametric curve?
Solve x′y″ − y′x″ = 0 for t (the numerator vanishing) and confirm the expression changes sign across the solution while x′ ≠ 0. Those parameter values mark where the curve's bending switches sides, the parametric version of f″ = 0 with a sign change.
Why does my ellipse show different concavity at different t?
Because it genuinely bends toward its center everywhere: on the upper arc that reads as concave down, on the lower arc concave up, in the y-versus-x convention. The formula tracks the arc the parameter currently traces; the sign changes as the curve rounds the vertices where x′ = 0.
Is there a third-derivative version?
Yes, by iterating the same recipe: differentiate d²y/dx² in t and divide by x′ once more. The expressions grow quickly and are rarely needed; applications wanting higher-order shape data almost always switch to curvature and its arc-length derivatives instead, which stay geometrically meaningful.