Sometimes a curve is too tangled to write as y = f ( x ) y = f(x) y = f ( x ) . Instead we let a third variable — a parameter t t t (think: time) — drive both x x x and y y y :
x = x ( t ) , y = y ( t ) x = x(t), \qquad y = y(t) x = x ( t ) , y = y ( t )
As t t t ticks forward, the point ( x ( t ) , y ( t ) ) (x(t), y(t)) ( x ( t ) , y ( t )) traces the curve. We still want the slope of that curve in the x y xy x y -plane. The trick: ==both x x x and y y y change because t t t changes==, so we measure how fast each moves with t t t and divide .
Intuition Speed ratio = slope
Imagine a bug walking along the curve. In a tiny instant d t dt d t :
it moves d x = x ˙ d t dx = \dot x\, dt d x = x ˙ d t horizontally,
it moves d y = y ˙ d t dy = \dot y\, dt d y = y ˙ d t vertically.
The slope of its path is "rise over run" = d y d x = y ˙ d t x ˙ d t \dfrac{dy}{dx} = \dfrac{\dot y\, dt}{\dot x\, dt} d x d y = x ˙ d t y ˙ d t . The d t dt d t cancels — slope doesn't care how fast you walk, only the ratio of vertical to horizontal speed.
We start from the chain rule , which is the only machinery we need.
y y y depends on x x x , and x x x depends on t t t :
d y d t = d y d x ⋅ d x d t \frac{dy}{dt} = \frac{dy}{dx}\cdot\frac{dx}{dt} d t d y = d x d y ⋅ d t d x
This is just the chain rule applied to y y y as a function of t t t through x x x . Now solve for the thing we want, dividing both sides by d x d t \dfrac{dx}{dt} d t d x (allowed when d x d t ≠ 0 \dfrac{dx}{dt}\neq 0 d t d x = 0 ):
Common mistake Steel-man the seductive wrong answer
It feels right to write
d 2 y d x 2 = ? y ¨ x ¨ \frac{d^2y}{dx^2} \stackrel{?}{=} \frac{\ddot y}{\ddot x} d x 2 d 2 y = ? x ¨ y ¨
because the first derivative was y ˙ / x ˙ \dot y/\dot x y ˙ / x ˙ — symmetry suggests "do it again." This is wrong.
Why it's tempting: the pattern looks self-similar.
The fix: d 2 y d x 2 \dfrac{d^2y}{dx^2} d x 2 d 2 y means differentiate d y d x \dfrac{dy}{dx} d x d y with respect to x x x — not with respect to t t t . You must go back through t t t using the chain rule.
Derivation. Let y ′ ≡ d y d x y' \equiv \dfrac{dy}{dx} y ′ ≡ d x d y (a function of t t t ). By definition:
d 2 y d x 2 = d d x ( d y d x ) = d d x ( y ′ ) \frac{d^2y}{dx^2} = \frac{d}{dx}\!\left(\frac{dy}{dx}\right) = \frac{d}{dx}(y') d x 2 d 2 y = d x d ( d x d y ) = d x d ( y ′ )
But we only know how to differentiate things w.r.t. t t t . Use the same parametric trick on y ′ y' y ′ :
d d x ( y ′ ) = d ( y ′ ) / d t d x / d t \frac{d}{dx}(y') = \frac{\,d(y')/dt\,}{\,dx/dt\,} d x d ( y ′ ) = d x / d t d ( y ′ ) / d t
Now y ′ = y ˙ / x ˙ y' = \dot y/\dot x y ′ = y ˙ / x ˙ , so differentiate it w.r.t. t t t by the quotient rule:
d d t ( y ˙ x ˙ ) = y ¨ x ˙ − y ˙ x ¨ x ˙ 2 \frac{d}{dt}\!\left(\frac{\dot y}{\dot x}\right) = \frac{\ddot y\,\dot x - \dot y\,\ddot x}{\dot x^{2}} d t d ( x ˙ y ˙ ) = x ˙ 2 y ¨ x ˙ − y ˙ x ¨
Divide by x ˙ \dot x x ˙ :
Worked example 1 — A point on a circle
x = cos t , y = sin t x = \cos t,\; y = \sin t x = cos t , y = sin t .
x ˙ = − sin t \dot x = -\sin t x ˙ = − sin t , y ˙ = cos t \dot y = \cos t y ˙ = cos t . Why? differentiate each w.r.t. t t t .
d y d x = cos t − sin t = − cot t \dfrac{dy}{dx} = \dfrac{\cos t}{-\sin t} = -\cot t d x d y = − sin t cos t = − cot t . Why this step? slope = speed ratio.
Check: at t = π / 2 t = \pi/2 t = π /2 the point is ( 0 , 1 ) (0,1) ( 0 , 1 ) , slope = − cot ( π / 2 ) = 0 =-\cot(\pi/2)=0 = − cot ( π /2 ) = 0 — the top of the circle is flat. ✓
x ¨ = − cos t , y ¨ = − sin t \ddot x = -\cos t,\ \ddot y = -\sin t x ¨ = − cos t , y ¨ = − sin t .
d 2 y d x 2 = ( − sin t ) ( − sin t ) − ( cos t ) ( − cos t ) ( − sin t ) 3 = sin 2 t + cos 2 t − sin 3 t = − 1 sin 3 t \dfrac{d^2y}{dx^2} = \dfrac{(-\sin t)(-\sin t) - (\cos t)(-\cos t)}{(-\sin t)^3} = \dfrac{\sin^2 t + \cos^2 t}{-\sin^3 t} = \dfrac{-1}{\sin^3 t} d x 2 d 2 y = ( − sin t ) 3 ( − sin t ) ( − sin t ) − ( cos t ) ( − cos t ) = − sin 3 t sin 2 t + cos 2 t = sin 3 t − 1 .
Why this step? plug into the boxed formula; numerator simplifies by sin 2 + cos 2 = 1 \sin^2+\cos^2=1 sin 2 + cos 2 = 1 .
Worked example 2 — A parabola
y = x 2 y=x^2 y = x 2 in disguise
x = t , y = t 2 x = t,\; y = t^2 x = t , y = t 2 .
x ˙ = 1 , y ˙ = 2 t ⇒ d y d x = 2 t = 2 x \dot x = 1,\ \dot y = 2t \Rightarrow \dfrac{dy}{dx} = 2t = 2x x ˙ = 1 , y ˙ = 2 t ⇒ d x d y = 2 t = 2 x . ✓ (matches direct differentiation)
x ¨ = 0 , y ¨ = 2 ⇒ d 2 y d x 2 = ( 2 ) ( 1 ) − ( 2 t ) ( 0 ) 1 3 = 2 \ddot x = 0,\ \ddot y = 2 \Rightarrow \dfrac{d^2y}{dx^2} = \dfrac{(2)(1)-(2t)(0)}{1^3} = 2 x ¨ = 0 , y ¨ = 2 ⇒ d x 2 d 2 y = 1 3 ( 2 ) ( 1 ) − ( 2 t ) ( 0 ) = 2 . ✓
Why care? it confirms the formula reproduces the ordinary answer when x = t x=t x = t .
Worked example 3 — Cycloid-type, where
x ¨ ≠ 0 \ddot x\neq 0 x ¨ = 0 matters
x = t − sin t , y = 1 − cos t x = t - \sin t,\; y = 1 - \cos t x = t − sin t , y = 1 − cos t .
x ˙ = 1 − cos t , y ˙ = sin t ⇒ d y d x = sin t 1 − cos t \dot x = 1 - \cos t,\ \dot y = \sin t \Rightarrow \dfrac{dy}{dx} = \dfrac{\sin t}{1-\cos t} x ˙ = 1 − cos t , y ˙ = sin t ⇒ d x d y = 1 − cos t sin t .
x ¨ = sin t , y ¨ = cos t \ddot x = \sin t,\ \ddot y = \cos t x ¨ = sin t , y ¨ = cos t .
d 2 y d x 2 = cos t ( 1 − cos t ) − sin t ⋅ sin t ( 1 − cos t ) 3 = cos t − cos 2 t − sin 2 t ( 1 − cos t ) 3 = cos t − 1 ( 1 − cos t ) 3 = − 1 ( 1 − cos t ) 2 \dfrac{d^2y}{dx^2} = \dfrac{\cos t(1-\cos t) - \sin t\cdot\sin t}{(1-\cos t)^3} = \dfrac{\cos t - \cos^2 t - \sin^2 t}{(1-\cos t)^3} = \dfrac{\cos t - 1}{(1-\cos t)^3} = \dfrac{-1}{(1-\cos t)^2} d x 2 d 2 y = ( 1 − cos t ) 3 cos t ( 1 − cos t ) − sin t ⋅ sin t = ( 1 − cos t ) 3 cos t − cos 2 t − sin 2 t = ( 1 − cos t ) 3 cos t − 1 = ( 1 − cos t ) 2 − 1 .
Why this step? used cos 2 + sin 2 = 1 \cos^2+\sin^2=1 cos 2 + sin 2 = 1 , then cos t − 1 = − ( 1 − cos t ) \cos t - 1 = -(1-\cos t) cos t − 1 = − ( 1 − cos t ) cancels one factor. Here the x ¨ y ˙ \ddot x\, \dot y x ¨ y ˙ term is essential — drop it and you get a wrong answer.
Common mistake Other classic traps
Forgetting x ˙ \dot x x ˙ can be zero. Vertical tangents (e.g. circle at ( ± 1 , 0 ) (\pm1,0) ( ± 1 , 0 ) ) make d y / d x dy/dx d y / d x undefined — that's geometry, not an error.
Mixing up which derivative goes on top. It's always y ˙ / x ˙ \dot y/\dot x y ˙ / x ˙ (the variable you're solving for , over the variable you differentiate against ).
Using x ˙ 2 \dot x^2 x ˙ 2 instead of x ˙ 3 \dot x^3 x ˙ 3 in the 2nd derivative. Steel-man: you'd be using d d t ( y ′ ) \frac{d}{dt}(y') d t d ( y ′ ) and forgetting the last ÷ x ˙ \div\dot x ÷ x ˙ .
Recall Feynman: explain to a 12-year-old
Imagine driving a car. A GPS tells you your east–west speed and your north–south speed separately. How steep is your path on the map? Just divide north-speed by east-speed — that's the slope. You never needed an equation for the road! For "how the slope itself is bending" (the second derivative), you can't just divide the accelerations ; you have to track how the slope changes as you move east, which means dividing by your east-speed one extra time — that's the mysterious "cube" x ˙ 3 \dot x^3 x ˙ 3 .
Mnemonic Remember the formulas
First: "dot-y over dot-x ."
Second: "Top differentiated by bottom, minus bottom by top, all over x ˙ \dot x x ˙ cubed " — same as quotient rule but with an extra x ˙ \dot x x ˙ underneath. Chant: "cube on the bottom or you've been fooled."
What is d y d x \dfrac{dy}{dx} d x d y for parametric x ( t ) , y ( t ) x(t),y(t) x ( t ) , y ( t ) ? d y / d t d x / d t = y ˙ x ˙ \dfrac{dy/dt}{dx/dt} = \dfrac{\dot y}{\dot x} d x / d t d y / d t = x ˙ y ˙ , valid when
x ˙ ≠ 0 \dot x\neq 0 x ˙ = 0 .
Why does the parameter d t dt d t cancel in d y / d x dy/dx d y / d x ? Slope is a ratio of vertical to horizontal change; the common factor
d t dt d t divides out, so walking speed is irrelevant.
Write the parametric second derivative formula. d 2 y d x 2 = y ¨ x ˙ − y ˙ x ¨ x ˙ 3 \dfrac{d^2y}{dx^2} = \dfrac{\ddot y\,\dot x - \dot y\,\ddot x}{\dot x^{3}} d x 2 d 2 y = x ˙ 3 y ¨ x ˙ − y ˙ x ¨ .
Why is d 2 y d x 2 ≠ y ¨ x ¨ \dfrac{d^2y}{dx^2}\neq \dfrac{\ddot y}{\ddot x} d x 2 d 2 y = x ¨ y ¨ ? Because
d 2 y / d x 2 d^2y/dx^2 d 2 y / d x 2 means differentiating
d y / d x dy/dx d y / d x w.r.t.
x x x (through
t t t ), giving a quotient-rule result divided again by
x ˙ \dot x x ˙ .
Where does the power x ˙ 3 \dot x^3 x ˙ 3 come from? x ˙ 2 \dot x^2 x ˙ 2 from the quotient rule on
y ˙ / x ˙ \dot y/\dot x y ˙ / x ˙ , times one more
x ˙ \dot x x ˙ from the final
÷ d x / d t \div\, dx/dt ÷ d x / d t .
What happens geometrically when x ˙ = 0 \dot x = 0 x ˙ = 0 ? Vertical tangent;
d y / d x dy/dx d y / d x is undefined (slope
→ ∞ \to\infty → ∞ ).
For x = cos t , y = sin t x=\cos t,\,y=\sin t x = cos t , y = sin t , find d y / d x dy/dx d y / d x . General method to get d 2 y / d x 2 d^2y/dx^2 d 2 y / d x 2 from d y / d x dy/dx d y / d x ? Differentiate
d y / d x dy/dx d y / d x w.r.t.
t t t , then divide by
d x / d t dx/dt d x / d t .
Chain rule — the engine behind both formulas.
Quotient rule — used to differentiate y ˙ / x ˙ \dot y/\dot x y ˙ / x ˙ .
Implicit differentiation — alternative when no parameter is available.
Tangents and normals — slope d y / d x dy/dx d y / d x gives tangent lines on parametric curves.
Concavity and second derivative — sign of d 2 y / d x 2 d^2y/dx^2 d 2 y / d x 2 tells concave up/down.
Cycloid , Parametric curves — natural homes for this technique.
differentiate w.r.t. x not t
via chain rule and quotient rule
Parameter t drives x and y
Chain rule dy/dt = dy/dx times dx/dt
First derivative dy/dx = ydot/xdot
xdot not zero else vertical tangent
Tempting wrong ydotdot/xdotdot
ydotdot xdot minus ydot xdotdot over xdot cubed
Intuition Hinglish mein samjho
Dekho, kabhi-kabhi curve itna complicated hota hai ki use seedha y = f ( x ) y=f(x) y = f ( x ) ke form me likhna mushkil ho jaata hai. Tab hum ek teesra variable t t t (jaise time) le aate hain, aur bolte hain x = x ( t ) x=x(t) x = x ( t ) , y = y ( t ) y=y(t) y = y ( t ) . Jaise t t t aage badhta hai, point ( x , y ) (x,y) ( x , y ) curve par chalta jaata hai. Slope nikalne ke liye simple soch: ek chhote se time d t dt d t me x x x kitna badla aur y y y kitna badla — dono ka ratio hi slope hai. Isi liye d y d x = y ˙ x ˙ \frac{dy}{dx}=\frac{\dot y}{\dot x} d x d y = x ˙ y ˙ , kyunki d t dt d t upar-neeche cancel ho jaata hai.
Ab sabse bada trap: second derivative. Bahut log galti se d 2 y d x 2 = y ¨ x ¨ \frac{d^2y}{dx^2}=\frac{\ddot y}{\ddot x} d x 2 d 2 y = x ¨ y ¨ likh dete hain — galat hai! Reason: d 2 y d x 2 \frac{d^2y}{dx^2} d x 2 d 2 y ka matlab hai d y d x \frac{dy}{dx} d x d y ko fir se x x x ke respect me differentiate karna, t t t ke respect me nahi. Isliye pehle y ˙ x ˙ \frac{\dot y}{\dot x} x ˙ y ˙ ko quotient rule se t t t ke respect me differentiate karo, phir ek aur baar x ˙ \dot x x ˙ se divide karo. Result milta hai y ¨ x ˙ − y ˙ x ¨ x ˙ 3 \frac{\ddot y\,\dot x-\dot y\,\ddot x}{\dot x^{3}} x ˙ 3 y ¨ x ˙ − y ˙ x ¨ — neeche cube (x ˙ 3 \dot x^3 x ˙ 3 ) hota hai, square nahi. Yaad rakho: "cube na ho to galat ho."
Yeh kyun important hai? Circle, cycloid, projectile path — yeh sab natural roop se parametric form me aate hain. Tangent line, concavity (up/down bending), aur physics me velocity-acceleration sab yahi se nikalta hai. Practice ke liye circle (x = cos t , y = sin t x=\cos t,y=\sin t x = cos t , y = sin t ) khud derive karo — top par slope 0 0 0 aana chahiye, jo physically sahi hai kyunki wahan curve flat hai.