Intuition The one core idea
Two quantities x and y can be tied together by an equation so that moving one forces the other to move. Implicit differentiation is just asking, "when x nudges forward by a hair, how far does y get dragged?" — and the answer to that question is the number d x d y .
This page assumes nothing . Every scratch of notation the parent note used is unpacked below, in the order you need it. If a symbol on the parent page looked like magic, it is decoded here.
A variable is a slot that can hold different number values. We name slots with letters — usually x for the number we choose , and y for a number that depends on our choice.
Picture a number line: x is a dot you can slide left or right. That sliding is the whole game — everything downstream is a reaction to x sliding.
A function is a rule that takes each input value of x and hands back exactly one output. We write y = f ( x ) , read "y equals f of x ": feed x into the machine f , out comes y .
Look at the figure: the input dot x on the bottom line feeds into a box labelled f , and an output dot y pops out on the left line. This picture is the reason we care about d x d y later — it shows y is chained to x , not free.
Definition Coordinate point
( a , b )
A point ( a , b ) is one dot on a flat sheet: go a steps right, then b steps up. The first number is the horizontal position, the second the vertical.
Definition Curve of an equation
The curve of an equation is the collection of all points ( x , y ) whose numbers make the equation true. For x 2 + y 2 = 25 it is the set of all dots exactly 5 away from the centre — a circle.
Notice in the figure: a vertical line cuts the circle in two places. That is the visual proof that a circle is not one function y = f ( x ) — a function may return only one y per x . This single picture is why the parent note needed a whole new technique.
Intuition Explicit vs implicit — the picture
Explicit y = f ( x ) : y sits alone, already solved. One clean output per input.
Implicit F ( x , y ) = 0 : x and y are stirred together in one pot; nobody is "solved for." The circle equation is implicit.
(Here F ( x , y ) is just a placeholder name for "one formula that eats both x and y " — it gets its own full definition in §7. For now read F ( x , y ) = 0 as "some mixed equation set equal to zero," like x 2 + y 2 − 25 = 0 .)
Δ x and Δ y
Δ (Greek capital delta) is shorthand for a change in . Δ x means "how much x moved," Δ y means "how much y moved as a result." They are plain subtractions: Δ x = x new − x old .
Picture two nearby dots on a curve. Slide right by Δ x ; the curve lifts you up by Δ y . The ratio Δ x Δ y is the steepness of the little step between them — rise over run.
A tangent line at a point on a curve is the straight line that just kisses the curve there — it touches at that one spot and, right around it, points the exact same direction the curve is heading. Think of a ball rolling along the inside of the curve: the flat floor it would momentarily rest on is the tangent.
In the figure, watch the amber secant lines: each joins our fixed white dot to a second dot further along. As the second dot slides in (the gap Δ x shrinks), the secant swings until it settles onto the single white tangent line. That settling motion is the limit, and its final steepness is the derivative. See Tangent and Normal Lines .
Definition Limit (informal)
A limit is the value a quantity homes in on as we squeeze something toward zero. Written lim Δ x → 0 , read "the value approached as Δ x shrinks to nothing." (See Chain Rule 's prerequisites for the full treatment.)
d x d y
The derivative is the slope Δ x Δ y after we let the two dots slide together until they touch:
d x d y = lim Δ x → 0 Δ x Δ y .
The letters d y and d x are "infinitely small" versions of Δ y and Δ x . The symbol answers exactly one question: "at this instant, how fast does y change per unit change in x ?" — and geometrically it is the steepness of the tangent line from section 4.
Intuition Why this tool and not just
Δ x Δ y ?
A plain ratio between two separate dots measures the average steepness across a gap. But a curve bends, so its steepness differs everywhere. Shrinking the gap to zero (the limit) is the only way to pin down the steepness at one exact spot — the slope of the tangent that just kisses the curve there.
Definition Notation zoo — all mean "derivative"
d x d y — Leibniz form, reads "d y over d x ."
y ′ — Lagrange form, read "y prime." Same object, shorter.
f ′ ( x ) — the derivative of the function f .
The parent note swaps between d x d y and y ′ freely; they are identical.
Intuition Why the power rule is true (the
x 2 picture)
Picture a square of side x ; its area is x 2 . Grow the side by a hair Δ x . Two thin new strips appear along two edges, each of area x ⋅ Δ x , plus a tiny corner Δ x 2 . So the area grows by 2 x Δ x plus a negligible corner. Growth-per-unit is Δ x 2 x Δ x = 2 x . The two strips are literally the 2 x . Same logic with a cube gives three slabs → 3 x 2 .
Intuition Why the constant rule is true
A constant is a flat horizontal line — no rise for any run. Steepness of a flat line is 0 . This is why d x d ( 25 ) = 0 in the circle example: the right-hand side never moves.
Intuition Why the chain rule is true (gear picture)
Imagine two gears in a chain: x turns y , and y turns g . If y spins d x d y times as fast as x , and g spins g ′ ( y ) times as fast as y , then g spins g ′ ( y ) ⋅ d x d y times as fast as x — the two speed-ratios simply multiply . That product is the toll every y pays. This is the single rule that makes implicit differentiation possible.
Intuition Why the product rule is true (rectangle picture)
Let a rectangle have width u and height v , so its area is uv . Nudge both: width grows by u ′ Δ x , height by v ′ Δ x . The area gains a strip along the bottom (v ⋅ u ′ Δ x ), a strip up the side (u ⋅ v ′ Δ x ), and a tiny negligible corner. Growth-per-unit is u ′ v + u v ′ — one strip from each side changing.
Common mistake "Inside changes at rate 1, so why the fuss?"
For d x d ( x 2 ) the inside is x , which changes at rate 1 with itself, so the toll is invisible (× 1 ). For d x d ( y 2 ) the inside is y , which changes at rate d x d y — the toll is not 1 , so it stays visible. Same chain rule, different toll.
F ( x , y ) and its partials
F ( x , y ) means "one formula that eats both x and y ." For the circle, F ( x , y ) = x 2 + y 2 − 25 , and the curve is where F = 0 .
F x = ∂ x ∂ F (read "partial F partial x ") = differentiate F treating y as a frozen constant .
F y = ∂ y ∂ F = differentiate treating x as frozen.
The curly ∂ warns "there's another variable I'm deliberately holding still."
Intuition What "freezing a variable" looks like
Picture F ( x , y ) as a landscape : height F over the flat ( x , y ) ground. To get F x you walk due east (only x changes, y pinned) and measure how steeply you climb. To get F y you walk due north (x pinned). Each partial is just an ordinary slope, but along one compass direction only — that is what the figure's two arrows show.
For F = x 2 + y 2 − 25 : freeze y , get F x = 2 x ; freeze x , get F y = 2 y . The parent's slick formula d x d y = − F y F x = − 2 y 2 x = − y x then falls out.
Definition The hypothesis that makes it legal —
F y = 0
The formula d x d y = − F y F x divides by F y , so it demands ==F y = 0 == at the point. The Implicit Function Theorem says exactly this: wherever F y = 0 , the tangled equation F ( x , y ) = 0 really does secretly define y as a smooth function of x near that point, and the slope formula is valid. Where F y = 0 , all bets are off — see the next box.
Intuition Division by zero has a geometric meaning
The circle slope is d x d y = − y x . At the top and bottom of the circle y = 0 , fine. But at the far left and right points ( ± 5 , 0 ) the denominator y = 0 — the formula explodes. This is not a mistake; it is the curve telling you its tangent there is vertical (straight up and down), which has no finite slope . A vertical line rises infinitely for zero horizontal run.
Definition Vertical tangent
A vertical tangent is a point where the curve is momentarily heading straight up/down, so d x d y is undefined (it "blows up to infinity"). At these points F y = 0 , so the Implicit Function Theorem's guarantee lapses — you often flip the roles and compute d y d x = − F x F y instead, which is finite there.
Common mistake Panicking at
d x d y = 0 − 5
Wrong reaction: "I made an error, division by zero!"
Right reaction: that answer is information — the tangent is vertical at that point. Report it as a vertical tangent, not as a failure. Check whether F y = 0 there to confirm.
sin , cos , tan
Trig functions turn an angle into a ratio on a circle. You only need that they have derivatives: d x d sin x = cos x , d x d cos x = − sin x . They appear in sin ( x y ) = x + y .
arcsin (inverse sine) — with its fenced domain, range, and derivative
arcsin x asks the reverse question of sin : "which angle has sine equal to x ?" Writing y = arcsin x is the same statement as sin y = x — that second form is implicit, which is exactly why implicit differentiation builds inverse-function derivatives.
Because sin only ever outputs values between − 1 and 1 , the input must obey ==x ∈ [ − 1 , 1 ] . And because infinitely many angles share a sine, we fence the answer to one branch: y ∈ [ − 2 π , 2 π ] ==. On that fenced range cos y ≥ 0 , which is precisely why cos y = + 1 − x 2 (positive root) when we differentiate. The result of that differentiation — the tool you actually carry — is
d x d arcsin x = 1 − x 2 1 .
See Derivatives of Inverse Functions .
ln (natural logarithm) — with its derivative
ln x asks "to what power must the special number e ≈ 2.718 be raised to get x ?" Its key algebra: ln ( a ⋅ b ) = ln a + ln b and ln ( a b ) = b ln a — it turns products and powers into sums . That is why logging both sides tames x x in Logarithmic Differentiation . Its derivative is the clean formula
d x d ln x = x 1 ,
and, since y is a function of x , the chain-rule version d x d ln y = y 1 ⋅ d x d y is what appears in the x x example.
The figure below draws the same dependency map three tiers deep: build the derivative (top), the four rules + partials (middle), the payoff technique and its uses (bottom). Tier 1 is load-bearing — knock out the chain rule and everything below collapses.
For readers who prefer the raw graph, the same structure in text form:
change delta x and delta y
tangent line kisses the curve
implicit equation F equals 0
partials Fx Fy with Fy not zero
tangents inverses log diff related rates
Test yourself — cover the right side and answer out loud.
What does y = f ( x ) say in plain words? Feed input x into rule f ; it returns exactly one output y .
Why is a circle not an explicit function? A vertical line hits it twice — two y -values for one x , which a function forbids.
What does Δ x mean? A change in x : x new − x old .
What is a tangent line, in one phrase? The straight line that just kisses the curve at a point, matching its direction there.
What question does d x d y answer? When x nudges by a hair, how fast does y change — i.e. the slope of the tangent.
Why do we take a limit instead of a plain ratio? To measure steepness at one exact point on a bending curve, not the average across a gap.
Why is the power rule d x d x 2 = 2 x true, in a picture? A square of side x gains two strips of area x Δ x when the side grows — the 2 x .
Why is the product rule true, in a picture? A rectangle uv gains one strip from each side changing: u ′ v + u v ′ .
State the chain rule for g ( y ) with y = y ( x ) . g ′ ( y ) ⋅ d x d y — two gear speed-ratios multiplied.
What does F x mean geometrically? The slope of the landscape F as you walk east, with y frozen.
What extra condition makes d x d y = − F y F x legal? F y = 0 at the point (the Implicit Function Theorem hypothesis; else you'd divide by zero).
What does d x d y = 0 − 5 at ( 5 , 0 ) on the circle mean? A vertical tangent — slope undefined/infinite, not an error.
What are the domain and range of arcsin x ? x ∈ [ − 1 , 1 ] and y ∈ [ − 2 π , 2 π ] , which forces cos y ≥ 0 .
State d x d arcsin x . State d x d ln x . x 1 .
Why does ln help with x x ? ln turns powers into products/sums: ln ( a b ) = b ln a , so a variable exponent becomes a plain product.
Chain Rule — the toll-collecting engine every d x d y passes through.
Product Rule — for mixed x y terms.
Derivatives of Inverse Functions — built from the implicit form sin y = x on its fenced range.
Logarithmic Differentiation — why ln tames variable exponents.
Implicit Function Theorem — the guarantee (needs F y = 0 ) that y ( x ) secretly exists.
Tangent and Normal Lines — where the slope d x d y gets used, including vertical tangents.
Related Rates — same chain-rule logic, differentiating in time.