The Chain Rule
If is a composition (an “outside” function of an “inside” function), then
Differentiate the outside function, leaving the inside alone, then multiply by the derivative of the inside. In Leibniz notation with and :
Outside is , inside is .
Two layers: around around .
Tip: Work from the outermost layer inward and multiply a new factor for each layer. You are never “done” until you have taken the derivative of the innermost expression.
Combining the chain rule with product and quotient rules
Real problems layer rules together. Identify the top-level structure first (is it a product? a quotient? a composition?), apply that rule, and use the chain rule wherever a composition appears inside.
Top level is a product .
Implicit Differentiation
When a curve is defined by an equation like (you cannot easily write ), treat as a function of and differentiate both sides with respect to . Every time you differentiate a , the chain rule forces an extra factor of . Then solve algebraically for .
Differentiate both sides with respect to :
At the point the slope is , so the tangent line is
Differentiate; the right side needs the product rule.
Tip: A tangent line is horizontal where (numerator ) and vertical where is undefined (denominator ).
Derivatives of Inverse Functions
If is one-to-one and differentiable, its inverse satisfies
The slope of at is the reciprocal of the slope of at the corresponding point . To use it: find the input with (so ), then compute .
First find with : testing gives , so . Then
so
Tip: You never need a formula for itself. You only need one matched pair with and the value .
Derivatives of the Inverse Trig Functions
The “co-” partners just carry a minus sign. Combine each with the chain rule for a composite argument.
Argument :
Tip: and share the radical ; and share ; the “sec/csc” pair share .
Logarithmic Differentiation
For messy products, quotients, and powers---especially a variable base raised to a variable power like ---take the natural log of both sides, use log laws to break it apart, then differentiate implicitly:
Remember , , .
The ordinary power rule and exponential rule both fail here---the base and exponent vary.
Tip: Any expression of the form (function) must be handled with logarithms (or by rewriting as ). You cannot use the power rule when the exponent contains .
Second Derivatives via Implicit Differentiation
To get implicitly: first find , then differentiate that expression again with respect to (using the quotient/product/chain rules), and substitute your formula for wherever it appears. Finally simplify using the original equation.
From we get . Differentiate again with the quotient rule:
using in the last step.
Tip: Your formula may legally contain ; always substitute the earlier result so the final answer is in terms of and only.
Going Deeper: Advanced Techniques
For a triple composition, multiply one derivative factor per layer, outside to inside. For :
And for :
Let , so on . Differentiate implicitly:
The same trick gives : from , , so .
For : differentiate to get , so . Then
since .
Variable base and variable exponent:
Implicit differentiation with respect to time is exactly the engine behind related rates. A sphere with gives, differentiating both sides in ,
Every quantity that changes gets its own “” factor by the chain rule---the same bookkeeping as putting on each .
If and , then the point on becomes on , and
Geometrically, reflecting the graph across reflects each tangent line, inverting its slope.
Any power with a constant exponent still uses the chain rule: . Any power with a variable exponent can be rewritten to avoid logarithmic differentiation:
This matches the log-differentiation result and is a useful check.
For find the tangent at . Differentiating,
At : , so the tangent line is .
Formulas, Proofs & Tips
What it means. Differentiate terms with the chain rule, then solve for . An inverse's slope is the reciprocal.
Example. : differentiating, , so .
Why it works. is a function of , so every term carries a factor. For inverses, differentiating gives .
Tip. Reflecting a graph across swaps rise and run, which is exactly why the slope inverts.