1. Evaluating f' at b for the derivative of an inverse
the whole questionWhat not to write
“, so .”
What to write
“, so ; and .”
Why: The theorem pairs on with on . The number is a true slope of , at : another point.
MATH 203 Calculus I • Concordia University, Montreal
This sheet is not a summary of section 3.8 of Thomas' Calculus: you have the course notes. It answers one question only, what makes students lose marks on the derivatives of inverse functions and of logarithms in MATH 203 at Concordia University, and which precise gesture avoids each loss.
The rules of the chapter fit in four lines. The marks go elsewhere: in the point where is evaluated for an inverse, and in the ALGEBRA done before differentiating, a root turned into a fractional power, a denominator turned into a minus sign, a law of logarithms that does not exist. A scientific calculator is allowed on the exams, so exact answers come first and the calculator checks them.
The thread of the chapter
Every derivative of section 3.8 comes from undoing: the slope of at is the reciprocal of the slope of at the SWAPPED point , and undoes products, quotients and powers, so the laws of logarithms, the true ones only, are applied BEFORE differentiating, and every base other than leaves its constant .
Write the pair on your copy before any derivative. It is the one line that prevents evaluating at , a loss worth the whole question.
The base never disappears from a derivative: it survives as , negative when . Only has , which is why and have derivatives with no constant.
Each row reads from left to right: the assumptions, then the result. A red cell is not an answer, it is the finding that the form settles nothing and the instruction to rewrite it. Every case is followed by a worked example.
Read a line as: the expression of the first column, under the condition of the second, may be replaced by the third. The red lines are rewritings that students write and that do not exist.
| Expression | Condition | Rewrite as |
|---|---|---|
| Example: Derivative at : . | ||
| signs unknown | ||
| Example: At : derivative , where alone does not exist. | ||
| Example: Derivative , equal to at . | ||
| Example: , equal to at . | ||
| , | ||
| Example: A constant: its derivative is everywhere on its domain. | ||
| none | rule that does not exist | |
| Example: At : , while . Same form, other result: but . What to do: Keep the sum inside: , which is at . | ||
| none | rule that does not exist | |
| Example: , while . Same form, other result: , while . What to do: The true law is ; or differentiate the difference term by term. | ||
Every legal rewriting keeps the function and changes its form. Test any rewriting at one value of with the calculator before differentiating: a fake law fails the test at once.
These are the errors I correct most often in session. Each one costs marks on a paper, even when the reasoning behind it is right.
What not to write
“, so .”
What to write
“, so ; and .”
Why: The theorem pairs on with on . The number is a true slope of , at : another point.
What not to write
“, so its derivative is .”
What to write
“Chain rule, inner function : .”
Why: , another function. The laws split products, quotients and powers, never sums.
What not to write
“”, then evaluating at .
What to write
“ on the whole domain; .”
Why: needs AND . At both factors are negative and their product is : the function exists, the expansion does not.
What not to write
“” or “.”
What to write
“, so the derivative is , equal to at .”
Why: The base enters through the constant , never through a factor . Only the natural logarithm differentiates with no constant.
What not to write
“.”
What to write
“ with : , equal to at .”
Why: The power rule needs a constant EXPONENT; here the exponent is the variable. Rewrite if in doubt.
What not to write
“.”
What to write
“: the cube root is the power , and it sits in the denominator.”
Why: Two separate algebra facts, and . Each one, missed, changes every later line.
What not to write
“, so the slope at is .”
What to write
“: , equal to at .”
Why: On the branch where , , and the chain rule gives . The absolute value extends the domain; it never enters the derivative.
What not to write
“, so .”
What to write
“, so .”
Why: The student's right side is the logarithm of the PRODUCT. The method is for products, quotients and powers; a sum is differentiated term by term.
What not to write
“The car loses per year, so .”
What to write
“, so per year.”
Why: The relative rate of is , not . The is the loss over a whole year; the instantaneous rate must be larger to lose it.
Look at the shape of the expression, or at the verb of the question, before writing anything
If the derivative of at a point, no formula for → find with , check , answer
Example: , : ,
If of a product, a quotient, a power or a root → expand with the laws of logarithms first, then differentiate term by term
Example:
If or with a constant base → the rule for or , times or divided by
Example:
If in the base AND the exponent of the same power → , product rule, multiply by
Example: :
If a product or quotient of many powers → logarithmic differentiation on , then multiply by
Example: :
If a limit of the form → take , bring out , exponentiate
Example:
If a limit with a log or an exponential → read it as
Example:
No L'Hôpital's Rule in this chapter: every limit here is a derivative read backwards or a logarithm brought out. And needs no logarithm: is in two DIFFERENT powers, and the product rule is shorter.
A marker ticks steps. Here they are in order, with the concluding sentence expected word for word.
When to use it: A one-to-one function is given by a formula that cannot be inverted by algebra
Concluding sentence
“Since , the point is on the graph of , and .”
The trap: Evaluating at , or using the point for the tangent to : both swap the pair back.
Marking: Typically 1 mark for one-to-one, 2 for the pair (a, b), 1 for f'(a), 1 for the reciprocal and the tangent.
When to use it: A heavy product or quotient of powers, or x in the base and the exponent of one power
Concluding sentence
“For , . Differentiating, , so .”
The trap: Stopping at : that is a relative rate, not the slope.
Marking: Typically 1 mark for the logarithm and its domain, 2 for the expansion, 2 for the differentiation, 1 for the multiplication by y.
Five minutes of checking recover more marks than one more problem started in a hurry.
Check the pair before inverting
For an inverse, plug a into f: it must return b exactly. If it does not, the reciprocal is taken at the wrong point.
, : , so and the slope of at is .
Test a rewriting at one value
Before differentiating an expanded logarithm, compute the original and the expansion at one convenient x with the calculator. They must agree.
At : , but : the expansion is false.
The sign of the base
A base below 1 gives a negative ln a, so a decaying quantity has a negative derivative. A positive rate for a car losing value is a lost sign.
, so for every .
Compare with a difference quotient
A calculator cannot differentiate, but it can compute (f(a + 0.001) - f(a))/0.001, which must be close to your f'(a).
For at : the quotient is about , the formula gives .
Let for . Find the equation of the tangent line to the graph of at the point where .
The equation cannot be solved for by algebra. Every step must be justified as on a MATH 203 final.
Step 1
and are strictly increasing on , so is strictly increasing, hence one-to-one: exists.
Why
The theorem is about an inverse that exists. A sum of increasing functions settles it in one line, with no derivative test, which comes later in the course.
Step 2
Find with : try , . So , the pair is on and on .
Why
Inspection is the method: makes the first value to try. Writing the pair is the step that prevents evaluating at .
Step 3
, so .
Why
The hypothesis is part of the theorem, and checking it is part of the marks.
Step 4
.
Why
The reciprocal of the slope at the SWAPPED point. The wrong line is what the marker looks for first.
Step 5
Tangent at : .
Why
The point is , not . Check: at the line gives .
The conclusion, written out
“The tangent line to the graph of at is .”
The classic mistake on this problem: Using gives the slope , and using the point gives a line through the wrong point: two independent losses, and a copy can make both.
You do not need a formula for the inverse. Find the number a such that f(a) equals the given b, usually by trying simple values like 0 or 1. Check that f prime of a is not zero. The derivative of the inverse at b is then one over f prime of a. The most common mistake is to evaluate f prime at b instead of at a.
Because 2 to the x equals e to the power x times ln 2, and the chain rule brings down the constant ln 2. Every base other than e leaves its natural logarithm in the derivative. The power rule does not apply, since the exponent is the variable. The same constant appears for log base 2, but in the denominator.
No. The laws of logarithms turn the log of a product into a sum, the log of a quotient into a difference and the log of a power into a product. Nothing splits the log of a sum. To differentiate ln of x squared plus 9, keep it whole and use the chain rule: the answer is 2x over x squared plus 9.
It is required when the variable appears in both the base and the exponent of the same power, like x to the power 1 over x. It is also the fastest route for long products and quotients of powers. It is useless for sums, and for a product like 3 to the x times x cubed, where the product rule is shorter. Always multiply by y at the end.
The base tends to 1 but the exponent grows without bound, so the form tells you nothing. Take the natural log: you get ln of 1 plus x, divided by x, which is a difference quotient of ln at 1 and tends to 1. Since the log tends to 1, the expression itself tends to e to the power 1, which is e.
© Ahmed Squalli Houssaini. Revision sheet published at www.letuteurscientifique.ca/en/fiches/math203-derivatives-logarithms. Free for personal and classroom use; republishing it elsewhere requires written permission (legal notice).