The AM-GM Inequality & Constrained Extremes
Master the Arithmetic Mean - Geometric Mean (AM-GM) inequality for constrained optimization without calculus: the fundamental duality between sums and products, equality condition requirements, the art of weighted term splitting, and guarding against out-of-domain phantom minima.
1. The Story & Mental Model: The Enclosed Field of Maximum Area
Consider an ancient farmer with a fixed length of fencing who wishes to enclose the largest possible rectangular grazing field. If the dimensions are length and width , the perimeter is fixed:
The farmer wishes to maximize the enclosed area, which is the product:
If the farmer makes the field long and narrow (), the area is a pathetic square meters. But if the farmer balances the dimensions equally into a perfect square (), the area surges to square meters!
Any inequality between the dimensions wastes perimeter while diminishing area. Symmetry is maximal efficiency.
This geometric intuition is formalized by one of the most powerful analytical engines in mathematics: the Arithmetic Mean – Geometric Mean (AM-GM) Inequality:
The Arithmetic Mean (the average sum) is always greater than or equal to the Geometric Mean (the average product). The two quantities achieve absolute equality if and only if all variables are completely identical:
In aptitude exams like CAT and IPMAT, calculus is often a clumsy, time-consuming instrument for optimization. AM-GM provides a swift algebraic scalpel that solves constrained maxima and minima in pure mental arithmetic.
2. The CAT Question Blueprint: How AM-GM Problems Are Disguised
Test setters design optimization questions around five classic blueprints:
-
The Pure Reciprocal Sum (Constant Product):
"Find the minimum value of for ."
The variable cancels cleanly in the product . The setter tests whether you can apply AM-GM directly: .
-
The Unbalanced Exponent Trap (Term Splitting):
"Find the minimum value of for ."
If you apply AM-GM to two terms, the product is , which is not a constant!
The setter tests whether you know how to split terms: divide into two equal halves: . Now the product of the three terms is purely numeric, unlocking the minimum via 3-variable AM-GM!
-
Constrained Product Maximization under Linear Sum:
"If for positive , find the maximum value of ."
Because has power 3 and has power 2, the setter expects you to partition into 3 equal pieces and into 2 equal pieces, enforcing equality across all 5 components.
-
Cyclic Multi-Variable Products:
"For positive reals , find the minimum value of ."
Applying AM-GM individually to each factor and multiplying the inequalities gives .
-
The Domain Boundary Ambush (The Phantom Minimum):
"Find the minimum value of for ."
AM-GM gives a minimum of at . But is outside the permitted domain ! The setter sets a trap for students who apply formulas without checking the equality feasibility condition.
3. Core Concepts, Formulae & Derivations
A. The AM-GM Inequality Statement
For any set of positive real numbers :
with equality holding if and only if:
B. The Dual Optimization Principles
-
Minimizing Sum under Constant Product:
If the product is a fixed constant, then:
The minimum sum is , achieved when every term equals .
-
Maximizing Product under Constant Sum:
If the sum is a fixed constant, then:
The maximum product is , achieved when every term equals .
C. The Art of Term Splitting (Weighted AM-GM)
To apply AM-GM effectively, the product of terms must be independent of the variables.
Rule 1: Reciprocal Splitting to Match Exponents
To minimize for :
We need the product of terms to cancel all powers of .
Since and , we must split into:
-
identical terms of the first kind:
-
identical terms of the second kind:
Total number of terms is .
The product becomes:
Rule 2: Power-Splitting for Product Maximization
To maximize subject to the linear constraint :
We split into equal parts:
We split into equal parts:
There are total terms summing to .
At the maximum product, all terms must be equal:
This master shortcut solves multi-variable constrained optimization in 10 seconds!
4. Common Pitfalls & Speed Shortcuts
The Positivity Violation:
AM-GM holds strictly for positive real numbers ().
If a problem allows negative numbers, AM-GM cannot be used directly. For instance, the expression has no minimum over all real numbers because as , . Over , the minimum is . Over , the maximum is .
The Unbalanced Product Trap:
Never apply AM-GM to terms whose product still contains the variable!
Writing is mathematically true, but completely useless because is not a constant bound! You must split so the variable cancels completely.
The Phantom Equality Check:
Every time you calculate a bound via AM-GM, you must verify that the equality condition is achievable within the problem's domain:
- Does yield a real number?
- Does the resulting lie within the specified range (e.g., )?
If equality cannot be attained, the true minimum lies at the boundary!
Formulas
| AM-GM Inequality | |
|---|---|
| Two-Variable Reciprocal Sum Minimum | |
| Master Power-Splitting Ratio | |
| Reciprocal Exponent Splitting Rule |
Traps and tips
- Positivity check: AM-GM is strictly valid only when all terms are strictly positive (a_i > 0).
- Variable cancellation in product: You cannot claim an extreme value unless all variables cancel out into a pure numeric constant in the GM product.
- Weighted splitting shortcut: To maximize x^a * y^b under c1x + c2y = K, set c1x / a = c2y / b = K / (a + b).
- Check domain feasibility: If the equality condition requires x = 5 but domain is restricted to x >= 8, AM-GM minimum is an unattainable phantom; check boundary x = 8.
Solved examples
Example 6.1: Reciprocal Sum Optimization with Variable Coefficient
Find the minimum value of the real-valued function: for all strictly positive real numbers , and determine the exact value of at which this minimum occurs.
Show solution
Step 1: Check conditions for the AM-GM Inequality.
We are given . Therefore, both terms are strictly positive:
Step 2: Apply the 2-variable AM-GM inequality.
For two positive real numbers and :
Substitute and :
Notice that the variable cancels out completely in the product:
Multiply both sides by 2:
Thus, the theoretical lower bound of the function is 24.
Step 3: Determine the condition for equality.
Equality in AM-GM holds if and only if the terms are equal:
Multiply both sides by :
Since :
Step 4: Verify the minimum value.
Evaluate :
The lower bound is achieved.
Summary of Results:
-
Minimum value of is 24.
-
Achieved at .
Takeaway: For Ax + B/x with x 0, the minimum value is always 2 sqrt(AB), achieved when Ax = B/x = x = sqrt(B/A).
Example 6.2: Unbalanced Exponent Power-Splitting
Determine the absolute minimum value of the expression: for all strictly positive real numbers , and find the value of where the minimum is attained.
Show solution
Step 1: Diagnose why direct 2-variable AM-GM fails.
If we apply AM-GM directly to the two terms and :
The right-hand side depends on ! This is not a constant bound and therefore does not provide the minimum value.
Step 2: Balance the powers of via Term Splitting.
The first term has (degree 2 in numerator).
The second term has (degree 1 in denominator).
To eliminate in the product, we need the powers to cancel: .
Therefore, we must split into two equal terms:
Now rewrite the expression as the sum of three positive terms:
Step 3: Apply the 3-variable AM-GM inequality.
For positive terms :
Compute the product inside the cube root:
Multiply both sides by 3:
Thus, .
Step 4: Check the equality condition.
Equality holds if and only if all three terms are equal:
Check the other equality: (identically true).
Since , this value is valid.
Step 5: Verify the value.
The minimum value of is 12, attained at .
Takeaway: Split terms to ensure variables cancel in the geometric mean: x^2 + 16/x becomes x^2 + 8/x + 8/x, yielding minimum 3 (64)^1/3 = 12 at x = 2.
Example 6.3: Constrained Product Maximization via Weighted Splitting
Let and be strictly positive real numbers satisfying the linear constraint: Find the maximum possible value of the product: and determine the values of and that achieve this maximum.
Show solution
Step 1: Analyze the objective and constraint.
-
Objective: Maximize .
-
Constraint: with .
Notice the powers in the objective function: has power , and has power .
Total degree of the product .
Step 2: Partition the terms in the constraint.
To create the factors and in the geometric mean product, we must split:
-
The term into equal pieces:
-
The term into equal pieces:
Rewrite the sum constraint as terms:
Step 3: Apply 5-variable AM-GM to these terms.
Raise both sides to the 5th power:
Step 4: Find the maximizing values of and via equality.
Equality holds when all 5 individual components are equal:
From this single relation:
Verify the constraint:
Step 5: Compute the maximum value of .
In exact fractional form:
Summary of Results:
-
Maximizing coordinates: and .
-
Maximum product value: .
Takeaway: To maximize x^p y^q subject to c1 x + c2 y = K, split into p parts of (c1 x/p) and q parts of (c2 y/q); all p+q parts equal K / (p+q).
Example 6.4: Multi-Variable Cyclic Product Inequality
For all strictly positive real numbers , determine the minimum possible value of the cyclic ratio: and state the condition under which this minimum occurs.
Show solution
Step 1: Apply AM-GM to each individual factor in the numerator.
Since , we can apply the 2-variable AM-GM inequality to each pairwise sum:
-
For and :
Equality holds if and only if .
-
For and :
Equality holds if and only if .
-
For and :
Equality holds if and only if .
Step 2: Multiply the three inequalities together.
Since both sides of all three inequalities are strictly positive, we multiply them directly:
Step 3: Evaluate the ratio .
Divide both sides by :
Thus, the minimum value of the ratio is 8.
Step 4: Establish the equality condition.
For equality to hold in the product of inequalities, equality must hold simultaneously in all three component inequalities:
Verification:
Set :
The minimum value of the ratio is 8, occurring when .
Takeaway: Apply AM-GM separately to symmetric cyclic factors (a+b = 2 sqrt(ab)), then multiply inequalities; equality holds when all variables are equal.
Example 6.5: Boundary Validity Check and the Phantom Minimum Trap
A risk assessment model optimizes capital allocation according to the function: where statutory regulatory constraints restrict the capital multiplier to the domain .
(i) Explain why naive application of AM-GM yields an incorrect minimum.
(ii) Determine the true minimum value of on the restricted domain .
Show solution
Step 1: Analyze the naive AM-GM calculation.
If one ignores the domain restriction and applies 2-variable AM-GM directly:
According to AM-GM, the theoretical minimum is .
However, equality holds if and only if:
Step 2: Identify the critical domain conflict.
The problem explicitly restricts the domain to:
The required point of equality, , does not belong to the allowed domain !
Therefore, the value is a phantom minimum—a mathematical lower bound that can never be reached by any admissible value of .
Step 3: Analyze the monotonic behavior of on .
Let us analyze how behaves as increases beyond .
Examine the difference for :
For any :
Thus, for all .
This proves that is strictly increasing throughout the entire domain .
Step 4: Compute the true minimum at the boundary.
Since is strictly increasing on , its smallest value must occur at the left boundary point:
Evaluate :
Summary of Results:
-
Naive AM-GM gives 10 at , which is unattainable because .
-
True minimum on occurs at the boundary , with value .
Takeaway: Always check whether the AM-GM equality condition lies inside the problem domain; if the turning point is excluded, the true minimum lies at the domain boundary.
Example 6.6: Three-Variable Reciprocal Sum Minimization
Let be strictly positive real numbers satisfying the linear equation: Determine the absolute minimum value of the reciprocal expression: and determine the values of that yield this minimum.
Show solution
Step 1: Set up the product of the given sum and the objective expression.
Let and .
Notice the matching coefficients:
-
First term pair: and
-
Second term pair: and
-
Third term pair: and
Let us define new transformed variables:
Then the constraint becomes:
And the objective expression becomes:
Step 2: Apply the Cauchy-Schwarz / Bergström Inequality.
For positive real numbers :
Compute the right-hand side:
Substitute :
Divide both sides by 18:
Thus, the minimum value of is 2.
Step 3: Find the condition for equality.
Equality in Cauchy-Schwarz holds if and only if the ratios are equal:
Equivalently:
Then:
Substitute into :
Thus:
Step 4: Recover the original variables .
Recall :
Notice the stunning symmetry: !
Step 5: Verify the minimum value.
Constraint check:
The minimum value of is 2, attained when .
Takeaway: Rewrite sum reciprocal products using normalized variables u=a, v=2b, w=3c; applying (u+v+w)(1/u + 4/v + 9/w) = (1+2+3)^2 = 36 yields minimum 36/18 = 2 instantly.
Practice questions
Question 1 · CAT Level
Let and be positive real numbers satisfying . What is the maximum value of ?
Show answer and explanation
Answer: (C) 62,208
To maximize , split into 3 equal terms () and into 2 equal terms (), totaling 5 terms.
The sum of these 5 positive terms is .
By the AM-GM inequality on these 5 terms:
Raising both sides to the 5th power gives .
Dividing by 4 yields .
Equality occurs when all terms are equal: , giving .
Question 2 · CAT Level
Find the minimum value of the function:
for all strictly positive real numbers .
Show answer and explanation
Answer: (C) 12
To eliminate the variable in the geometric mean product, the powers in numerator and denominator must balance.
Split into two equal terms: , giving three positive terms: .
Their product is , which is a constant.
By the AM-GM inequality on these three terms:
Multiplying by 3 gives , with equality when .
Question 3 · CAT Level
Let and be positive real numbers such that . What is the minimum possible value of the product:
Show answer and explanation
Answer: (D) 64
Substitute into each factor: .
Alternatively, rewrite as by AM-GM.
Multiplying cyclically across all three terms gives .
Since , , and , their product gives .
Thus .
Equality holds when , where .
Question 4 · CAT Level
Find the minimum value of the expression:
for all real numbers .
Show answer and explanation
Answer: (C) 13
Shift the linear term to match the denominator by rewriting as :
Since , the term is strictly positive.
Apply the AM-GM inequality to the two positive terms and :
Thus , which implies .
Equality holds when .
Question 5 · CAT Level
Let and be strictly positive real numbers such that . What is the minimum possible value of:
Show answer and explanation
Answer: (B) 12
The exponents in the constrained product indicate how to partition the linear terms:
Split into 2 equal parts (), into 3 equal parts (), and into 1 part (), forming 6 positive terms.
The sum of these 6 terms is .
The product of these 6 terms is .
Applying the AM-GM inequality on these 6 terms:
Equality holds when all 6 terms are equal: , yielding .
Try real CAT questions
Study this concept with interactive practice in PrepSoc.
Open the concept library