Bounding techniques
This chapter rigorously introduces fundamental bounding techniques crucial for quantitative analysis in inequalities and estimation. Mastery of upper and lower bounds, range finding, and estimation with parameters is essential for solving complex problems and constitutes a frequently examined component of the CMI BS Hons curriculum.
---
Chapter Contents
|
| Topic |
|---|-------| | 1 | Upper bounds | | 2 | Lower bounds | | 3 | Range finding | | 4 | Estimation with parameters |---
We begin with Upper bounds.
Part 1: Upper bounds
Upper Bounds
Overview
Upper-bound questions ask you to show that an expression cannot exceed a certain value. In olympiad-style and exam-style algebra, the real challenge is not just to produce some bound, but to produce a sharp or natural one and justify it cleanly. These problems frequently rely on AM-GM, Cauchy-type thinking, completing the square, or simple monotonicity arguments. ---Learning Objectives
After studying this topic, you will be able to:
- Prove upper bounds using standard inequalities.
- Distinguish a valid upper bound from the least upper bound.
- Detect sharpness by checking equality cases.
- Rewrite expressions into forms where the maximum is visible.
- Use domain restrictions correctly in upper-bound problems.
Core Idea
A number is an upper bound for an expression on a domain if
for every allowed value of the variable.
Standard Upper-Bound Tools
If an expression can be written as
then
so
Hence the maximum value is .
Many upper bounds come from standard inequalities under conditions like:
- fixed sum
- fixed product
- bounded interval
Example:
If is fixed, then
This follows from
If lies in a bounded interval, then expressions like
can often be bounded directly.
For example, for real ,
To prove
it is often enough to show
Then use squares or standard inequalities on .
Sharpness
If you prove
and there exists some allowed value where , then is sharp.
If equality never occurs, then may still be an upper bound, but not the maximum.
Minimal Worked Examples
Example 1 Find the maximum value of Since , Equality occurs at . So the maximum value is . --- Example 2 Find the maximum value of Rewrite: Hence Equality occurs at . So the maximum is . ---Upper Bounds from AM-GM
AM-GM is often used indirectly.
If is fixed and , then
Reason:
so
hence
Common Patterns
- Quadratic opening downward:
- Product with fixed sum:
- Expressions on a bounded interval:
- Symmetric expressions:
reduce with substitutions or identities
Common Mistakes
- ❌ Proving an expression is less than something large and calling it the maximum
- ❌ Forgetting to check equality
- ❌ Using AM-GM without nonnegativity
- ❌ Losing signs while completing the square
CMI Strategy
- Try to rewrite the expression as a constant minus a square.
- If a sum is fixed, try to bound the product.
- If the domain is bounded, use it fully.
- Check whether the bound is attained.
- Separate “upper bound” from “maximum” in your final statement.
Practice Questions
:::question type="MCQ" question="The maximum value of for real is" options=["","","",""] answer="C" hint="Use ." solution="Since , we have Equality occurs at . Hence the maximum value is , so the correct option is ." ::: :::question type="NAT" question="For real , find the maximum value of ." answer="1" hint="Complete the square." solution="We write Since , Equality occurs at . Hence the maximum value is ." ::: :::question type="MSQ" question="Which of the following statements are true?" options=["If for all allowed values, then is an upper bound of ","If equality occurs in , then is a possible maximum value","For real , ","Every upper bound is the least upper bound"] answer="A,B,C" hint="Distinguish upper bound from best upper bound." solution="1. True by definition.Summary
- Upper bounds are often found by rewriting as a constant minus a nonnegative quantity.
- Completing the square is one of the cleanest methods.
- Fixed-sum problems often lead to product upper bounds.
- An upper bound is not automatically the exact maximum.
- Equality cases determine sharpness.
---
Proceeding to Lower bounds.
---
Part 2: Lower bounds
Lower Bounds
Overview
Lower-bound problems ask for a guaranteed minimum value of an expression. In exam settings, the goal is usually not brute-force optimization but identifying the right inequality, rewriting the expression into a nonnegative form, or using symmetry and equality conditions effectively. CMI-style questions often test whether you can see why a quantity cannot go below a certain value. ---Learning Objectives
After studying this topic, you will be able to:
- Recognize common techniques for proving lower bounds.
- Use nonnegative-square arguments to derive bounds.
- Apply AM-GM, Cauchy-type ideas, and basic algebraic manipulation carefully.
- Track equality cases correctly.
- Decide whether a claimed lower bound is sharp.
Core Idea
A number is called a lower bound of an expression if
for all values of the variables in the allowed domain.
A lower bound is called sharp or best possible if equality is achieved for some valid choice of variables.
Main Techniques
The most basic source of lower bounds is
From this we get:
The last inequality follows from
For nonnegative real numbers,
equivalently,
More generally, for positive numbers,
Other High-Value Sources of Lower Bounds
- For all real ,
- For positive ,
- For nonnegative ,
- For all real ,
- For positive ,
Equality Cases Matter
Typical equality cases:
- in , equality when
- in AM-GM for two variables, equality when
- in
,
equality when
A bound is incomplete if the equality case is ignored.
Lower Bound by Rearrangement
Suppose you think an expression should be at least . Then try to write
as a sum of squares or another nonnegative quantity.
This is one of the most reliable methods in algebraic inequalities.
Lower Bound by AM-GM
For positive :
-
-
These are often useful when the expression contains reciprocal terms.
Lower Bounds with Fixed Sum or Fixed Product
If a sum is fixed, then quadratic expressions often have bounds via square expansion.
Example:
If
,
then
So controlling gives a bound.
If a product is fixed, AM-GM or substitution often becomes natural.
Minimal Worked Examples
Example 1 Find a lower bound for Rewrite: So the least possible value is . --- Example 2 For positive , find a lower bound of By AM-GM, So the lower bound is , attained when ---Common Traps
- ❌ Using AM-GM when variables can be negative
- ❌ Giving a lower bound without checking whether it is attained
- ❌ Treating a lower bound as the minimum without domain check
- ❌ Expanding everything when completing the square would be cleaner
CMI Strategy
- Check domain first: are variables real, positive, or nonnegative?
- Ask whether completing the square is possible.
- Look for AM-GM if there are positive terms or reciprocal structure.
- Try to rewrite as a nonnegative expression.
- Always state where equality holds.
Practice Questions
:::question type="MCQ" question="For real , the least possible value of is" options=["","","",""] answer="B" hint="Complete the square." solution="We write Since , we get Equality holds at . Hence the correct option is ." ::: :::question type="NAT" question="For positive , find the least possible value of ." answer="2" hint="Use AM-GM." solution="By AM-GM, Equality holds when Hence the least possible value is ." ::: :::question type="MSQ" question="Which of the following are always true?" options=[" for all real "," for all positive "," for all nonnegative "," for all real "] answer="A,B,C" hint="Check the domain and source of each inequality." solution="1. True, from .Summary
- Lower bounds often come from nonnegative squares or AM-GM.
- Rewriting is usually more important than raw calculation.
- Equality conditions are part of the answer.
- Domain restrictions determine which tools are valid.
- A sharp lower bound is one that is actually attained.
---
Proceeding to Range finding.
---
Part 3: Range finding
Range Finding
Overview
Range finding is the art of determining all possible values a given expression can take under stated conditions. In exam problems, this is rarely a brute-force computation. The real task is to combine domain restrictions, algebraic rewriting, standard inequalities, and equality cases in a clean way. CMI-style questions often test whether you can convert a messy expression into one whose minimum or maximum is visible. ---Learning Objectives
After studying this topic, you will be able to:
- Identify the domain before attempting to find a range.
- Rewrite expressions into forms that reveal lower or upper bounds.
- Use squares, AM-GM, and substitutions to find exact ranges.
- Track equality cases correctly.
- Distinguish between a bound and the actual range.
Core Idea
The range of an expression or function is the set of all values it can attain as the variable runs over its domain.
Step 1: Always Find the Domain
Before finding a range, check where the expression is defined.
Typical restrictions:
- requires
- requires
- requires
- rational substitutions may introduce extra restrictions
Main Range-Finding Tools
If an expression can be written as
then
so the minimum value is .
Examples:
For positive numbers ,
This is especially useful for expressions like
If a repeated part appears, set
and rewrite the expression in terms of , while also finding the range of .
Example:
If , then complete the square:
Some high-frequency facts:
- for real ,
- for ,
- for real ,
Equality Cases Matter
If you prove
that only gives a lower bound.
To claim the minimum is exactly , you must also show that is possible.
Minimal Worked Examples
Example 1 Find the range of Complete the square: Since , we get Equality occurs at . So the range is --- Example 2 Find the range of for . By AM-GM, Equality occurs at . As , . So the range is ---Typical Patterns
- Quadratic:
- Reciprocal expressions:
- Square root expressions:
analyze domain first, then monotonicity or squaring
- Symmetric expressions:
use substitutions or standard inequalities
- Rational expressions:
sometimes solve for in terms of and force the discriminant to be nonnegative
Discriminant Method
For a rational or algebraic expression, set
and rearrange to get an equation in .
Then require that equation to have real solutions, so its discriminant must satisfy
This often produces the exact range of .
Common Mistakes
- ❌ Finding a bound without checking whether it is attained
- ❌ Ignoring the domain
- ❌ Using AM-GM when variables are not positive
- ❌ Claiming a minimum for a quadratic without completing the square correctly
CMI Strategy
- Write the domain first.
- Look for a square, AM-GM pattern, or substitution.
- Try to reduce the number of moving parts.
- Track when equality occurs.
- State the final range in interval form whenever possible.
Practice Questions
:::question type="MCQ" question="The range of for real is" options=["","","",""] answer="B" hint="Complete the square." solution="We write Since , the minimum value is , attained at . Hence the range is , so the correct option is ." ::: :::question type="NAT" question="For , find the minimum value of ." answer="4" hint="Use AM-GM." solution="By AM-GM, Equality occurs when so and since , we get . Hence the minimum value is ." ::: :::question type="MSQ" question="Which of the following statements are true?" options=["For real , ","For , ","The range of over all real is ","A lower bound is always the minimum"] answer="A,B" hint="Check equality cases carefully." solution="1. True, because a square is always nonnegative.Summary
- Range finding begins with the domain.
- Completing the square and AM-GM are the most common tools.
- A bound becomes an extremum only when equality is attainable.
- Substitution and discriminant methods are powerful in harder questions.
- A clean range argument is usually short but very precise.
---
Proceeding to Estimation with parameters.
---
Part 4: Estimation with parameters
Estimation with Parameters
Overview
Estimation with parameters means finding bounds or ranges for expressions that depend on one or more variables together with an extra parameter. The parameter may be fixed, chosen freely, or determined so that an inequality always holds. In exam-level problems, the real challenge is to understand how the parameter changes the strength of the estimate. ---Learning Objectives
After studying this topic, you will be able to:
- Estimate expressions containing parameters by isolating the parameter-dependent part.
- Find parameter values for which an inequality holds for all real numbers or all positive numbers.
- Use discriminant, AM-GM, and Cauchy-based arguments to handle parameter bounds.
- Separate the cases where the parameter is fixed from those where it must be optimized.
- Avoid incorrect conclusions caused by ignoring the domain.
Core Idea
A parameter is a constant whose value affects the shape or size of an expression.
Typical tasks are:
- find the least or greatest value of an expression in terms of a parameter
- find all parameter values for which an inequality is true
- choose the parameter so that a bound becomes best possible
Standard Strategy
When a parameter appears:
- decide whether the parameter is fixed or must be found
- isolate the variable part of the expression
- use a standard inequality or algebraic condition
- check equality cases and domain restrictions
Estimating Quadratic Expressions with a Parameter
If an expression has the form
then for a fixed parameter inside or , a lower bound can often be found by:
- completing the square, or
- using the discriminant condition for non-negativity
Completing the Square with Parameters
For expressions like
write
So the minimum value is
This is one of the most useful school-level estimation methods.
Parameter in Rational or Symmetric Bounds
If a problem asks for a constant such that
or
for all allowed values, then:
- guess the equality case
- test if the expression can be rewritten around that case
- choose the parameter to make the inequality sharp
Using AM-GM with Parameters
For positive variables,
When a parameter is present, choose and so that the parameter sits naturally inside one or both terms.
For example, to estimate
for ,
AM-GM gives
when .
Using Cauchy / Titu with Parameters
If
appears, and denominators are positive, then think of Titu-type lower bounds:
Minimal Worked Examples
Example 1 Find the minimum value of for fixed parameter . Complete the square: Since , the minimum value is --- Example 2 Find all real such that for all real . For non-negativity of a quadratic for all real , we need the discriminant to be non-positive: Hence the parameter range is . ---Equality Case as a Guide
When a parameter is to be chosen optimally, equality often indicates the best candidate.
If the guessed equality case is wrong, the parameter choice is usually not sharp.
Common Patterns
- find the minimum of an expression in terms of a parameter
- find all such that an inequality holds for all real
- estimate or similar forms
- determine the best constant in an inequality
- compare several parameter ranges using discriminant or square completion
Common Mistakes
- ❌ forgetting whether the parameter is fixed or variable
- ❌ using AM-GM when positivity is not guaranteed
- ❌ missing domain restrictions like
- ❌ forgetting to check equality cases
- ❌ using discriminant incorrectly when the leading coefficient is not positive
CMI Strategy
- Ask first: am I bounding in for fixed , or solving for ?
- For quadratics, complete the square before doing anything else.
- For “true for all real ”, think discriminant.
- For expressions like with , think AM-GM immediately.
- For the best constant, look for the equality case and then justify it.
Practice Questions
:::question type="MCQ" question="For fixed real parameter , the minimum value of is" options=["","","",""] answer="B" hint="Complete the square." solution="We write Since , the minimum value is Hence the correct option is ." ::: :::question type="NAT" question="Find the least value of for ." answer="6" hint="Use AM-GM." solution="For , by AM-GM, Equality occurs when Therefore the least value is ." ::: :::question type="MSQ" question="Which of the following statements are true?" options=["If for all real , then ","For , whenever ","The minimum value of is ","AM-GM can be applied to for all real without any sign conditions"] answer="A,B,C" hint="Use discriminant, AM-GM, and square completion." solution="1. True. The discriminant condition gives .Summary
- Parameter estimation is about understanding how constants affect bounds.
- Completing the square is the fastest tool for quadratic estimation.
- Discriminant is essential when an inequality must hold for all real numbers.
- AM-GM gives sharp lower bounds in positive-variable parameter problems.
- Equality cases usually reveal the best possible parameter or bound.
---
Chapter Summary
Definition of Bounds: Understand upper and lower bounds, and differentiate them from supremum (least upper bound) and infimum (greatest lower bound). The existence of a bound does not guarantee its attainment.
Techniques for Finding Bounds: Employ calculus (derivatives for extrema), algebraic manipulation (e.g., completing the square, factoring), and fundamental inequalities (e.g., AM-GM, Cauchy-Schwarz, Triangle Inequality, Jensen's Inequality).
Range Finding: Determine the set of all possible output values of a function or expression by systematically identifying its minimum and maximum achievable values within its domain.
Estimation with Parameters: Analyze how bounds are influenced by unknown parameters, often requiring consideration of parameter space, worst-case scenarios, or specific conditions for optimality.
Tightness of Bounds: Emphasize the importance of finding the tightest possible bounds, as these provide the most accurate and useful information for estimation, error analysis, and optimization.
Worst-Case/Best-Case Analysis: Bounding techniques are critical for establishing the extreme limits of a system's behavior, providing guarantees on performance or safety.
---
Chapter Review Questions
:::question type="MCQ" question="Consider the function for . What is its maximum value?" options=["", "", "", ""] answer="" hint="Rearrange the expression to form a quadratic equation in and analyze its discriminant for real solutions." solution="Let . Rearranging gives , which simplifies to . For real values of , the discriminant must be non-negative: . This simplifies to . Factoring the quadratic yields , so or . Since the parabola opens upwards, the inequality holds for . Thus, the maximum value of is ."
:::
:::question type="NAT" question="Given positive real numbers such that , what is the minimum value of ?" answer="9" hint="Consider applying the Cauchy-Schwarz Inequality or AM-HM Inequality." solution="By the Cauchy-Schwarz Inequality (or Titu's Lemma, a direct consequence):
Equality holds when . Thus, the minimum value is ."
:::
:::question type="MCQ" question="Consider the function for a real constant . If the least upper bound of is , what is the value of ?" options=["", "", "", ""] answer="" hint="Rewrite by adding and subtracting in the numerator, then analyze its behavior as varies." solution="We can rewrite the function as:
Let . Since , . So .
If : As (i.e., ), . As , . In this case, the maximum value (least upper bound) is .
If : As (i.e., ), . As , . In this case, the maximum value (least upper bound) is .
If : for all , so the least upper bound is .
Given that the least upper bound of is , we must have the first case, where , and thus ."
:::
---
What's Next?
This chapter on bounding techniques provides foundational tools for rigorous analysis. The principles learned here are indispensable for Inequalities, where the focus shifts to proving relationships between expressions, often by establishing bounds. Furthermore, these techniques are directly applied in Estimation to quantify uncertainty, determine error margins, and approximate complex values, forming the bedrock for robust problem-solving in mathematical finance.