Lecture from: 27.05.2024 | Video: Video ETHZ
Review: Partial Fraction Decomposition – Breaking Down Complex Fractions
Quick Recap: The Big Idea of Partial Fractions When we have a rational function, which is a ratio of two polynomials , and we want to integrate it, the strategy is to break it down into a sum of simpler fractions. These simpler pieces are ones we usually know how to integrate (like or , etc.).
Key Assumptions:
- The degree of the numerator is strictly less than the degree of the denominator . (If not, we do polynomial long division first to get a polynomial part plus a new fraction that does satisfy this).
- We assume is monic (its leading coefficient is 1). (If not, we can divide both and by that leading coefficient).
Case 1: “All roots of are real”
If all the roots of the denominator are real numbers, we can factor like this: Here:
- are the distinct real roots of .
- is the multiplicity of the root (how many times appears as a factor).
The Setup for Partial Fractions: The theory tells us that we can then write our original fraction as a sum of terms, where for each distinct root with multiplicity , we include terms with denominators : The are constants that we need to find. The inner sum means that if a root has multiplicity , we’ll have terms in our partial fraction expansion corresponding to that root. For example, if , we’d have terms for , , , and .
Clicker Question: Setting up a Partial Fraction Decomposition
What is the correct setup for the partial fraction decomposition of ? Here, .
The roots are:
- with multiplicity .
- with multiplicity .
So the setup is: (We need terms for and because of the factor).
Finding the Coefficients: Multiply both sides by the common denominator :
Now, we equate the coefficients of the powers of :
- Coefficient of (constant term): .
- Coefficient of : . Since , .
- Coefficient of : . Since , .
So, .
Integrating: (Using for the constant of integration to avoid confusion with from the coefficients).
Case 2: ” also has non-real (complex) roots”
What if the Denominator Doesn’t Factor Nicely into Real Linear Terms? If (our denominator polynomial with real coefficients) has complex roots, they always show up in conjugate pairs. For example, if is a root, then must also be a root. When we multiply the factors corresponding to these conjugate pairs, we get an irreducible quadratic with real coefficients: . This kind of quadratic (like or ) doesn’t have real roots.
If has non-real roots, they always appear in complex conjugate pairs (since is a real polynomial).
A pair of complex conjugate roots, say and (where ), corresponds to an irreducible quadratic factor in the real factorization of : This quadratic term has no real roots (its discriminant is negative).
So, the general real factorization of (monic) will look like: where:
- are the distinct irreducible quadratic factors, each with multiplicity .
- are the distinct real linear factors, each with multiplicity .
The General Setup for Partial Fractions (including complex roots)
The partial fraction decomposition for (with ) will now include:
- Terms for each real root with multiplicity :
- Terms for each irreducible quadratic factor with multiplicity : (Notice the numerator for quadratic factors is linear: ).
The overall setup is: The coefficients are found by multiplying by the common denominator and comparing coefficients of powers of , just like before. It leads to a system of linear equations.
How do we integrate terms like ?
First, we try to make the numerator related to the derivative of the “inside” of the quadratic part, or a constant. Write . (This is like completing the square for the linear term).
So the term becomes:
-
Integrating the first part: . Use substitution: . Then . The integral becomes , which we know how to integrate (it’s , leading to if , or if ).
-
Integrating the second part: . Use substitution: . Then . The integral becomes . The integral is from our earlier discussion (Example 5.2.7 (5)), which can be solved using reduction formulas (often involving integration by parts or other tricks) to eventually get arctangents and rational functions of .
It’s a systematic process, though it can be algebraically intensive!
Example: A Denominator with an Irreducible Quadratic
Let’s integrate . Here .
Roots: (real, multiplicity 1). The factor is an irreducible quadratic (like , so ). Multiplicity 1.
Setup: Multiply by : .
Compare coefficients:
- : .
- : .
- : . Since , .
So, .
- : Let . This is .
- .
- .
So, the integral is .
Integration of Convergent Series
The Big Question: Can we swap the order of limit and integration? That is, if , is it true that ?
Example: When Swapping Fails with Pointwise Convergence
Consider a sequence of “tent” functions .
Each forms a triangle with base and height . The area of this triangle is . So, for all . Therefore, .
Now, what about the pointwise limit of ? For any fixed , if is large enough such that , then . If , for all . So, for all as . The pointwise limit function is . Then .
We have . So, in this case, we cannot swap the limit and the integral. The problem here is that while pointwise, the “mass” of the function (its integral) doesn’t go to zero; it just gets squeezed into a narrower and taller spike. This is not uniform convergence.
Theorem: Swapping Limit and Integral with Uniform Convergence
Uniform Convergence Saves the Day! If a sequence of integrable functions converges uniformly to a function , then this “bad behavior” doesn’t happen. The limit function will also be integrable, and we can swap the limit and integral.
Let be a sequence of bounded, Riemann integrable functions such that converges uniformly to . Then is also bounded and Riemann integrable, and: In other words: “You can swap and under uniform convergence.” (Proof: see script. It uses the definition of uniform convergence and properties of integrals.)
Clicker Question: Pointwise Convergence and Integrability
Let be bounded, integrable, and pointwise as , where is bounded.
-
Is then integrable? No! Consider the sequence of functions where if is one of the first rational numbers in (in some fixed enumeration of ), and otherwise. Each is integrable (it’s non-zero at only finitely many points, so ). The pointwise limit is (the Dirichlet function). We know the Dirichlet function is not integrable. (Compare with Example 5.1.6).
-
If is integrable, does it then hold that ? No! See the “tent functions” example at the beginning of this note. There, pointwise, is integrable (), but .
-
If is integrable and , does uniformly? No! Consider on . Pointwise limit . This is integrable, . . . So, holds. However, does not converge uniformly to on .
-
If uniformly, is integrable and ? Yes! This is exactly what Theorem 5.5.1 states.
Application: Integration of Power Series Term by Term
Integrating Infinite Polynomials Piece by Piece Since power series converge uniformly on any closed interval strictly inside their radius of convergence, this theorem allows us to integrate a power series term by term, just like we do with finite polynomials.
Corollary: Term-by-Term Integration of Power Series
Let be a convergent power series with a positive radius of convergence . Then for any : The resulting series for the integral also has the same radius of convergence .
Example: The Series for (Value of the Alternating Harmonic Series)
We know the geometric series . This converges for , so for . Radius of convergence is .
Integrate from to , where : .
Using term-by-term integration:
This series for also has radius of convergence 1.
What happens at the boundary ?
The series becomes , which is the alternating harmonic series.
We know is continuous at (from the left), so .
Abel’s theorem on power series states that if a power series converges at an endpoint of its interval of convergence, then the function defined by the series is continuous up to that endpoint.
The alternating harmonic series converges (by Leibniz criterion).
Therefore, we can say:
Example: The Series for
We know . This converges for , so for . Radius .
Integrate from to , where : .
Using term-by-term integration:
This series for also has . At , the series is . This is Leibniz’s formula for .
Since is continuous at and the series converges at (by Leibniz criterion), we have:
Improper Integrals
Extending the Idea of Area: Unbounded Regions or Functions So far, we’ve only defined integrals where the interval is finite and the function is bounded on that interval. “Improper integrals” allow us to consider two new scenarios:
- Integrating over an unbounded interval (like , , or ).
- Integrating a function that becomes unbounded at one or more points within a finite interval.
Goal: Extension of the integral to unbounded domains or unbounded functions.
Example: How can we understand ?
Intuition: An antiderivative of is . So maybe ?
This looks promising!
Definition: Improper Integrals
Let be an interval, and be a function that is bounded and Riemann integrable on every compact (closed and bounded) subinterval . (Often, is simply continuous on ).
Let and (where can be and can be ).
-
Case 1: “J is open or unbounded to the right” (e.g., , , ). This means but (if is or ) or . is (improperly) integrable on if the limit exists in (i.e., is a finite real number). If it exists, we define .
-
Case 2: “J is open or unbounded to the left” (e.g., , , ). This means (if is or ) or , but . is (improperly) integrable on if the limit exists in . If it exists, we define .
-
Case 3: “J is open or unbounded on both sides” (e.g., , , ). This means and . is (improperly) integrable on if for some (any) , both improper integrals (as in Case 2) and (as in Case 1) exist in . In this case, we define: (The value of this improper integral is then independent of the choice of ).
Convergence/Divergence: If an improper integral exists in (i.e., is finite), we say the improper integral converges. Otherwise, we say it diverges.
Examples of Improper Integrals
-
: This is Case 1 with . . So, (it converges).
-
for . This is Case 1 with . Consider .
- If : . . So for , the integral diverges.
- If : .
Now we need to look at .
- If (i.e., ): . Integral diverges.
- If (i.e., ): . In this case, the limit of the integral is .
Conclusion for :
- Converges to if .
- Diverges if .
-
for . Here the potential problem is at if . This is Case 2 with . Consider .
- If : . . So for , the integral diverges.
- If : .
Now we need .
- If (i.e., ): . The limit of the integral is .
- If (i.e., ): . Integral diverges.
Conclusion for :
- Converges to if .
- Diverges if .
Continue here: 25 Integral Test for Series Convergence, Stirling’s Formula to Approximate Factorial, Summary