Skip to main content

Subsection 8.9.2 Reviewing the Idea

So far, we’ve shown straightforward examples of induction proofs of claims about integer arithmetic. But induction can do way more than that. Recall the basic idea. We’ve shown it again in this rock climbing picture. If we can get started, and if, having made one step, we can always make the next one, we can show that we can make any number of steps.

Big Idea

Mathematical induction is a powerful tool for proving that some property holds of natural numbers or anything that can be characterized by them.

Prove that, for n  1, if the plane is cut by n distinct lines, the interiors of the regions bounded by the lines can be colored with red and blue so that no two regions sharing a common line segment as a boundary will be colored identically.

Here’s an example that illustrates the idea. In this case, n = 5. To see whether you believe the claim, try adding one more line to the figure. Can you recolor it as required?

Here’s what happened when we did that:

To prove the claim, we start by clearly articulating P(n):

P(n)  If the plane is cut by n distinct lines, the interiors of the regions bounded by the lines can be colored with red and blue so that no two regions sharing a common line segment as a boundary will be colored identically.

Basis step: P(1) is true since, if the plane cut by one line, two regions are formed. One may be colored red and the other blue. Thus the two regions are colored differently.

Inductive step: For n  1, we prove P(n)  P(n+1).

Given the plane cut by n+1 distinct lines, select any one line and remove it. The plane is then cut by n distinct lines. By the inductive hypothesis, the interiors of the regions bounded by the lines can be colored with red and blue so that no two regions sharing a common line segment as a boundary will be colored identically.

Now reintroduce the n+1st line. Choose one side of that line. Reverse the color of all regions on the chosen side.

Consider any line segment s that corresponds to a boundary of some region we’ll call r. Then s must lie either on the n+1st line or one of the other n lines but not both since the lines are distinct.

  • If s lies on the n+1st line, then the n+1st line has cut across what was a larger region, which was thus divided into r and some other region t. The larger region previously had a single color. But the color on one side (but not the other) of the n+1st line (and thus s) has been flipped. So the two regions (r and t) on opposite sides of s must have different colors.

  • The other case is that s lies on one side or the other of the n+1st line. (It can’t cross the n+1st line because, if it did, it would be two separate boundary segments.) Since the regions on either side of s previously had different colors, they still do because, after the introduction of the n+1st line, either they were both unchanged or they were both reversed.

In both cases, the interiors of the regions bounded by the n+1st line can be colored with red and blue so that no two regions sharing a common line segment as a boundary will be colored identically.

So, by the principle of Mathematical Induction, our claim is true.

Exercises Exercises

1.

1. Prove by induction that, for n ≥ 3, the sum of the interior angles of a convex polygon of n vertices is (n – 2)⋅π. Recall that a convex polygon is a polygon every one of whose interior angles is less than π (180°).

Consider:

(A) (B)

(A) is convex. (B) isn’t, because of the angle labeled λ. It is an inside angle, but it is larger than 180°.

The first step in writing our proof should be a clear articulation of P(n). Write one.

The next step is the base case. What value of n should be used for the base case?

Now write your proof that P holds of the base case.

Now write your proof of the inductive step. You must prove:

For n ≥ 3, if P(n) then P(n+1).

Answer.
P(n)  The sum of the interior angles of a convex polygon of n vertices is (n – 2). The next step is the base case. What value of n should be used for the base case? Enter it in the box. We can have a box and we can check the answer, I think. The correct answer is 3. Now write your proof that P holds for the base case. Click here to see one: When clicked, should see: P(3) is true. A polygon with 3 vertices is a triangle and the sum of the interior angles of any triangle is  (180). So  = (3 – 2) = . Now write the inductive step of your proof. You must prove: For n  3, if P(n) then P(n+1). For a hint, click here. If clicked, should see: Consider any convex polygon of n+1 vertices. Label the vertices p1, p2, p3, …, pn, pn+1 in counterclockwise order. Draw the line segment p1pn. (Really do draw all of this. A picture helps.) What can you say about the two convex polygons that you’ve now got? If you’d like another hint, click here. If clicked, should see: Consider the drawing from the last hint. We have a convex polygon of n+1 vertices. Its vertices are labeled vertices p1, p2, p3, …, pn, pn+1 in counterclockwise order. We’ve drawn the line segment p1pn. Notice that the original polygon of n+1 vertices is the union of the convex polygon p1, p2, p3, …, pn, with vertices, and the triangle with vertices p1, pn, and pn+1. What do we know about the sum of the interior angles of a polygon with vertices? What do we know about the sum of the interior angles of a triangle? When you’re ready to see a complete proof, click here. If clicked, should see: The proof is by induction on n. Consider any convex polygon of n+1 vertices. Label the vertices p1, p2, p3, …, pn, pn+1 in counterclockwise order. Draw the line segment p1pn. Notice that the original polygon of n+1 vertices is the union of the convex polygon p1, p2, p3, …, pn, with vertices, and the triangle with vertices p1, pn, and pn+1. By the inductive hypothesis, the polygon p1, p2, p3, …, pn, with vertices has a sum of interior angles equal to (n - 2). The triangle with vertices p1, pn, and pn+1 has (as do all triangles) a sum of interior angles equal to . Thus the sum of all interior angles is: (n - 2) +  = (n - 1)  = ((n + 1) – 2)  So, P(n)  P(n+1). So, by the principle of Mathematical Induction, our claim is true.

2.

Prove that, for n ≥ 1, \((\sum_{i = 1}^{n}{\ (4i - 3)) = n(2n - 1)}\text{.}\)

Start by writing an explicit statement of P(n).

Base case. Prove it.

Induction step: Write out the specific claim that we must prove.

Now write out the proof.

Answer.
P(n)  (∑_(i=1)^n▒〖(4i-3))=n(2n-1)〗 Base case. Prove it. Click here to see a proof. When clicked, should see: For n = 1, we have: (∑_(i=1)^1▒〖(4i-3))=41-3=1=1(21-1)〗 Induction step: Write out the specific claim that we must prove. Click here to check it before moving on. When clicked, should see: For n  1, ((∑_(i=1)^n▒〖(4i-3))=n(2n-1))〗  ((∑_(i=1)^(n+1)▒〖(4i-3))=(n+1) (2(n+1)-1)〗) Now write out the proof. Click here if you need a hint. When clicked, should see: Write out an expression for ∑_(i=1)^(n+1)▒〖 (4i-3) 〗. Since this is a summation, can you write it as the sum of the first n terms plus the (n+1)st term? If you’re still stuck, click here for another hint. When clicked, should see: (∑_(i=1)^(n+1)▒〖 (4i-3))〗 = (∑_(i=1)^n▒(4i-3) ) + 4(n+1) – 3 Now see if you can manipulate the right hand side to get the result you need. Try writing out the expression n(2n – 1), substituting (n+1) for n. That’s what you need to derive. It helps to know where you’re going. When you’re ready to see a complete proof, click here. When clicked, should see: The proof of the claim is by induction on n: Base case: 1 is the base case: For n = 1, we have: ∑_(i=1)^1▒〖(4i-3)=41-3=1=1(21-1)〗 Inductive Step: Prove: For n  1, ((∑_(i=1)^n▒〖(4i-3))=n(2n-1))〗  ((∑_(i=1)^(n+1)▒〖(4i-3))=(n+1) (2(n+1)-1)〗) [1] ∑_(i=1)^(n+1)▒〖 (4i-3) 〗 = ∑_(i=1)^n▒(4i-3) + 4(n+1) – 3 [2] = n(2n-1) + 4(n+1) – 3 Using the induction hypothesis [3] = 2n2 - n + 4n + 4 - 3 [4] = 2n2 - n + 4n + 1 [5] = 2n2 + 4n + 2 – (n+1) We know we need the (n+1). [6] = 2(n+1)2 – (n+1) [7] = (n + 1) (2(n + 1) – 1) Factoring out (n + 1). Thus, by the Principle of Mathematical Induction: For n  1, ∑_(i=1)^n▒〖(4i-3)=n(2n-1)〗.

3.

Prove that, for n ≥ 1, the product of any n odd integers is odd.

We need to start by writing an explicit statement of P(n). Before we can do that, we need to define a notation for products. We’ll use this:

\(\prod_{i = 1}^{n}{f(i)}\) is the product, as i goes from 1 to n, of f(i).

Π is to multiplication what Σ is to addition.

So let oi be the ith odd number in some arbitrary list of odd numbers. (Note that they don’t have to come in order: 1, 3, 5, … . Our claim applies to any n odd numbers.) Then:

\(\prod_{i = 1}^{n}o_{i}\) is the product of the n odd numbers in the given list.

Using this notation, write a definition of P(n).

Base case. Prove it.

Induction step: Write out the specific claim that we must prove.

Write out the proof of the inductive step.

Answer.
P(n)  For any list of n odd numbers, (∏_(i=1)^n▒o_i ) is odd. Base case. Prove it. Click here to see a proof. When clicked, should see: For n = 1, we have just one odd integer. The product of a single odd integer must of course be odd. Induction step: Write out the specific claim that we must prove. Click here to check it before moving on. When clicked, should see: For n  1: ((∏_(i=1)^n▒〖o_i)〗 is odd)  ((∏_(i=1)^(n+1)▒〖o_i)〗 is odd) That this claim is true when n = 1 is trivial. To get past that, we need to establish that it is true when we actually have to multiply two numbers together. So, before we start to do the induction, let’s prove a lemma: Lemma: The product of exactly two odd integers must be odd. Proof of lemma: Recall that every odd integer can be represented as 2k+1 for some integer k. So let the two odd integers be 2i+1 and 2j+1, for some integers i and j. Their product is: (2i+1) (2j+1) = 4ij + 2j + 2i + 1 = 2(2ij + j + i) + 1 = 2k + 1 Letting k = (2ij + j + i). So this number is odd. Now we can continue the induction proof of our claim. Write out the proof of the inductive step. Click here if you need a hint. When clicked, should see: The product of (n+1) odd numbers is the product of n of them multiplied by the (n+1)st one. Use this as your starting point. If you’re still stuck, click here for another hint. When clicked, should see: We should write an algebraic expression of the product of n+1 odd integers: (∏_(i=1)^(n+1)▒〖o_i)〗 = (∏_(i=1)^n▒o_i ) (on+1) Now see if you can use the inductive hypothesis to say something about the expression on the right. When you’re ready to see a complete proof, click here. When clicked, should see: The proof of the claim is by induction on n: Base case: 1 is the base case: For n = 1, we have just one odd integer. The product of a single odd integer must of course be odd. Inductive step: Prove: For n  1: ((∏_(i=1)^n▒〖o_i) 〗is odd)  ((∏_(i=1)^(n+1)▒〖o_i)〗 is odd) (∏_(i=1)^(n+1)▒〖o_i)〗 = (∏_(i=1)^n▒o_i ) (on+1) The inductive hypothesis guarantees that (∏_(i=1)^n▒o_i ) is odd. And the last odd integer is, of course odd. So we have the product of two odd integers, which, by the lemma we just proved, must be odd. So, by the Principle of Mathematical Induction: For n  1, the product of any n odd integers is odd.

4.

This problem is interesting, but harder than most of the ones we’ve done. We suggest that you try it, but do not get worried if you’re not sure you’ve got it.

Define the harmonic numbers hi, for i ≥ 1 as:

hi = \(1 + \frac{1}{2} + \ \frac{1}{3} + \ldots + \ \frac{1}{i}\)

Or, using summation notation, we can say:

hi = \(\sum_{j = 1}^{i}\left( \frac{1}{j} \right)\)

These numbers have been studied for hundreds of years, as has the infinite harmonic series:

\(1 + \frac{1}{2} + \ \frac{1}{3} + \frac{1}{4} + \ \frac{1}{5} + \ldots\)

The harmonic series does not converge to any fixed value. As i grows, so does hi, although it does grow very slowly. In this problem, we’ll use mathematical induction to prove that this is so. To do that, we’ll prove a claim about the harmonic numbers hk, where k is a power of two. Specifically:

For any integer n ≥ 0, \(h_{2^{n}\ }\)≥ 1 + \(\frac{n}{2}\)

First, you may want to convince yourself that this claim is believable. Write out the values of \(h_{2^{0}}\text{,}\) \(h_{2^{1}}\text{,}\) and \(h_{2^{2}}\text{.}\)

Let n = 2. What is \(h_{2^{n}}\) (i.e., h4) to 2 decimal places?

To check our claim, what is 1 + \(\frac{n}{2}\) ? (n is still 2.)

So, at least in the case of n = 2, we have that \(h_{2^{n}\ }\)≥ 1 + \(\frac{n}{2}\text{.}\)

Now on to our proof. Write out an explicit statement of the proposition P that we are trying to prove.

Base case: Prove P(0).

Induction step: We need to prove:

(\(h_{2^{n}\ }\)≥ 1 + \(\frac{n}{2}\)) → (\(h_{2^{n + 1}\ }\)≥ 1 + \(\frac{n + 1}{2}\))

Write this proof.

Answer.
We need to prove something about h_(2^(n+1) ). So start by using the definition of the harmonic series to write an equation of the form: h_(2^(n+1) ) = _________________ You will need to use the … notation to do that. Once you’ve got the expression, see if you can use the induction hypothesis (and probably some algebra) to derive the desired result. If you need another hint, click here. When clicked, should see: We start with: h_(2^(n+1) ) = 1+1/2+ 1/3+⋯+ 1/2^n + 1/(2^n+1)+1/(2^n+2) +⋯+ 1/2^(n+1) Now see if you can use the definition of h_(2^n )to simplify this in preparation for using the induction hypothesis. If you need another hint, click here. When clicked, should see: We’ve already got: h_(2^(n+1) ) = 1+1/2+ 1/3+⋯+ 1/2^n + 1/(2^n+1)+1/(2^n+2) +⋯+ 1/2^(n+1) Using the definition of h_(2^n ), we can rewrite it as: h_(2^(n+1) ) = 〖 h〗_(2^n ) + 1/(2^n+1)+1/(2^n+2) +⋯+ 1/2^(n+1) Now see if you can use the induction hypothesis to turn this into an inequality. Then use algebra to complete the result. If you need another hint, click here. When clicked, should see: So far, we have: h_(2^(n+1) ) = 1+1/2+ 1/3+⋯+ 1/2^n + 1/(2^n+1)+1/(2^n+2) +⋯+ 1/2^(n+1) h_(2^(n+1) ) = 〖 h〗_(2^n ) + 1/(2^n+1)+1/(2^n+2) +⋯+ 1/2^(n+1) Using the induction hypothesis, we get: h_(2^(n+1) )  (1+n/2) + 1/(2^n+1)+1/(2^n+2) +⋯+ 1/2^(n+1) Now we need one more important insight. Look at the remaining terms in the summation. Suppose that n were 3. We’d have: 1/(9 )+ 1/(10 )+⋯+ 1/16. There would be exactly 2n, i.e., 8, terms. For any n, this is so. Further notice that each of those terms is greater than 1/2^(n+1) . Since we are working with an inequality, we can replace that summation with another expression that is less than it. So we can write: h_(2^(n+1) )  (1+n/2) +(2^n ) (1/2^(n+1) ) Now see if you can get the desired result in a couple more straightforward steps. When you’re ready to see a complete proof of the inductive step, click here. When clicked, should see: h_(2^(n+1) ) = 1+1/2+ 1/3+⋯+ 1/2^n + 1/(2^n+1)+1/(2^n+2) +⋯+ 1/2^(n+1) h_(2^(n+1) ) = 〖 h〗_(2^n ) + 1/(2^n+1)+1/(2^n+2) +⋯+ 1/2^(n+1) Using the induction hypothesis, we get: h_(2^(n+1) )  (1+n/2) + 1/(2^n+1)+1/(2^n+2) +⋯+ 1/2^(n+1) There are 2n, remaining terms, and each is greater than: 1/2^(n+1) . So we can write: h_(2^(n+1) )  (1+n/2) +(2^n ) (1/2^(n+1) )  (1+n/2) + 1/2  1+ (n+1)/2