Let's try to learn together the basis of calculus!

Proof by induction

About proof by induction

When we work with new concepts in mathematics, we often start from the basics and then add more and more tools in our toolbox. Or we add recipes to our cookbook, if you prefer. One thing very peculiar with sequences is that they are indexed by positive integers. This offers the possibility to use a proof technique called proof by induction. We use this proof in the following situations:

  • When we know that a property is true for a certain value to start (very often at n=0n=0 or n=1n=1), let’s call this value n0n_0.
  • When we know that if the property is true for a given value nn (ideally n>=n0n >= n_0), then it is also true for the next value n+1n+1. If the two conditions are satisfied, then we can conclude that the property is true for all values n>=n0n >= n_0.

(Note: For those who worked in programming, this is very similar concept to a recursion.)

Demonstration of the theorem

Goal

In Mathematics we start from some basic properties (called axioms) and then we prove other properties (called theorems), which afterwards we can use directly to prove other theorems or properties. It’s a bit like a craftmen who has to build tools, then with those tools he can build other tools and so on. As a maths practitioner, we have to remember all of the tools and choose which one is the best to use for a given problem.

The Proof

In order to prove the theorem, we will use a technique called proof by absurdity. This technique consists in assuming that the theorem is false and then showing that this leads to a contradiction. As the contradiction is not true, we can conclude that the theorem is true. (Because if it was false we would have a paradox).

Let’s restate the starting assumptions:

  1. The property is true for n0n_0.
  2. We know that if the property is true for any value nn bigger than n0n_0, then it is also true for n+1n+1. The theorem then states that the property is true for all values n>=n0n >= n_0. Let’s assume that the theorem is false. This means that there exists several values strictly bigger than n0n_0 such that the property is false.

Let’s take the smallest value nfn_f such that the property is false. Because we know that nfn_f is the smallest value such that the property is false, we know that nf1n_f-1 is true. We can apply the second property of this theorem, which states that if the property is true for nf1n_f-1, then it is also true for nf1+1=nfn_f-1+1=n_f. But this is a contradiction, because we assumed that the property was false for nfn_f.

So we can conclude that the theorem is true, there is indeed no value nfn_f strictly bigger than n0n_0 such that the property is false.

And here we are: we have made our first demonstration of this series! 🎉🎉🎉

Using the theorem

First example

Preamble

we will go very slowly in this first example. It is very important to understand the mechanics of a mathematics proof.

  • We write very carefully what we know, and what we want to prove.
  • Each time we want to use a theorem:
    • We have to recall the theorem’s necessary conditions
    • clearly show that they are satisfied, and then apply the theorem.
    • Recall the theorem’s conclusion and apply it to our case to make one step forwards
  • And we repeat the above until we reach our conclusion.

Let’s goooooooo

We will study a partical case of sequence, called arithmetic sequence. It is very often defined recursively as an+1=an+da_{n+1} = a_n + d, where dd is a constant. We want to find a direct expression for this sequence. In order to get the intuition for a direct expression, let’s look at the first few elements:

  • a0=a0a_0 = a_0 ok - thank you captain obvious
  • a1=a0+da_1 = a_0 + d
  • a2=a1+d=a0+d+d=a0+2da_2 = a_1 + d = a_0 + d + d = a_0 +2d
  • a3=a2+d=a2+d=a0+3da_3 = a_2 + d = a_2 + d = a_0 + 3d
  • a4=a3+d=a0+4da_4 = a_3 + d = a_0 + 4d

It seems that we can express the sequence as an=a0+nda_n = a_0 + nd.

Let’s try to prove this by induction.

  1. We know that the property is true for n=0n=0, because a0=a0+0da_0 = a_0 + 0d.
  2. Let’s try to prove that if the property is true for nn, then it is also true for n+1n+1.

We know that an=a0+nda_n = a_0 + nd. (That’s the assumption) Let’s try to compute an+1a_{n+1}:

  • an+1=an+da_{n+1} = a_n + d by definition of the sequence
  • But we know how to express ana_n with our assumption, so we can replace it:
  • an+1=(a0+nd)+da_{n+1} = \lparen a_0 + nd \rparen + d
  • an+1=a0+(n+1)da_{n+1} = a_0 + (n+1)d

And here we are, we have proved that the property is true for n+1n+1.

Let’s check again the hypothesis of the induction theorem:

  • If a property is true for a given value n0n_0 (we proved this above in 1.)
  • If a property is true for nn then it is also true for n+1n+1 (we proved this above in 2.)
  • Then the induction theorem states that the property is true for all values nn0n\geq n_0.

So we can conclude that the property is true for all values of n0 n \geq 0.

Exercises

  1. Try to prove again the induction theorem, without looking at the proof above. It will be a good exercise to understand the induction theorem and the basics mechanics of proof demonstration.

  2. Let’s look at another example, the geometric sequence. It is defined recursively as an+1=anra_{n+1} = a_n * r, where rr is a constant.

This is what happens when you have a saving account with a constant interest rate, and you want to see how much money you will have after nn years. Can you find a direct expression for this sequence? Can you prove that your expression is correct, using the induction theorem?

  1. Let’s look at another example, the sum of the first n integers. It is defined recursively as Sn=Sn1+nS_n = S_{n-1} + n, where S0=0S_0 = 0. The general formula for this is
Sn=n(n+1)2S_n=\frac{n * (n+1)}{2}

Try to compute the first 5 or 6 terms and see if the general expression is correct. Can you prove that this expression is correct, using the induction theorem?