Some students are not convinced that a proof by mathematical induction is a proof. I have given the analogy of dominoes toppling but still some remain unconvinced. Is there very convincing way of introducing mathematical induction? I need something which will have an impact. Are there any real life applications of induction?

Martin Sleziak
  • 50,316
  • 18
  • 169
  • 342
  • 2,726
  • 2
  • 27
  • 33
  • 16
    It's great that they are not convinced! It means you did a good job. What makes you think induction is generally valid beyond the game of moving around letters on a piece of paper, which are associated with this concept? (Please don't use induction in your answer.) – Nikolaj-K Jun 18 '13 at 12:01
  • 1
    Related: See [this question](http://math.stackexchange.com/questions/145189/examples-of-mathematical-induction), which asks for examples of mathematical induction that do *not* involve summations and such. You might have better luck with one of them. (E.g. because some of these seem more concrete and it may be more obvious to the students that you're generating new facts, not just assuming what you want to prove.) – ShreevatsaR Jun 18 '13 at 12:57
  • Is it really necessary to teach mathematical induction non-math majors? What would you expect them to *really* understand without going thru naive set theory and Peano's axioms first? – Andrea Mori Jun 18 '13 at 17:15
  • 4
    @AndreaMori Huh? Why would you need Peano's axioms to understand induction? – Jack M Jun 18 '13 at 19:51
  • 28
    Start with the base case of teaching one student (left as an exercise). Now suppose you can teach $n$ students. – Najib Idrissi Jun 18 '13 at 20:32
  • @JackM: Well, because the Induction Principle is actually one of the Peano's Axioms and characterizes the ordering property of the natural numbers. My impression is that it is hard to accept and intimately understand the working of proofs without some background on these ideas. – Andrea Mori Jun 18 '13 at 21:43
  • 6
    @AndreaMori Counting is also one of Peano's Axioms. Kindergartner's can count. PA only formalizes induction after you realize it's obviously true, just like every (traditional) axiomatic system. – Jack M Jun 18 '13 at 21:51
  • 1
    @JackM : "obviously true"??? it's an axiom! – Andrea Mori Jun 18 '13 at 21:54
  • @AndreaMori It's an axiom in PA. In the "axiom system" of common sense, it's a theorem. – Jack M Jun 18 '13 at 21:57
  • 13
    How do you get everyone in the school to know your secret? Tell one person. – Blue Jun 18 '13 at 23:03
  • 1
    @NickKidman: Unless you are willing to admit that language is nothing more than sounds according to some "game", utterly meaningless in any way, I certainly don't see the point of your comment. – Apostolos Jun 19 '13 at 12:22
  • @Apostolos: I can't think of anything which is truly meaningless. I can use language to tell the person next to me to hand me over a spoon because I want to eat my delicious cake, so there is the practical implication. Even if I think I live in the matrix and there is, in fact, no spoon - talking about spoons is practical and therefore language has meaning. The question is formulated as the problem of how to teach the students induction, with the implication that it's a valid principle. I think there is no conclusive argument for it being true, so it shouldn't be taught as some sort of fact. – Nikolaj-K Jun 19 '13 at 13:09
  • 2
    @NickKidman: The natural numbers are by definition the least infinite inductive set. By the fact that it is least, follows that induction is true on them (this requires no induction argument). Hence it is a valid principle and true for the natural numbers. It is also provable (again without any use of induction) that it holds in any well ordered set. For example the dominoes provably satisfy induction, in the sense that if the first domino falls and every domino that falls makes the next one fall, then all dominoes fall. – Apostolos Jun 19 '13 at 16:00
  • @Apostolos: I'm not arguing against the possibility to do math which most people will agree upon. Just against unreflectedly relating it to quantity in the real world, just because you can start to enumerate them. – Nikolaj-K Jun 19 '13 at 16:08
  • @NickKidman: This is why I mentioned the language analogy to begin with. I hope my analogy will become clearer now. I guess my choice of the world "meaningless" wasn't a good one: Are you willing to admit that we should not teach children the meanings of words as "some sort of fact"? Would it be "great" if a student offered nonsense text as his answer in a question? Are you against unrelfectedly relating language to things in the real world? Also, do you believe that not all dominoes will fall? Do you have any evidence of that? – Apostolos Jun 19 '13 at 16:31
  • @Apostolos: Why do you use the word "admit" here, am I in a defensive position? Why would I be against teaching the meaning of the word induction? There is a difference in the meaning (referent) of the word and the validity of the associated concept. And nonsense depends on the context. If "nonsense" is that which will result in the students get bad grades (because the concensus is that it's useless), then I guess it would mean to not point it out. And I don't believe I will ever see induction fail, but that doesn't mean it holds. But let's go the chat if you really want to discuss further. – Nikolaj-K Jun 19 '13 at 16:44
  • @AndreaMori Any student who studies modern logic can benefit from understanding mathematical induction. As an example, consider proofs of the deduction metatheorem or Zeman's proof of what he calls the semisubstitutivity of implication http://www.clas.ufl.edu/users/jzeman/modallogic/chapter02.htm. Or try proving that for if "p" is a wff, then p preceded by any natural number of negation symbols is also a wff. – Doug Spoonwood Jun 23 '13 at 03:13
  • 1
    See also [Dominoes and induction, or how does induction work?](http://math.stackexchange.com/questions/19485/dominoes-and-induction-or-how-does-induction-work) – bobobobo Jun 24 '13 at 20:09
  • @Blue that is a hilarious comparison :P – Mr Pie Mar 04 '20 at 03:10

29 Answers29


I'm surprised everyone seems to be in general agreement that the domino analogy is good. It seems poor to me: what have dominos got to do with proving things? It's a nice metaphor, but it has simply nothing to do with the issue at hand. I don't remember it having been helpful when I came across it before understanding induction, either. Maybe I'm a special case in that regard.

Which brings me to my answer: I would look for theorems that admit a particularly simple inductive proof. I would also relate those proofs to the students in a more conversational language. Ie, rather than:

First, prove it for 0. Then, prove that if it is true for n, it is true for n + 1. So it must be true for all n.

There are two points here that may be confusing:

  1. The notion of proving an "if then" statement. I suspect many students initially find it hard to see that an "if then" can even be considered a "statement" at all, people are more used to thinking of them as combinations of statements. The idea of proving "A if B", without even considering whether A is true, seems bizarre. Supposing I proved that if there were a unicorn, it would like tea.
  2. The use of a variable $n$ may be an unnecessary distraction for some students with low mathematical ability, many students remain uncomfortable with variables all the way through high school. This is admittedly a less important point.

To beat issue (1), work by example. State a simple theorem and propose to prove that it's true for all numbers (don't say "for all n"! Just point to the number in the statement, written on the board, and say "we'll prove it's true no matter what this is"). Prove it for zero. Then, without proving or stating the inductive step as a separate statement, go right ahead show "well because it's true for 0, it must be true for 1, because...". Then, "but because it's true for 1, we can use the same argument to show it must be true for 2, after all...". Repeat maybe once more. Finish up by pointing out that as the same argument works for any number, you can keep going as far as you need to and the statement is true for all numbers.

Imagine you had never been taught induction, that you were living in an age before abstract mathematics, and remember the principle meaning of the word proof. A proof is what you use to convince your fellow human that something must be true. If you were simply trying to do that, you wouldn't be fussing around with separating out the inductive step as a proposition in its own right, you'd use the much more easily grasped (and much more convincing!) sort of proof by example above. The illustration of why that means it's true for all $n$ is in the language of the proof itself: it's true because we can keep doing the thing we just did two or three times.

Jack M
  • 26,283
  • 6
  • 57
  • 113
  • 12
    +1 for showing (at least trying to) how and why the induction step appears, instead of postulating it and then moving on to convince the class that it works. – Rolazaro Azeveires Jun 18 '13 at 20:24
  • 5
    Yeah, most proofs by induction are awful that way. "Suppose this complicated formula is true for $n$..." What? *Why!?* – Jack M Jun 18 '13 at 21:52
  • 5
    I always disliked the dominoes analogy too, it seemed to me to miss the point. +1 for the explanation in the last two paragraphs, _that's_ the way to explain things! – Josh Chen Jun 18 '13 at 21:59
  • 3
    Do you think the well-ordering principle is intuitive? Why do we take it as an axiom? Why do we accept anything about natural numbers? I consider induction as something that we accept. (Not to object your opinion!) – Gil Jun 19 '13 at 02:02
  • 3
    @GilYoungCheong I never claimed all axioms are intuitive or that only intuitive facts should be taken as axioms. I only claimed that induction happens to be an example of an axiom that *is* intuitive. (Also, I think you meant to reply to my comments under the OP) – Jack M Jun 19 '13 at 02:18
  • @JackM I was just trying to defend domino analogy. Now I understand what you discern. – Gil Jun 19 '13 at 02:33
  • 1
    @Jack M: “Suppose this complicated formula is true for n...” – Touché! – reminiscent of the following joke: An chemist, a physicist, and a mathematician are stranded on an island when a can of food rolls ashore. The chemist and the physicist comes up with many ingenious ways to open the can. Then suddenly the mathematician gets a bright idea: "Assume we have a can opener ..." – Mike Jones Sep 02 '17 at 06:50
  • +1 I like your answer and especially helpful comments on how to teach. But I am sure that the example of toppling dominoes is a nice explanation for induction. It is not just a metaphor. – Cyriac Antony Apr 15 '19 at 11:19

I use the domino analogy myself, and I think it is the most useful I've seen.

As a different approach I also do the following, which is more difficult to understand, but does serve as a way of introducing certain other useful properties of systems of numbers. Consider the set of counterexamples: i.e. the set of positive integers $n$ such that $P(n)$ is false. If this set is non-empty, then it must have a smallest element. Why? At this point I usually find the youngest person in the crowd by running the "anyone younger than $x(\in\mathbb{N})$ years raise your hand"-algorithm once. After that you can pinpoint the exact point, where the induction had to fail but didn't. Then you can conclude that the contrapositive assumption that the set of counterexamples was non-empty is the culprit.

Of course:

  • You also need to have explained proof by contradiction.
  • You need to explain that no matter how large the crowd of counterexamples (even infinite!), the "raise your hand" -algorithm will terminate after finitely many steps. Nowadays many students have seen computers and algorithms in school, and actually seem to follow this - might not have worked nearly as well 40 years ago.
  • You can also immediately start modifying the process, and discuss, why/how "raise your hand" algorithm might fail, if A) the number of participating people is infinite, and B) we count their ages with infinite precision instead of in full years. Alternately (may be better?) you can return to this "memorable" example, when you discuss the finer points of the order relation of real numbers.

Do it in many ways. One explanation may work for some, another for somebody else. Some may not get the point of any of them, but them's the breaks.

Jyrki Lahtonen
  • 123,594
  • 24
  • 249
  • 599
  • 1
    I use this approach in freshman calculus, because I **will** get to completeness of the real number system a few weeks after doing induction, and thus need to discuss the non-existence of a smallest element in some sets of numbers. – Jyrki Lahtonen Jun 18 '13 at 12:46

I quite like the domino analogy.

The problem with teaching induction - this is from a UK point of view but it probably applies everywhere - is that there is a formal way of setting it out which they have to use, but only makes sense if you're familiar with the construction of the natural numbers.

In particular the formal inductive argument is roughly as follows.
1. There is some set $S = \{n\in\mathbb N : P(n) \text{ is true}\}$.
2. $0\in S$.
3. $k+1\in S$ for every $k\in S$.
4. $\mathbb N$ is defined as the smallest set such that 2. and 3. hold. Therefore $S=\mathbb N$.

But because induction is used so often it's enough to do steps two and three and write induction somewhere on the sheet. In the UK students get a little proof template which they have to learn, but you can pass the exam without understanding it.

A more intuitive explanation of induction is as a contradiction.

Suppose that there is some $n\in\mathbb N$ such that $P(n)$ is false then there must be a smallest $k$ such that $P(k)$ is false. We can show that $k\neq 0$ by checking that $P(0)$ is true, therefore we must have $P(k-1)$ is true, but $P(k)$ is false. But we have shown that $P(k-1)\Rightarrow P(k)$ which is a contradiction.

Notice that for this version is similar to Euclid's proof that there are infinitely many primes, you can even turn that one into an induction, with $P(k)$ the statement that there are at least $k$ primes. You might use this as an introduction.

You should try to get across the idea that induction is shorthand for a deeper argument, rather than present it as an argument in itself. I can't remember all the good examples (I'm not a teacher) but there are some examples of a formulae that are quite tricky to work out, but once you've got them the proof by induction is very quick. So induction is good for lazy people.

not all wrong
  • 15,712
  • 2
  • 32
  • 56
  • 5,299
  • 20
  • 31

Have you tried just showing how it works in special cases?

For example suppose you have proved the basis and induction steps for $P(n)$. Now we can show that $P(3)$ holds:

  • $P(0)$ holds because that's the basis step.
  • $P(1)$ holds because $P(0)$ holds and the induction step says that if $P(0)$ holds then $P(1)$ holds.
  • $P(2)$ holds because of the induction step and because $P(1)$ holds.
  • $P(3)$ holds because of the induction step and because $P(2)$ holds.

Do this for a few cases and you one should see that if you take any $n$ you can always "climb the ladder" from $0$ to $n$ this way.

  • 1,948
  • 10
  • 16

Maybe try a tiling proof by induction to demonstrate that the assumption of the inductive hypothesis is not "cheating".

My favourite is using induction to prove the tiling of a $2n\times 2n$ L-shape is possible by using $2\times 2$ L-shapes. You can motivate it by saying "look, the shape can be split into $4$ L-shapes of the next size down, if we knew we could tile those then we could tile the whole thing!".

  • 10,610
  • 1
  • 22
  • 37

I've taught a discrete math for computer science course for about two years now and use the analogy of "the wave" at sporting events:

  • Some person starts off the wave.
  • If the person to your left does the wave, you do the wave as well.

I usually have my students actually do the wave to get a sense for why induction works - something starts everything off (the base case), and as long as there is "energy" propagating forward, it will keep propagating forward (the inductive step). People love it.

To motivate complete induction, I have people do the following “mathematicalisthenics” exercise. I tell the first person in each row to stand up, then tell each person to stand up when everyone before them in their row is standing. People quickly get the idea and everyone ends up standing. I’ve found that this eliminates confusion as to why complete induction works, though of course students still need lots of practice to determine where and when to use complete induction.

Since this is a discrete math for computer science course, I often continue onward by talking about induction as a "machine." You start off with a proof that the result holds for 0. Then, you build a magic machine that takes as input a proof that the result holds for some number n, and it produces a proof that the result holds for some number n + 1. CS students are pretty good at getting the idea that you could just sit in a loop and feed in the proof for 0 to get a proof for 1, then a proof for 1 to get a proof for 2, etc.

Hope this helps!

  • 8,825
  • 3
  • 35
  • 77
  • 1
    Good point about "the wave" - and it reminds me of the phenomenon of the sudden freezing over of a super-cooled lake when a pebble is thrown into it. – Mike Jones Jul 13 '13 at 14:17

Use the analogy of dominoes, and then give the real-life example of the sinking of the Titanic: The management of the Titanic realized that the ship was doomed when they realized that the bulkhead that was being flooded would be completely flooded, and that when a given bulkhead was completely flooded, the next bulkhead would undergo the same fate.

Talk about impact...

Mike Jones
  • 4,310
  • 1
  • 35
  • 39

Here's what I've said in student handouts (for non-mathematicians, so hopefully this is accessible to doubting mathematicians too). It echoes some other answers, but the very slightly different presentation might be helpful for some readers.

Suppose we want to show that all natural numbers have some property $P.$ We obviously can't give separate proofs, one for each $n$, that $n$ has $P$, because that would be a never-ending task. So how can we proceed?

One route forward is to appeal to the principle of arithmetical induction. Suppose we can show that (i) $0$ has some property $P$, and also that (ii) if any given number has the property $P$ then so does the next: then we can infer that all numbers have property $P$.

Let's borrow some logical notation, use $\varphi$ for an expression attributing some property to numbers, and put the principle like this:


(i) $\varphi(0)$ and
(ii) $\forall n(\varphi(n) \to \varphi(n + 1))$,

we can infer $\forall n\varphi(n)$.

Why are arguments which appeal to this principle good arguments? Well, suppose we establish both the base case (i) and the induction step (ii). By (i) we have $\varphi(0)$. By (ii), $\varphi(0) \to \varphi(1)$. Hence we can infer $\varphi(1)$. By (ii) again, $\varphi(1) \to \varphi(2)$. Hence we can now infer $\varphi(2)$. Likewise, we can use another instance of (ii) to infer $\varphi(3)$. And so on and so forth, running as far as we like through the successors of $0$ (i.e. through the numbers that can be reached by starting from zero and repeatedly adding one). But the successors of $0$ are the only natural numbers. So for every natural number $n$, $\varphi(n)$.

The arithmetical induction principle is underwritten, then, by the basic structure of the number sequence, and in particular [this is the crucial point!] by the absence of 'stray' numbers that you can't get to step-by-step from zero by applying and reapplying the successor function.

Peter Smith
  • 49,532
  • 3
  • 58
  • 138

There is no better analogy; but some of your students may not actually have played with dominoes in this way, so it doesn't come home to them. I suggest bringing some (real) dominoes into the classroom. With a bit of hand-waving, you can explain that the line can be extended as far as you like. Only the ultrafinitists in your audience will doubt you.

John Bentin
  • 16,449
  • 3
  • 39
  • 64

I remember that I felt more convinced with mathematical induction more after seeing it being derived from the "well ordering principle" of the natural numbers. At that time, the "well ordering principle" was more intuitive for me than mathematical induction.

Perhaps your students will find the well ordering principle more natural than mathematical induction, and will get more convinced if they see it being derived from a possibly more "intuitive" concept.

  • 19,303
  • 5
  • 49
  • 116

From a layman's perspective, the two essential constituents of induction are a starting point and neighbourhood. With natural numbers, the starting point is proving P(1) is true, and the neighbourhood step is proving P(n) is true implies P(n+1) is true. Here is a simple application of induction:

If you have a line of coloured balls, how do you prove that they are all the same colour? You note the colour of the first ball and it is the same colour as itself, so P(1) is true. To check the rest of the balls, you simply make sure that any ball is the same colour as the previous ball. If all balls up to the previous were the same colour, then all balls up to the current are the same colour. If you can check that, you have shown that P(n) is true implies P(n+1) is true. All balls are now shown to be the same colour.

  • 201
  • 1
  • 2

I like to emphasize that the induction principle (or, more precisely, its contrapositive) is just expressing the idea that we can (in principle) get from 0 to any natural number by counting (in steps of 1, as usual). So if some statement is true about 0 but false about some other number $n$, then, by watching what happens to the statement as we count up to $n$, we will see the statement change from true to false at least once (maybe more often, if it changes back to true). So we get some $k$ such that the statement is true for $k$ but not for $k+1$.

Separately from this, I also follow the procedure in nonpop's answer, showing that, if a statement is true for 0 and preserved to successors then we can deduce it first for 1 and then for 2 and then for 3, etc.

Andreas Blass
  • 65,963
  • 4
  • 76
  • 146
  • 1
    I like this answer. Remembering OP is asking for something that leaves an impression, you should expand (in simple english) more on why induction is analogous to the idea that you can count to any given number – Cyriac Antony Apr 15 '19 at 11:06

My experience has been that students who grasp how induction is supposed to work generally have little trouble accepting that it does work; the real problem is getting them to understand the idea in the first place. First playing around a bit with simple recursively defined functions can be helpful.

I generally mention the domino analogy, but I tie it very closely to the actual workings of simple induction. That is, proving that $P(k)\to P(k+1)$ is analogous to making sure that the $k$-th domino is close enough to the $(k+1)$-st domino to knock it over. Then proving $P(0)$ knocks over domino $0$, which by virtue of $P(0)\to P(1)$ knocks over domino $1$, and so on. If we’ve done a little with logic, as we often have in a discrete math course, I might actually talk through the fact we’re repeatedly using modus ponens and actually derive $P(0)\to P(4)$, say. Without some such detailed explanation, it really isn’t (and shouldn’t be) at all convincing.

Time permitting, I also like to prove a few concrete cases of the induction step, starting with the first two or three after the base case. Then I’ll point out that the argument was pretty much the same each time and show that I can do exactly the same thing to get from $100$ to $101$, say. At that point most students are reasonably comfortable with the idea that I could fill in the gap to conclude that the $P(n)$ holds through $n=101$. Then we can usually go to the general case without losing too many students: we’re just ‘automating’ the process of getting from $0$ to $n$ for arbitrary $n$.

Whether they convince or not, these two approaches are useful in trying to get across the idea that a proof by induction is often just a legitimate way to say and so on, and they do convince quite a few students.

However, my main serious argument is that any proof by induction is just a proof that there is no minimal counterexample to the theorem and hence no counterexample at all (provided, of course, that we’re doing induction over a well-founded relation): I’ve found that almost all students readily accept that a non-empty set of integers that is bounded below has a least element. I make a point of presenting some induction arguments in those terms, too, though I also present some in more traditional format.

This point of view has the virtue of covering all kinds of induction: weak induction, strong induction, structural induction, and transfinite induction. It even covers some arguments that aren’t usually taught as proofs by induction, like the usual proof of irrationality of $\sqrt2$. Those who get it at all generally seem to find it quite convincing.

Brian M. Scott
  • 588,383
  • 52
  • 703
  • 1,170

How about giving an analogy related to genetics .Say if we can prove that if the parents are tall, so will be their children. Then if a person is tall, so will be his children, his grandsons and so on.

Martin Sleziak
  • 50,316
  • 18
  • 169
  • 342
  • 4,992
  • 18
  • 28

As another possible approach for these young skeptics, you might pose the question, what are the essential properties of the natural numbers? Steer the discussion to these fundamental points (or just present them):

(1) 1 (or 0, depending on preference) is a number.

(2) For every number, there is a unique next number, with "next" being a function. $next(1)=2, next(2)=3$, etc.

(3) If a number has a predecessor, it is unique.

(4) 1 has no predecessor.

Draw a diagram something like:

$1\rightarrow 2 \rightarrow 3 \rightarrow 4 \rightarrow ...$

Does this diagram fit all these of 4 essential features? Yes.

Then add 1 or more other points off to the side that are linked in a circular graph not connected to the first diagram, using letters as the nodes.

$a \rightarrow b$

$\uparrow \space \space \space \space \downarrow$

$d\leftarrow c$

Taken together with the previous diagram, the 4 essential features still fit. Somehow, we must introduce another property to eliminate the possibility of such side loops or chains. And this is precisely what the principle of induction does.

Suppose you have a subset $P$ of natural numbers. Suppose $1\in P$. Suppose that if $x\in P$, then $next(x)\in P$. Then argue that $P$ must contain every element of the chain $1 \rightarrow 2\rightarrow...$, but not any of the side loop. You can't get there from here (starting at 1). So, the last requirement is that we must have $P=N$.

Dan Christensen
  • 13,194
  • 2
  • 23
  • 44

In my experience, there are two main problems students have with induction:

  1. They cannot clearly write down the inductive hypothesis.

  2. They think that mathematical induction is a technique used (only) to prove summation formulas for series of numbers.

I have been teaching an algorithms class recently. I have seen many students who have in fact had a discrete mathematics course in which mathematical induction was covered, but who are completely baffled by inductive arguments applied to finite graphs. I spend a lot of time talking about the inductive hypothesis, and insisting that they write down explicitly what it is in their proofs. We work out lots of problems in elementary graph theory in class. Eventually, most of them do get the idea, but it seems to be a very subtle and sophisticated notion for students.

I'm particularly interested in this because these are all Computer Science students, and every one of them knows how to write a recursive function. And even though (from my point of view, at least) the understanding needed to write a recursive function is pretty much the same as that needed to write an inductive proof, it evidently doesn't seem that way to most of my students. I too would really like to understand better exactly what the difficulty is as seen by the students. I have some ideas, as I've indicated above, but I do have the feeling that I'm missing something big.

And by the way, I think that toy examples aren't useful. I can't imagine dominoes helping anyone understand how to write an inductive proof. But maybe I've misunderstood how that explanation is used?

Carl Offner
  • 287
  • 2
  • 4
  • 1
    A very belated comment: I think "Why do students have so much trouble with mathematical induction?" is perhaps the most interesting math education question I know, and I regard it as "unsolved". I agree that both of the difficulties you enunciate are difficulties, but I don't think your list is exhaustive. In fact I think that "understanding what mathematical induction means / why it works" is a more basic difficulty. The domino explanation does not help students write an induction proof, but it does help some students understand what induction means / why it works...to a degree. – Pete L. Clark Jul 24 '15 at 05:34

I like the "Stairway to Heaven" example:

Given an (semi-)infinite staircase from the ground up forever. It is promised that if any stair is white, then the stair above it is also white. Once you find the first (or any) white stair, then all the rest above it forever will also be white.

  • 237
  • 1
  • 5
  • What does semi-infinite mean? If it's going up from the ground forever, then it is going forever ad *infinitum* - it is infinite. Stopping at one step doesn't make the staircase *end* there, but just allows us to choose a point on that infinite staircase. It doesn't change its magnitude of infinity. I don't mean to be a nitpick, but often the little things arouse the most puzzlement. – Mr Pie Mar 04 '20 at 03:15
  • 1
    @MrPie "Semi-infinite" distinguishes the single-sided infiniteness of the Natural Numbers from the double-sided infiniteness of the Integers. It is the same distinction as between a ray and a line. – John Mar 04 '20 at 11:24
  • Oh. Well, I wouldn't call it "semi-infinite"... but then again... I see what you mean. Thanks for the explanation :) – Mr Pie Mar 04 '20 at 11:25

I was never "convinced" that mathematical induction was an absolute truth of universe, and it bothered me so much until I thought of it as an axiom of natural numbers. Just as some people were not comfortable accepting axiom of choice (and many magical proofs based on it), I don't think it is a good idea to find a way to "convince" students about it.

For students who want to understand "why" every single statement in mathematics is "true", I think it is important to understand that mathematics is not always about "truth" but it is rather about logical conclusions of certain axioms.

I feel comfortable writing proofs with induction, but I am still amazed about elementary proofs that use induction. Not only induction, but the existence of maximal element of finite linearly ordered set or arguments based on counting finite sets also fascinate me, but I do not try to make it "rigorous" by going deeper on those principles. These are beautiful principles, but that does not mean that we can "prove" why they work.

  • 911
  • 6
  • 15
  • Well, it is better thought of as a theorem rather than an axiom of natural numbers ... unless you think the same about the statement "every set of natural numbers has a minimum element" (which many people do) – Cyriac Antony Apr 15 '19 at 10:59
  • I upvoted this answer because I agree with you. Teachers should not be forcing students to agree with them. I think other countries should adopt the Finnish education system and use a student centered approach. I myself find the universal rule of induction dubious because it lets you prove the existence of $2\uparrow \uparrow 8$. Once you've shown that for all natural numbers $n$, $2^n$ exists, then after you prove that $2\uparrow \uparrow 7$ exists, using the fact that for all natural numbers $n$, $2^n$ exists, you can show that $2^{2\uparrow \uparrow 7}$ exists and is therefore – Timothy Aug 11 '19 at 03:34
  • $2\uparrow \uparrow 8$. However, I don't have a feel for the real meaning of it because I have not yet visualized $2\uparrow \uparrow 8$. However, I believe that if researchers used a very weak system of pure number theory to get as much information as they need to get jobs such as engineering done which you can be confident enough that all the properties describable in that system satisfy, it would still work well enough. It seems to be working so there is probably some underlying reason such as that weak proof system of pure number theory being reliable. When you create a stronger system by – Timothy Aug 11 '19 at 03:42
  • introducing a new property, you can't automatically assume that that property satisfies induction also. Some people think of the property of induction as a meaningful statement that all subsets of the natural numbers satisfy induction. One problem I have with that is that I find it dubious that the class of all subsets of the natural numbers even is a set and not a proper class. I like working in pure number theory better. I also believe that the fundamental laws of the universe are something like a Conway's game of life, yet can simulate a universe whose observations can be explained by – Timothy Aug 11 '19 at 03:46
  • calculus where it is consistent with our observations that time is continuous. I don't believe the real reasons for what we observe come from real number induction. That would be a very wierd complicated set of fundamental laws. – Timothy Aug 11 '19 at 03:51

You could modify the domino analogy by pointing out that, if some of the dominos are not properly set up, and wouldn't be knocked over by the previous one, it becomes no longer true that $\phi(k)\implies\phi(k+1)$, and hence the induction step will not be provable. You could also attack Zeno's paradox, first proving that if there's space between one $\sum_{k=0}^k\frac{1}{2^i}$ and $2$, there's space between $\sum_{k=0}^{k+1}\frac{1}{2^i}$ and $2$. A realworld application isn't as effective or to the point as a scenario they can vividly imagine, and Zeno situations are something many people have thought about and sought to logically establish.

Loki Clock
  • 2,113
  • 11
  • 13

I think the problem here is that mathematical induction may come across to some people as a bit abstract. The domino effect seems quite a decent example, except that there is a lot more explaining to do than just to say that one tile falls and others will follow.

The best way to make your students accept mathematical induction is to make them "suffer" through the pains of calculations. For example,

"Show that for each $n$, $\sum_{i=1}^{n}i=\frac{n(n+1)}{2}$."

and ask them to calculate by hand. It will be very tedious and not satisfactory, because you may have verified for n=10, but you still have a lot more verification to do. Yet this statement is true for all $n$!

You can then show them that one does not have to suffer by using a trick: splitting up the sums $\sum_{i=1}^{n-1}i+n$ and ask them to use the previous results that they have established, and show them that this trick works in general for all n.

Using this as an example, you can then show them the proof in general and how it leads to mathematical induction. A mathematical induction in short is a way of telling people how the machinery of calculation works (that is it is based on the result of the previous calculation)

  • 2,141
  • 16
  • 26

Tell your students that $\omega$ is the smallest inductive set

Edoardo Lanari
  • 3,474
  • 12
  • 27

You are in line at Taco Bell. The first person in line orders a cheesy gordita crunch.

Also, if the person in front of you orders a cheesy gordita crunch, you order it too, because it sounds so good.

How many people in line order cheesy gordita crunches?

Brian Rushton
  • 12,904
  • 11
  • 57
  • 93

I think that what your students really don't understand is the Axiom of Induction, one of the Peano Axioms, which defines the term Natural Numbers.

A proof that using mathematical induction contains two part:

Part 1: Prove that the desired proposition satisfies the requirement of Axiom of Induction, which is usually showed in a fashion like "base case ... case n ... case n+1".

Part 2: Once Part 1 is finished, the QED is just a sentence away: By Axiom of Induction, the desired proposition holds for all natural numbers.

The correctness of such a proof is no more mysterious than any other proofs: In Part 1, you made sure that the desired proposition satisfies an already-true proposition (in this case, Axiom of Induction), thus you can apply the already-true proposition to the desired proposition to get a true proposition, which is Part 2.

I don't think you can help your students understand the Axiom of Induction, it's called an axiom for a reason. It's a formal statement of the intuition we have on natural numbers, and it takes a lot to understand an intuition.

Not an ID
  • 837
  • 1
  • 6
  • 13

I think what you need is an intuitive exaplanation for mathematical induction. The example of toppling dominos is good. But, there is an even better one.

Suppose you have a ladder. All that mathematical induction says is that if you can climb onto the first step of the ladder, and from any step you can climb to the next step, then you can climb to the top of the ladder.

See, it is very logical. That is how most people climb a ladder, right :)

(I saw this in a writing of some mathematician; unfortunately I don't remember the source. Here is an overlong article in medium: An Intuitive Understanding : Mathematical Induction. I didn't like it myself; but we know taste of people differ.)

Cyriac Antony
  • 462
  • 5
  • 17

Bearing in mind that mathematical induction is an axiom, it can be interpreted as "If there are no appearing obstacles/boundaries to stop us, then we can walk as we wish". (not a translation/modification of any kind of quote, just threw off my head).

Metin Y.
  • 1,344
  • 1
  • 11
  • 22

you could use programming for induction proof:


#simple func which returns sum of the arguments
>>> def func(x, y):
...     return x + y
#range from 0 to 9
>>> r1 = range(10)
>>> r1
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
#range from -10 to -1
>>> r2 = range(-10, 0 , 1)
>>> r2
[-10, -9, -8, -7, -6, -5, -4, -3, -2, -1]
#range from 20 to 29
>>> r3 = range(20,30,1)
>>> r3
[20, 21, 22, 23, 24, 25, 26, 27, 28, 29]

we know that:

if a>0 and b>0 than a+b>0 
if a<0 and b<0 than a+b<0 
if a>0 and b<0 than a+b>0 (if -b < a) ... etc

now we could see it:

#map parameters from list1(x) and list2(y) and call func with func(x,y)
>>> map(lambda x, y: func(x,y), r1, r2)
[-10, -8, -6, -4, -2, 0, 2, 4, 6, 8]
>>> map(lambda x, y: func(x,y), r3, r2)
[10, 12, 14, 16, 18, 20, 22, 24, 26, 28]
>>> map(lambda x, y: func(x,y), r3, r1)
[20, 22, 24, 26, 28, 30, 32, 34, 36, 38]

i hope it helps.


Since we are discussing how to prove non-provable statements, I think you should first make sure that the students understand that, even though you are trying your best to not use a proof in this case, proofs are absotively, posolutely desirable in almost every single other case.

Next, I think the important thing to stress is that, for those few statements that you can't prove (axioms), the acceptance/nonacceptance of those statements depends on their usefulness. I consider an axiom to be useful if it can prove many interesting statements without making any contradictions. For instance, many mathematicians accept the axiom of choice for precisely this reason: although it is unintuitive, it is still profoundly useful without being contradictory.

In other words, you should challenge your students to live their (mathematical) lives without ever applying the induction axiom. Show them how much math they would be missing out on - how many useful and elegant theorems they would be unable to prove without inducting. Also stress that induction is not contradictory (so far as we know); if your students really hate the induction axiom, they should try to derive a contradiction from it.

Matthew W.
  • 179
  • 1
  • 6
  • 1
    Wait a minute... How is the axiom of choice more useful than the axiom of determinacy? Isn't this still completely subject to one's personal opinion? Furthermore, as you know, we don't even know that ZF is consistent, so we can't say that ZF is useful by your criterion, because it may turn out to be contradictory. The reason why people accept ZF is generally because they can do what they *want* with it. – user21820 Dec 26 '13 at 10:56

There is also the parking analogy. Let us consider a parking with size infinite. We assume that we already know that in front of every green car there is a green car. And we look at the first car and it is a green car so what should be the color our of all cars?

  • 5,945
  • 1
  • 11
  • 27

A mathematical axiom should be self-evident. It should be unnecessary to explain why something is self-evident.

One should prove mathematical induction based on the self-evident proposition that every set of natural numbers has a least element.

I have found that students immediately understand the truth of that proposition whereas they do not immediately understand the principle of mathematical induction.

Mathematical induction is better approached as a theorem rather than an axiom.

John Wayland Bales
  • 19,649
  • 2
  • 17
  • 48
  • If something is self-evident to you and yet your students doesn't understand it, that means we need to come up with a *better explanation*. I understand that you explain it indirectly. But OP is asking for something that leaves an impact on students. – Cyriac Antony Apr 15 '19 at 11:01
  • i don't know why the above remark has a negative rating. It seems quite logical to derive mathematical induction through the Well Ordering Principle as stated above. – matqkks Apr 15 '19 at 14:44