# Irrational Numbers

1. Feb 22, 2009

### timjones007

why do irrational numbers exist? I am well familiar with the proof that irrational numbers exist, but why do they?

2. Feb 22, 2009

### Hurkyl

Staff Emeritus
Your question doesn't really make sense. If you know the proof, then what's your problem?

Are you using the word "why" in some unusual way? If so, you really should have said that up front....

3. Feb 22, 2009

### Unknot

Well, your question does seem odd, but my guess is that you want to ask a philosophical question.

Let me ask you a question. Do rational numbers exist? how do you know this?

4. Feb 22, 2009

### CRGreathouse

For me, the intuitive answer is "because there aren't nearly enough rationals to 'fill in' all the gaps".

5. Feb 22, 2009

### Unknot

Well, I don't understand why people think rational numbers exist and some numbers don't. It's just easier to think that all numbers are mathematical constructs and real numbers are simply, yes, way to fill in gaps between rational numbers.

I wonder what the op thinks of complex numbers.

6. Feb 22, 2009

### csprof2000

It's actually a very interesting question the OP is getting at, and one I've often thought about myself. How much information do we need to have about a number before we can consider the number to be well-defined? Are all numbers which provably exist well-defined under our definitions of well-definedness of a number? Is there a definition of the well-definedness of numbers?

One can also talk about whether numbers are computable or not. It's interesting that the real numbers are most incomputable... what does this mean? What can even be meant by incomputable number?

I think it's an interesting discussion. To the OP: do you think that sqrt(2) exists, and in what sense do you think this? I mean, we both know that there is a proof that it is not rational. It's a relatively tame irrational number. Why do you feel the way you do? I could enjoy this conversation.

7. Feb 22, 2009

### timjones007

no, i don't think sqrt(2) exists. This is my reason: sqrt(2) is just a symbol for it's decimal representation which is 1.414213562..., and the decimal places continue on infinitely.

So, if we will never reach the last digit in the decimal places for sqrt(2), how can we multiply it by itself.

It's the same logic that goes with the fact that we can't multiply any number by infinity. For example, 0 x infinity is an ideterminate form, because although we know logically that you will get 0 if you keep multiplying 0s, we will never finish multiplying 0 infinitely many times so we say that it is undefined.

In other words, sqrt(2) by definition is a number that you multiply by itself in order to get 2. However, we will never be able to get that number so it should be undefined for the same reason that infinity is undefined.

8. Feb 22, 2009

### CRGreathouse

So you don't accept 1/9 = 0.111111... or 1/10 = 0.10000... either?

What about this: I define "foo" as an ordered pair (a, b) where (a, b) = (c, d) iff (a - c)(b - d) = 0 and the operations plus and times are defined by (a, b) + (c, d) = (a + c, b + d) and (a, b) * (c, d) = (ac + 2bd, bc + ad). Do "foo"s exist?

How about "bar"s, where "bar" is an ordered pair (a, b) where (a, b) = (c, d) iff ad = bc and the operations plus and times are defined by (a, b) + (c, d) = (ad + bc, bd) and (a, b) * (c, d) = (ac, bd)? Do "bar"s exist?\

Maybe "baz", where a "baz" is (a) where (a) = (b) iff a - b = 0 and the operations plus and times are defined by (a) + (b) = (a + b) and (a) * (b) = (ab). Do "baz"s exist?

Last edited: Feb 22, 2009
9. Feb 22, 2009

### Unknot

So tell me, what does "exist" mean? Do you think 2 exists? How so? Is there a realm of numbers where 2 exists but sqrt(2) doesn't?

what about sqrt(2) km? Does that exist?

10. Feb 22, 2009

### HallsofIvy

It might be good to point out that while asking if a number is "well-defined" it makes no sense to focus entirely on the decimal representation of the number, as timjones007 does in #7. The decimal representation of a number is just that- a representation- and has little to do with the properties of the number itself.

11. Feb 22, 2009

### Hurkyl

Staff Emeritus
I would be very surprised if he was actually asking interesting questions about formal language, computability theory, or anything like that. I think he simply doesn't have a clear understanding of what others (and he) means by 'number', and lacking such clarity, is flailing about with his intuition.

12. Feb 22, 2009

### csprof2000

"no, i don't think sqrt(2) exists. This is my reason: sqrt(2) is just a symbol for it's decimal representation which is 1.414213562..., and the decimal places continue on infinitely. "

So you take the definition of a number as its decimal representation? This would take a little elaboration to take into account the (very valid) objection raised by CRGreathouse. For instance, you could say that a number is well defined if its decimal representation repeats with a string of digits of finite length L for all places N at least N_0 to the right of the decimal. This covers repeating decimals (1/9 = 0.1111... letting N_0 = 1 and the string of digits being "1", and 1/10 = 0.1000... is covered letting N_0 = 2 and the string of digits being "0", etc.) Obviously, the choice of 10 as the base doesn't make any difference... you could allow this to vary as well.

But the real problem with that is that you're taking the properties of rational numbers and saying that's what makes a number well-defined. Does that make sense? I mean, if we are trying to show that irrational numbers are not well defined, it's a little self-serving to equate a property of rational numbers with well-definedness. Savvy?

"So, if we will never reach the last digit in the decimal places for sqrt(2), how can we multiply it by itself. "

Well, the problem with this is that, as CRG said, 1/10 = 0.100... and this technically also goes on forever... perhaps a better way of saying what you're thinking is that you have a finite set of rules with which you can always generate the next digit in the decimal (or some other sensible) representation. For instance, 1/10 is well-defined because I can say "tenths' place 1, all other places 0" and you can use the two rules to write out the number to any desired number of digits. Does this sound alright, tim?

The only snag with that, of course, is that sqrt(2) is also well defined by this definition of well-definedness. Consider this: sqrt(2) can be found as follows:

sqrt(n)::
x := 0
p := n // could be made more efficient, but who cares?

for p = n to p_min
begin

while x <= n
begin
x = x + 10^p
end
x = x - 10^p

end

Let's see this operating on n = 2.

x = 0.
x = 100, p = 2.
x = 0, p = 2.
x = 10, p = 1.
x = 0, p = 1.
x = 1, p = 0.
x = 2, p = 0.
x = 1, p = 0.
x = 1.1, p = -1
x = 1.2, p = -1
x = 1.3, p = -1
x = 1.4, p = -1
x = 1.5, p = -1
x = 1.4, p = -1
x = 1.41, p = -2
etc.

As you can see, this will always allow you to find the nth decimal digit in a finite number of steps... so you would need a stricter definition than the one I provided to exclude sqrt(2).

13. Feb 22, 2009

### csprof2000

What does everybody else think about what it takes to define a number? Do numbers have to have a value? If so, and you know a number exists for which you cannot possibly find its value... does this mean anything?

14. Feb 22, 2009

### Hurkyl

Staff Emeritus
A type of number system is defined by a list of properties. If a particular set* has those properties, it's a model of that number system, and we would call its elements numbers (of the appropriate type).

(The properties don't have to be complete -- though the definitions for common number systems like the integers or the reals are complete in the appropriate sense)

*: Or type or class or language or whatever foundational gadget you want to use.

Once you actually have an actual, concrete list of properties to work with, you can usually answer simple questions relatively easily. e.g. it's fairly straightforward to show that

* in the rational numbers, 2 doesn't have a square root. (what would its factorization be?)

* in the real numbers, 2 does have a square root. (construct it as the least upper bound)

* for fields the question is undecidable -- some fields do and some fields don't have a square root of 2. (as shown by the previous two examples)

15. Feb 23, 2009

### Take_it_Easy

You are right! sqrt(2) does not exist.
And in fact also 1 does not exist.
1 is the multiplication of 13/7 and 7/13 now
13/7 and 7/13 are just symbols for their decimal representations which are
13/7 = 1,85714285......
7/13 = 0,53846153......
and the decimal places continue on infinitely.

So, if we will never reach the last digit in the decimal places for these two numbers, how can we multiply them together???

In other words, 1 is a number that you get multiplying 13/7 by 7/13.
However, we will never be able to get that number so it should be undefined for the same reason that infinity is undefined.

16. Feb 23, 2009

This is an interesting question and I think it is one that has been discussed since ancient times.According to David Wells (the penguin dictionary of curious and interesting numbers) pi is the only irrational and transcedental number that occurs naturally.People here have been using root 2 as an example and I have been trying to think of an example where this number can be given a unit.Suppose we were told that a square had an area of root 2 metres squared.Does this mean anything when such a square cannot be consructed or have I picked on a dopey example?

17. Feb 23, 2009

### csprof2000

That could be a dopey example.
What about the distance between opposite corners of a square of area 1?

Hurkyl:
I see what you're saying, but I think the problem we're all having is in communicating. I agree that you're absolutely right about numbers... a very clear and thoughtful exposition.

However, I think that the OP means to talk about the value of numbers, not their properties... to know what the number is, not whether it is there or not. I mean, 2 *is* an integer, but how big is 2? We can get to 2 using a finite number of logical steps. Is sqrt(2) a real number? The OP didn't think so, but perhaps after my last post he will agree that sqrt(2) must exist as well... since we can get as close as we like to it on a whim. But in what sense do the numbers which we cannot find values for have these values - even if we know the number must exist?

I apologize that the discussion is a little vague. I'd love to give you an example of such a number, but obviously I can't... I don't know, maybe the reason this topic isn't more mainstream is that it's a rabbit hole, makes no sense, and has no good answer.

18. Feb 23, 2009

Yes csprof2000,yours is a good example,so numbers like root 2 come up but we can't measure them.It is a rabbit hole.

Last edited: Feb 23, 2009
19. Feb 23, 2009

You can construct a square of side 1 by drawing the sides and then you can draw in the two lines from opposite corners.You can't do it the other way round though by drawing the two lines first.Sorry but I am not exactly sure what I am getting at here,just chucking a few thoughts in as they come.This rabbit hole can make the brain ache-time for a nice cup of tea.

20. Feb 23, 2009

### csprof2000

No, you're right, it is sort of strange. Although, I challenge you to draw a line that's exactly 1 unit long, and prove that you got that right...

Basically, how do you get anything exactly right? How close is close enough to be exactly right?

21. Feb 23, 2009

### Hurkyl

Staff Emeritus
But now, you're not doing mathematics anymore -- you've crossed over into physics, or possibly epistemology.

22. Feb 23, 2009

### HallsofIvy

What do you mean we can't measure them? We can construct and measure a length $\sqrt{2}$ as well as we can a length 1. It is true that we cannot write that out in terms of decimal numerals, but that is a problem with the numeration system, not the number.

23. Feb 23, 2009

### HallsofIvy

What exactly do you mean by "draw a line that's exactly 1 unit long"? In Euclidean geometry, we simply declare a segment to have length 1 and base everything else on that. I can then construct a segment that has length exactly $\sqrt{2}$. (The physical "compasses" and "straight edge" represent the mathematics that is going on. Physical measurement is "approximate". Mathematical construction is not.) If you want to continue in this line, you should discuss $\sqrt[3]{2}$ which is not a "constructible" number!

In any case, $\sqrt{2}$, and even $\sqrt[3]{2}$ are as well defined as "1", "2", or "1/2".

24. Feb 23, 2009

### CRGreathouse

Not with compass and straightedge, anyway. But it's possible with a marked straightedge, a http://www.museo.unimo.it/labmat/trisetin.htm [Broken], or origami... :tongue2:

Last edited by a moderator: May 4, 2017
25. Feb 23, 2009

### csprof2000

"What exactly do you mean by "draw a line that's exactly 1 unit long"?"

Exactly what you said in your first post after me, if you think about what the meaning of that is. The point is both 1 and sqrt(2) are equally hard to nail down to infinite precision. I was just using the same language as Dad to say it, and you were using other language.

And what about the discussion of the value of numbers? Do numbers have to have value? And do you have to be able to find the value, at least in principle?