Proof/Derivation


By Ali Korotana (P4195) on Monday, May 7, 2001 - 08:33 pm :

This may seem like an odd question but, if proof is about why things work, is there a difference between proof and derivation? Does a derivation from basic principles satisfy as a proof?
Thanks
Ali


By Oliver Samson (P3202) on Monday, May 7, 2001 - 09:40 pm :

I'd have thought a derivation is one form of proof, so yes, it does satisfy as a proof, but there is a difference between the two.


By Anonymous on Monday, May 7, 2001 - 09:46 pm :

This is an interesting question. In any axiomatic system (collection of statements derived from stated principles) a statement is either TRUE or FALSE. Like you say proof can be seen as nothing more than derivation, it is a method of convincing us that a statement is TRUE. I think there is quite a lot of philosophical argument about whether a statement is 'TRUE' or 'FALSE' before it is proved or disproved, also about whether a statement that is NOT TRUE is necessarily FALSE. I don't know very much about this but would like to learn more, I've seen one or two similar posts on the topic, perhaps NRICH could get an expert to do an article on the subject.

There is also the interesting question of Godel's theorems that state that in any axiomatic system there exists a statement that cannot be proved or disproved.

Fascinating...


By Olof Sisask (P3033) on Tuesday, May 8, 2001 - 05:11 pm :

On a slightly separate note, does this work as a proof?

Let S be the set of all positive integers about which there are no interesting facts. S then has a least element, x. But then there is an interesting fact about x, therefore no such set can exist.

Seems kinda weird to me...

Olof


By Michael Doré (Michael) on Tuesday, May 8, 2001 - 06:20 pm :

In other words every number has an interesting property :)


By Anonymous on Tuesday, May 8, 2001 - 06:23 pm :

Actually, isn't this a version of Russell's paradox?


By Michael Doré (Michael) on Tuesday, May 8, 2001 - 06:27 pm :

No, it's not really the same. Russell's paradox is based on the axiom of abstraction, formulated by Frege in 1893. It says:

"Given any property, there exists a set whose members are just those entities possessing that property."

Now soon it was discovered that this was a nonsensical axiom - i.e. it was not self-consistent and it was quickly abandoned. Can you see why? There are two well-known proofs. One of them is Russell's paradox Here it is:

By the axiom of abstraction we can find a set S which contains all entities such that

i) each entity in S is a set
ii) each set in S is not a member of itself.

(So the set of all words is in S because the set of words is not a word itself. But the set of all things apart from words is not a word so it is a not a member of S.)

Now consider the question is S a member of S? Well if it is then it isn't and if it isn't then it is. We have our contradiction.

There is another way of showing the axiom of abstraction is nonsensical; this one's called Cantor's Paradox.

By the axiom of abstraction, we can find a set S that contains all entities - i.e. is the universal set. Now by Cantor's theorem the set of subsets of S (called P(S)) is of higher cardinality than S, so there is no surjection from S to P(S). But as S is the universal set P(S) is just a subset of S, so there is obviously a surjection from S to P(S) contradicting Cantor's Theorem, which can be proved. We have a paradox.

These prove that the axiom of abstraction is inconsistent. It has been replaced by the axiom of separation which states that for any property there exists a set whose members possess just that property and whose members all belong to some other set X. Can you see how this gets around these two?

Just for completeness, here are a few more paradoxes.

Liar's Paradox

It is believed to have been put forward by the Cretan philosopher Epimendies around 500 BC. Suppose someone says:

"I am lying."

This statement contradicts itself. For if it is true, then it is false and if it is false then it is true. This is an example of a grammatically correct sentence which is not logically self-consistent.

Grelling's Paradox

Put forward by K. Grelling in 1808. We call an adjective autological if it has the property of itself. So for instance, the word polysyllabic is polysyllabic (as it has > 1 syllables) therefore it is autological. Also English is English, so English is an autological word too. On the other hand the word French is not French so it is called heterological. (The definition of a heterological word is one which is not autological.) Monosyllabic is another heterological word. Now a question: is the word "heterological" autological or not? Well if it is then this implies it is heterological, contradicting the fact it is autological. But if it is heterological then it is not heterological so it is autological which is again a contradiction. We have a paradox.

(Clearly, this is very, very similar to Russell's paradox.)

Richard's Paradox

Put forward by J. Richard in 1905. We are first of all going to specify an enumeration of the English phrases that denote real numbers.

First take all one-syllable phrases representing numbers. Order these in terms of the magnitude of the number. Next write down the words of two syllables (again ordering these in magnitude) then three syllables etc. So we now have an enumeration of phrases denoting real numbers. Call the nth phrase in the list the nth Richard number.

Now consider the phrase denoting the real number r, such that for all integers n, the nth decimal place of r is 2 if the nth decimal place of the nth Richard number is 1. Otherwise the nth decimal place is 1.

This expression should be a Richard number but if it's the kth then it differs in the kth decimal place to the kth Richard number which is a contradiction.

Berry's Paradox

Formulated by G. Berry in 1906. Let f(n) be equal to the least number of syllables you need to communicate the number n. So f(1) = 1 because the number 1 can be communicated in one syllable - namely "one". Now consider the minimal integer n such that f(n) > 18. If you do a bit of calculation it appears that this number is one hundred and eleven thousand seven hundred and seventy seven (i.e. 111777) which is 19 syllables. However notice that you can also specify the number in eighteen syllables as follows: it is "the least integer not nameable in fewer than nineteen syllables". If you look at the expression in quotes it uniquely names the number in question and yet the quoted expression is only 18 syllables. So n can be specified in 18 syllables which is a contradiction so we have a paradox.


By Michael Doré (Michael) on Tuesday, May 8, 2001 - 07:07 pm :

Going back to the earlier discussion I think Godel's proof says something like in any axiomatic system involving arithmetic there are always undecidable statements. Clearly there are trivial axiomatic systems with no undecidable statements.


By Olof Sisask (P3033) on Tuesday, May 8, 2001 - 07:50 pm :

Do you happen to know of any mathematical (non-set specific) statements that have been proven to be unprovable (if that's a word)?

Nice list of paradoxes by the way.

Olof


By Kerwin Hui (Kwkh2) on Tuesday, May 8, 2001 - 08:01 pm :

Provability (if that is a word) depends on what axioms you use. For example, it was shown, by Godel in 1938, that the axiom of choice is consistent with the framework of ZF set theory. Further, in 1963, Cohen proved that the negation of axiom of choice can also be consistent with ZF. Hence the Axiom of Choice is undecidable in ZF. I think it was also proven that the (generalised) continuum hypothesis is independent of ZF+Choice.

There is a theorem which states that a statement A is provable in the system N if and only if it is true in all models of N. Godel and Cohen's work basically consisted of constructing a model of ZF in which AC is true and false respectively.

Kerwin


By Olof Sisask (P3033) on Tuesday, May 8, 2001 - 08:18 pm :

What's the continuum hypothesis?

Thanks,
Olof


By Kerwin Hui (Kwkh2) on Tuesday, May 8, 2001 - 08:22 pm :

The continuum hypothesis states that, if a subset of R, the reals, is infinite, then it either bijects with N, the naturals, or R itself. Basically this means there are no intermediate infinities between the size of N and the size of R.

Kerwin


By Olof Sisask (P3033) on Tuesday, May 8, 2001 - 08:30 pm :

Sorry I'm not quite sure what bijects means - could you perhaps give a brief explanation? Thanks.

Olof


By Kerwin Hui (Kwkh2) on Tuesday, May 8, 2001 - 08:44 pm :
Bijects means there is a one-to-one mapping, and the mapping is a bijection. For example, the set of all positive even numbers bijects with the set of natural numbers (pretty obviously).

I perhaps should have stated that the general continuum hypothesis states

Àn+1=2Àn

where Àn (aleph n) means the nth infinity (with À0 being the cardinality (size) of the naturals)

Kerwin


By Anonymous on Tuesday, May 8, 2001 - 08:47 pm :

Sets A and B biject if there is a mapping A-> B which takes no two elements of A to the same element of B and takes some element of A to each element of B.

If A and B are finite then this says the number of elements in A is the same as the number of elements in B. We generalise and say that any two sets have 'the same number of elements' (cardinality) if and only if they biject.