I am in process of writing essay about cryptography and math behind it. I know that φ(n)=(p1)(q1), but would it be true if p and q are not primes but just ordinary factors of n?

1http://en.wikipedia.org/wiki/Euler's_totient_function#Euler.27s_product_formula – lab bhattacharjee Dec 06 '13 at 07:59

It couldn't be true, could it? Because $6 \times 35 = 10 \times 21$, but $5\times 34 \ne 9 \times 20$. – TonyK Dec 06 '13 at 08:09

In general, $\phi(mn) = \phi(m) \phi(n) \frac{d}{\phi(d)}$ where $d=\gcd(m,n)$. See http://math.stackexchange.com/questions/114841/proofofaformulainvolvingeulerstotientfunction. – lhf Dec 06 '13 at 10:24
3 Answers
In general,
If the integer $n\ge1$ has the prime factorization $n=p^{k_1}_1p^{k_2}_2 \cdot\cdot\cdot p^{k_r}_r$, then $$\phi(n)=n\left(1\frac{1}{p_1}\right)\left(1\frac{1}{p_2}\right) \cdot\cdot\cdot \left(1\frac{1}{p_r}\right)$$.
There is a paper coming out on RoseHulman Institute of Technology Math Journal on Dec 15 about this function if you are interested. The latest issue is here,
http://www.rosehulman.edu/mathjournal/v14n1.php.
You will want to wait until the 15th for the next one. The paper was written by undergrads so it should very accessible.
Update: This is the link to the above mentioned paper: http://www.rosehulman.edu/mathjournal/archives/2013/vol14n2/paper6/v14n26pd.pdf
 952
 6
 20

There does not seem to be any mention of any paper on this topic at that link. – Tobias Kildetoft Dec 06 '13 at 08:41

As I mentioned, it is being published the 15th, it is not out yet. – ReverseFlow Dec 06 '13 at 08:42

Ahh. You might want to make it more clear that the link is not to anything about that paper then. – Tobias Kildetoft Dec 06 '13 at 08:43

Decompose n into prime factors (a)^r,(b)^s,(c)^t... and use the result phi(p^x) = p^x p^x1.
 63
 4

If those are supposed to be the primes dividing $n$, then either this is wrong or you need to assume that $n$ is squarefree. – Tobias Kildetoft Dec 06 '13 at 08:39

Yes sorry decompose to n's prime powers then use phi(p^n) = p^n  p^n1 – stevie Dec 06 '13 at 09:10
Suppose $n=p_1^{k_1}p_2^{k_2}...p_m^{k_m}$, where $p_i,i=1,...,k$ are distinct primes.
Then $\phi(n)=n(1\frac{1}{p_1})(1\frac{1}{p_2})\cdot...\cdot (1\frac{1}{p_m})$.
 1,646
 1
 13
 15

Please consider at least defining the symbols you use. As written this is not actually correct (calling a number $p$ does not mean it is a prime, and unless specified, one might have $p_i = p_j$ for $i\neq j$). – Tobias Kildetoft Dec 06 '13 at 08:51

$\phi$ is the Euler totient function. It is indeed true. Proof first it is multiplicative (it is not strongly multiplicative), then calculate it for prime powers. – Marc Palm Dec 06 '13 at 09:13

1@MarcPalm The claim is only true if the $p_i$ are distinct primes. Just because a notation suggests this to be the case does not mean it should not be pointed out. – Tobias Kildetoft Dec 06 '13 at 10:12

@TobiasKildetoft An edit was made to take care of things. Thanks for pointing it out. – epsilon Dec 06 '13 at 10:43