MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am afraid to continue to ask trivial things but really I do not know how to proceed, so I ask the experts:

A specially multiplicative function, is a function $f$ from positive integers to complex numbers, that satisfies: $$ f(n)f(m) = \sum_{d \mid \gcd(m,n)} f(\frac{mn}{d^2})g(d) $$ for all $m,n$, where $g$ is a completely multiplicative function, i;e. $$ g(rs) = g(r)g(s) $$ for all positive integers $r,s.$

Seems well known that any specially multiplicative function $F$ is the Dirichlet convolution of $2$ completely multiplicative functions. (We write, say, $F = r * s$). That is:

$$ F(n) = r * s (n) = \sum_{d \mid n} r(d) s(n/d) $$ with $r,s$ completely multiplicative functions.

I checked that

$$ \sigma = id * z $$

where $\sigma$ is the sum of divisors's function, (that is specially multiplicative with a explicit $g$), $$ id(n) = n $$ for all $n$ and $$ z(n) = 1 $$ for all $n.$

But I am (unfortunately) unable to find completely multiplicative functions $a,b$ such that

$$ \tau = a * b $$

where $\tau$ is Ramanujan's tau function.

Question 1 : Somebody can display such $a,b$ ???

Probably, there is a standard construction of such $a,b$ for any specially multiplicative function.

Question 2 : What is the recipe to get such $a,b$ for a given specially multiplicative function ???

share|cite|improve this question
up vote 3 down vote accepted

I don't understand your definition of "specially multiplicative"; if $m$ and $n$ are coprime, it gives $f(m)f(n)=mn$, which is not true of the sum of divisor function...

However, the Ramanujan $\tau$-function satisfies $\tau(m)\tau(n)=\sum_{d\mid (m,n)}{\mu(d)\tau(mn/d^2)}$, which might be what you mean?

In any case, $\tau$ is indeed the Dirichlet convolution of two totally multiplicative functions: factor the local $p$-factor of the Dirichlet generating series, which is $1-\tau(p)p^{-s}+p^{11-2s}$, as $(1-a(p)p^{11/2-s})(1-b(p)p^{11/2-s})$, then the totally multiplicative functions mapping $p$ to $a(p)$ and $b(p)$ satisfy $a* b=\tau$.

There is therefore (uncountably) many possibilities for these functions (with $a(p)$ being chosen out of two possibilities for every $p$), so it is hard to get a simple formula (especially a formula that does not involve factoring $n$ into primes).

share|cite|improve this answer
Apologies! : I forgetted the "f" in the right hand side. Would edit and correct this. Thanks a lot for the answer. For question 2, is the same thing ? – Luis H Gallardo Jun 13 '11 at 20:24

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.