Im looking for real entire functions f(x) such that :

f(x) = f(x+2pi) = 0 + a1 x - a2 x^2 + a3 x^3 + a4 x^4 - a5 x^5 + ...

where an > 0 and the pattern continues {+,+,-}.

This comes from the idea of understanding the signs of the nth derivatives of certain functions , being tetration related or more classical functions.

For instance sin and cos have the pattern {+,-} ( omitting 0's ).

SO it seems natural to ask about the pattern {+,+,-}.

Consider the many theta functions and fourier series that seems natural.

I have not seen this question before.

Maybe this is easy and there might even be an elementary f(x).

Maybe calculus 101 or trig 101 etc.

But here is how I tried to consider it.

Btw this also comes from my lost notebook.

f(x) = A(x) + B(x)

A(x) = 0 + A1 sin(x) + A2 sin(2x) + A3 sin(3x) + ...

B(x) = 0 + B1 cos(x) + B2 cos(2x) + B3 cos(3x) + ...

A(x) = 0 + (A1 + A2 2^1 + A3 3^1 + ...) x^1/1! -

(A1 + A2 2^3 + A3 3^3 + ...) x^3/3! +

(A1 + A2 2^5 + A3 3^5 + ...) x^5/5! - ...

(A1 + A2 2^(2k+1) + A3 3^(2k+1) + ...) x^(2k+1)/(2k+1)!

B(x) = 0 + (B1 + B2 + B3 + ...) x^0/1! -

(B1 + B2 2^2 + B3 3^2 + ...) x^2/2! +

(B1 + B2 2^4 + B3 3^4 + ...) x^4/4! - ...

(B1 + B2 2^(2k) + B3 3^(2k) + ...) x^(2k)/(2k)!

Now I have an infinite system of equations/inequalities that reminds me of dirichlet series.

Matrix methods ??

I noticed {+-+-+-}*{+--+--}={++---+}

{+-+-+-}*{+---++}={++-++-}

I thought it might help.

Is it possible for such an f(x) to exist ?

Or do we need that 50 % have sign + and 50% have sign - ?

I vaguely remember someone saying related stuff about truncated Taylor series leading to some limitations.

A related question is how fast the derivatives of a 2pi periodic function can grow ?

Related : c_1 + c_2 2^k + c_3 3^k + ... c_n n^k ~ exp(exp(k)) ?

If we allow c_n to be 0 often it seems we might get an approximation from the Taylor series of exp(x).

Otherwise I dont know.

I was inspired to share this idea because of the thread :

http://math.eretrandre.org/tetrationforu...hp?tid=882

regards

tommy1729

f(x) = f(x+2pi) = 0 + a1 x - a2 x^2 + a3 x^3 + a4 x^4 - a5 x^5 + ...

where an > 0 and the pattern continues {+,+,-}.

This comes from the idea of understanding the signs of the nth derivatives of certain functions , being tetration related or more classical functions.

For instance sin and cos have the pattern {+,-} ( omitting 0's ).

SO it seems natural to ask about the pattern {+,+,-}.

Consider the many theta functions and fourier series that seems natural.

I have not seen this question before.

Maybe this is easy and there might even be an elementary f(x).

Maybe calculus 101 or trig 101 etc.

But here is how I tried to consider it.

Btw this also comes from my lost notebook.

f(x) = A(x) + B(x)

A(x) = 0 + A1 sin(x) + A2 sin(2x) + A3 sin(3x) + ...

B(x) = 0 + B1 cos(x) + B2 cos(2x) + B3 cos(3x) + ...

A(x) = 0 + (A1 + A2 2^1 + A3 3^1 + ...) x^1/1! -

(A1 + A2 2^3 + A3 3^3 + ...) x^3/3! +

(A1 + A2 2^5 + A3 3^5 + ...) x^5/5! - ...

(A1 + A2 2^(2k+1) + A3 3^(2k+1) + ...) x^(2k+1)/(2k+1)!

B(x) = 0 + (B1 + B2 + B3 + ...) x^0/1! -

(B1 + B2 2^2 + B3 3^2 + ...) x^2/2! +

(B1 + B2 2^4 + B3 3^4 + ...) x^4/4! - ...

(B1 + B2 2^(2k) + B3 3^(2k) + ...) x^(2k)/(2k)!

Now I have an infinite system of equations/inequalities that reminds me of dirichlet series.

Matrix methods ??

I noticed {+-+-+-}*{+--+--}={++---+}

{+-+-+-}*{+---++}={++-++-}

I thought it might help.

Is it possible for such an f(x) to exist ?

Or do we need that 50 % have sign + and 50% have sign - ?

I vaguely remember someone saying related stuff about truncated Taylor series leading to some limitations.

A related question is how fast the derivatives of a 2pi periodic function can grow ?

Related : c_1 + c_2 2^k + c_3 3^k + ... c_n n^k ~ exp(exp(k)) ?

If we allow c_n to be 0 often it seems we might get an approximation from the Taylor series of exp(x).

Otherwise I dont know.

I was inspired to share this idea because of the thread :

http://math.eretrandre.org/tetrationforu...hp?tid=882

regards

tommy1729