MathOverflow will be down for maintenance for approximately 3 hours, starting Monday evening (06/24/2013) at approximately 9:00 PM Eastern time (UTC-4).

## Regarding Kolmogorov’s Superposition Theorem

Hi Experts,

I have question regarding Kolmogorov's Superposition Theorem:

It is known that: Let ${f(x_1,x_2,...,x_m): \Re^m :=[0,1]^m \to \Re}$ be an arbitrary multivariate continuous function. From Kolmogorov’s Superposition Theorem we have the following representation:

${f(x_1,x_2,...,x_m)= \sum_{q=0}^{2m} \Phi_q (\sum_{p=1}^m \phi_{p,q}(x_p))}$

with continuous one-dimensional outer functions ${\Phi_q}$ and inner functions ${\phi_{p,q}}$. All these functions are defined on real line. The inner functions ${\phi}$ are independent of function ${f(x_1,x_2,...,x_m)}$.

Question is: Is it possible to find inner functions ${\phi_p{(x_p)}}$ which is independent of $q$, that satisfies the superposition theorem:

${f(x_1,x_2,...,x_m)= \sum_{q=0}^{2m} \Phi_q (\sum_{p=1}^m \phi_p (x_p))}$

Where ${\Phi_q, \phi_p, N}$ can be selected and defined where appropriate.

It is critical to our works on nonlinear control, and we look forward to your advises on possible solutions, tips, related documents,etc.

Thank You! Wang Tao

-

If I understand correctly, that doesn't seem possible. If the inner functions are independent of $q$, then the sum of outer functions collapses to a single function $\Phi$ with $\Phi(\cdot) = \sum_{q=0}^{2m} \Phi_q(\cdot)$. So the stronger form of the theorem that you are looking for would be equivalent to:

For every dimension $m$, there exists continuous functions $\phi_1,\phi_2,\cdots ,\phi_m$ from $[0,1]$ to $\mathbb R$ such that, any continuous function $f\colon [0,1]^m \rightarrow \mathbb R$ can be written as $$f(x_1,x_2,...,x_m)= \Phi_f \left(\sum_{p=1}^m \phi_p (x_p)\right)$$ for some continuous function $\Phi_f$ from $\mathbb R$ to $\mathbb R$.

But by taking $f_i$ with $f_i(x_1,x_2,...,x_m)=x_i$ respectively, it can be shown that, that would imply the existence of a continuous map $F\colon [0,1]^m\rightarrow \mathbb R$ which is also one-to-one (since all the coordinates can be recovered from it). But a continuous map from an open set in $\mathbb R^m$ to $\mathbb R$ cannot be one-to-one for $m>1$, so we get a contradiction.

-

Hi AgCl

Thank you very much for the reply. It helps to save my time and effort. As this topic is important to me, I want to think about my answer to you carefully. Allow me some days to draft a proper reply to you?

Warm Regards Wang Tao

-
@Wang Tao: usually you can comment on answers submitted to questions that you asked. In this case you apparently have two separate accounts, which is why you posted this as an answer. I've requested your original account be merged into your new account. – Willie Wong Nov 30 2010 at 12:08

The inner functions can be constructed indepent of q, which means only one outer function phi is needed. however phi is discontinuous in this case.

-