On the approximation of quasiperiodic functions with Diophantine frequencies by periodic functions

Kai Jiang,Shifeng Li,Pingwen Zhang

2024-04-22

Abstract:We present an analysis of the approximation error for a $d$-dimensional quasiperiodic function $f$ with Diophantine frequencies, approximated by a periodic function with period $[0,L)^d$. When $f$ has a certain regularity, its global behavior can be described by a finite number of Fourier components and has a polynomial decay at infinity. The dominant part of periodic approximation error is bounded by $O(L^{-1/s})$, where $L$ belongs to the best simultaneous approximation sequence of Fourier frequencies and $s$ is the number of different irrational elements in Fourier frequencies. Meanwhile, we discuss the optimal approximation rate. Finally, these analytical results are verified by some examples.

Number Theory

What problem does this paper attempt to address?

### What problem does this paper attempt to solve?

This paper aims to solve the problem of error analysis when high - dimensional quasi - periodic functions are approximated by periodic functions. Specifically, the author focuses on how to use periodic functions to approximate a d - dimensional quasi - periodic function with Diophantine frequencies within a finite domain and analyze the error of this approximation.

#### Main research contents:

1. **Relationship between quasi - periodic functions and periodic functions**:
- A quasi - periodic function \( f(x) \) can be represented by its Fourier series as:
\[
f(x)=\sum_{\lambda \in \Lambda} a_{\lambda} e^{i2\pi \lambda \cdot x}, \quad x \in \mathbb{R}^d
\]
where \( a_{\lambda} \) are Fourier coefficients and \( \Lambda \) is the set of Fourier exponents.
- Use a periodic function \( f_p(x) \) to approximate \( f(x) \), and the periodic function can be represented as:
\[
f_p(x)=\sum_{\ell = 1}^D b_{\ell} e^{i2\pi h_{\ell} \cdot x/L}, \quad x \in [0, L)^d
\]

2. **Error analysis**:
- **Truncation error**: When \( f(x) \) has a certain regularity, the truncation error can be estimated by \( O(N^{\kappa-\alpha}) \), where \( \alpha > \kappa > d/2 \).
- **Rational approximation error**: It is mainly controlled by Diophantine approximation theory, and the error is \( O(L^{- 1/s}) \), where \( s \) is the number of different irrational elements and \( L \) belongs to the optimal simultaneous approximation sequence.

3. **Optimal approximation rate**:
- The paper discusses the optimal approximation rate and verifies these theoretical results through some examples.

4. **Application of Diophantine conditions**:
- The concept of Diophantine numbers is introduced to ensure that the irrationality of Fourier exponents satisfies certain conditions, thereby ensuring the controllability of the approximation error.

#### Research significance:

- **Theoretical contribution**: It provides a strict and systematic theoretical analysis of the periodic approximation problem of quasi - periodic functions, filling the gap in this field.
- **Practical application**: The periodic approximation method is widely used in solving problems of quasi - periodic systems, such as complex structure modeling in materials science, physics and other fields.

Through the above analysis, the paper not only deepens the understanding of the periodic approximation of quasi - periodic functions, but also provides theoretical support for practical applications.