Smallest common denominators for the homogeneous components of the Baker-Campbell-Hausdorff series

Harald Hofstätter

DOI: https://doi.org/10.48550/arXiv.2012.03818

2020-12-08

Abstract:In a recent paper the author derived a formula for calculating common denominators for the homogeneous components of the Baker-Campbell-Hausdorff (BCH) series. In the present work it is proved that this formula actually yields the smallest such common denominators. In an appendix a new efficient algorithm for computing coefficients of the BCH series is presented, which is based on these common denominators, and requires only integer arithmetic rather than less efficient rational arithmetic.

Number Theory

What problem does this paper attempt to address?

The problem that this paper attempts to solve is about the least common denominator of homogeneous components in the Baker–Campbell–Hausdorff (BCH) series. Specifically, in previous work, the author has already derived a formula to calculate the common denominator of the homogeneous components of the BCH series. In this paper, the author proves that this formula can actually give the **least** common denominator of these homogeneous components.

### Problem Background

The BCH series is formally defined as:
\[ H = \log(e^A e^B) = \sum_{k = 1}^{\infty}\frac{(-1)^{k + 1}}{k}(e^A e^B - 1)^k \]
where \( A \) and \( B \) are non - commutative variables and the coefficients are rational numbers. The BCH series can be written as the sum of homogeneous components:
\[ H=\sum_{n = 1}^{\infty}H_n \]
Each homogeneous component \( H_n \) can be represented as a linear combination of words \( w \) of length \( n \), that is:
\[ H_n=\sum_{w\in\{A, B\}^n}h_w w \]
where \( h_w \) is the coefficient of the word \( w \) in \( H \).

### Main Results

The main result of the paper is Theorem 1, which shows that for any positive integer \( n \) and prime number \( p\geq2 \), define:
\[ l(n, p)=\max\{t:p^t\leq s_p(n)\}=\left\lfloor\log_p(s_p(n))\right\rfloor \]
where \( s_p(n) \) is the sum of the digits in the \( p \)-ary expansion of \( n \). Then define:
\[ d_n = \prod_{p\text{ prime}, p < n}p^{l(n, p)} \]
Then \( n!d_n \) is the least common denominator of the coefficients of all words of length \( n \) in the BCH series, that is:
\[ \text{lcm}\left\{\text{denom}(\text{coeff}(w, H)):w\in\{A, B\}^n\right\}=n!d_n \]

### Solution Method

To prove this result, the author constructs a specific word \( w(n, p)\in\{A, B\}^n \) such that the denominator of its coefficient satisfies:
\[ v_p(\text{denom}(\text{coeff}(w(n, p), H)))=v_p(n!)+l(n, p)=v_p(n!d_n) \]
where \( v_p(k) \) represents the highest power of \( p \) by which \( k \) can be divided.

### Application

In addition, the paper also provides a new and efficient algorithm based on these least common denominators for calculating the coefficients of the BCH series. This algorithm only requires integer operations instead of the less efficient rational number operations, thus improving the computational efficiency.

In summary, this paper solves the problem of how to accurately find the least common denominator of the homogeneous components of the BCH series and provides an efficient calculation method.