Abstract:We provide evidence for a conjecture of Yamamura that the truncated logarithmic polynomials \[ F_n(x) = 1 + x + \frac{x^2}{2} + \cdots + \frac{x^n}{n} \] have Galois group $S_n$ for all $n \geq 1$.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to verify a conjecture of Yamamura, that is, for all \(n \geq 1\), the Galois group of the truncated logarithmic polynomial \(F_n(x) = 1 + x+\frac{x^2}{2}+\cdots+\frac{x^n}{n}\) is \(S_n\) (the symmetric group). Specifically, Yamamura conjectured that the discriminant \(\text{disc}(F_n)\) of these polynomials is not the square of a rational number.
### Background and Problem Description
In the literature [12], the authors studied two families of polynomials:
\[F_n(x) = 1 + x+\frac{x^2}{2}+\cdots+\frac{x^n}{n},\]
\[f_n(x) = 1-\frac{x^2}{2!}+\cdots+(-1)^n\frac{x^{2n}}{(2n)!}.\]
These two polynomials are the truncated Maclaurin polynomials of the functions \(F(x) = 1-\log(1 - x)\) and \(f(x)=\cos(x)\) respectively. Based on Schur's research results on truncated exponential polynomials [11], they explored the irreducibility and Galois groups of these polynomials.
For \(F_n(x)\), they proved that \(\text{Gal}_Q(F_n)\) contains \(A_n\) (the alternating group), but the conclusion of \(\text{Gal}_Q(F_n)\cong S_n\) is controversial. Yamamura pointed out in his comment that this conclusion may be incorrect and proposed the conjecture that \(\text{Gal}_Q(F_n)\cong S_n\) holds for all \(n\geq1\).
### Main Contributions
The main contributions of this paper are as follows:
1. **Provide evidence to support Yamamura's conjecture**: By calculating and analyzing the discriminant of \(F_n(x)\), prove that it is not the square of a rational number.
2. **Handle discriminants in specific cases**: In particular, when \(n\equiv0, 2, 3\pmod{4}\) or \(n\equiv1\pmod{4}\) and \(n\) is a power of an odd prime, prove that \(\text{disc}(F_n)\notin\mathbb{Q}^\times_2\).
3. **Introduce new infinite families of integers**: For some \(n\) satisfying \(n\equiv1\pmod{4}\), give new infinite families of integers such that \(\text{disc}(F_n)\) is not the square of a rational number.
### Formula Presentation
- Definition of discriminant:
\[
\text{disc}(P)=(-1)^{\binom{n}{2}}a_n^{-1}\text{Res}(P, P')
\]
where \(P(x)\) is a univariate polynomial of degree \(n\), and \(a_n\) is its leading coefficient.
- Derivative of \(F_n(x)\):
\[
F'_n(x)=1 + x+x^2+\cdots+x^{n - 1}
\]
whose roots are non - trivial \(n\)-th roots of unity.
- Discriminant of \(F_n(x)\):
\[
\text{disc}(F_n)=(-1)^{\binom{n}{2}}n\prod_{\{\theta : F'_n(\theta)=0\}}F_n(\theta)
\]
- Let \(L_n=\text{lcm}\{1, 2,\ldots\)