0810.0382/HM4.tex
1: % HoshiMiyake.tex (2009/06/06) 
2: % RIMS LaTeX style files
3: % 
4: % Please use the RIMS style files "rims-bessatsu.cls" and "crop.sty" when
5: % submitting papers for RIMS Kokyuroku Bessatsu.  Since the content of these
6: % files determines the style of "RIMS Kokyuroku Bessatsu", changing
7: % their contents may cause incompatibilities with the style of other
8: % papers, and can also make index-creation impossible.  Never change the
9: % content.  However, crop.sty is essential only to the printing process,
10: % so the initial draft can be made without using it.  RIMS style files
11: % are designated for LaTeX2e, and although they can be used with
12: % LaTex2.09, we strongly recommend the use of LaTeX2e.
13: % 
14: %  Please use all or part of the following template
15: % "bessatsu.tex" as a guide to style when preparing papers, of course
16: % changing the file names.
17: %
18: % 
19: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
20: %%%%%%   bessatsu.tex for rims-bessatsu.cls        %%%%%%
21: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
22: %%%
23: %%% This file is based on rimstp27.tex.
24: %%% Modified by T. A. on Dec. 21, 2006.
25: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
26: % Please insert comments, if any, describing the paper here.
27: %
28: %
29: %%%\documentclass[cropon,leqno,draft,secnum]{rims-bessatsu}%%% only Printer
30: %%%\documentclass[cropon,leqno,draft,ssecnum]{rims-bessatsu}%%% only Printer
31: %%%\documentclass[cropon,leqno,draft,nsecnum]{rims-bessatsu}%%% only Printer
32: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
33: % The above 3 lines are for the editor's use only.  Please leave commented out.
34: %
35: %
36: \documentclass[secnum,leqno,draft]{rims-bessatsu}%%% eqno resets every section 
37: %\documentclass[ssecnum]{rims-bessatsu}%%%eqno resets every subsection 
38: %\documentclass[nsecnum]{rims-bessatsu}%%%eqno never resets 
39: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
40: % Select one from the above 3 choices to specify the style of formula 
41: % numbering.  
42: %
43: %
44: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
45: %%% options %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
46: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
47: %%%[  secnum,nsecnum,ssecnum,fleqn,cropon]
48: %%%  use [leqno] option with amsmath package
49: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
50: \usepackage{amssymb, amsmath}%%% option packages
51: %%%%%%%%%%%%%%%%%%%%%%%%%%%
52: \usepackage{graphics}%%% option packages
53: \usepackage[dvips]{graphicx}%%% option packages
54: \usepackage{eepic}%%% option packages
55: \usepackage{epic}%%% option packages
56: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
57: %
58: % Uncomment the above to use the amssymb and amsmath package.  
59: % If you use amsmath, please add [leqno] to \documentclass.
60: %
61: %
62: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
63: %%%%%% Define Lemma, Proposition, Theorem, etc.(Bold 9. Italic)
64: \newtheorem{theorem}{Theorem}[section]
65: \newtheorem{lemma}[theorem]{Lemma}
66: \newtheorem{proposition}[theorem]{Proposition}
67: \newtheorem{corollary}[theorem]{Corollary}
68: \newtheorem*{problem}{Problem}
69: \newtheorem*{isom-prob}{Field isomorphism problem of a generic polynomial}
70: \newtheorem*{int-prob}{Field intersection problem of a generic polynomial}
71: %\newtheorem{thmx}{Theorem}%%% Definition for multi number series
72: %\renewcommand{\thethmx}{\Alph{thmx}}
73: %%\newtheorem{thmy}{Theorem}
74: %%\renewcommand{\thethmy}{\arabic{thmy}}
75: %%%
76: %%%%%% Define Definition, Example etc. (Bold 9. Roman)
77: \theoremstyle{definition}
78: \newtheorem*{definition}{Definition}
79: \newtheorem{example}[theorem]{Example}
80: \newtheorem{remark}[theorem]{Remark}
81: \newtheorem*{acknowledgments}{Acknowledgments}
82: %%%%%% Define Note, Remark etc. (Italic 9. Roman)
83: \theoremstyle{remark}
84: %\newtheorem{rem}{Remark}%%%
85: \newtheorem*{rem}{Remark}%%% * means no numbering
86: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
87: %
88: %\addtocounter{page}{10}   %editors must define this.
89: %
90: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
91: %%%    personal macros
92: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
93: \newcommand{\bs}{\mathbf{s}}\newcommand{\bt}{\mathbf{t}}
94: \newcommand{\ba}{\mathbf{a}}\newcommand{\bb}{\mathbf{b}}
95: \newcommand{\bx}{\mathbf{x}}\newcommand{\by}{\mathbf{y}}
96: \newcommand{\Dels}{\Delta_\mathbf{s}}\newcommand{\Delt}{\Delta_\mathbf{t}}
97: \newcommand{\Dela}{\Delta_\mathbf{a}}\newcommand{\Delb}{\Delta_\mathbf{b}}
98: \newcommand{\betas}{\beta_\mathbf{s}}\newcommand{\betat}{\beta_\mathbf{t}}
99: \newcommand{\betaa}{\beta_\mathbf{a}}\newcommand{\Es}{E_\mathbf{s}}
100: \newcommand{\Gs}{G_\mathbf{s}}\newcommand{\Gt}{G_\mathbf{t}}
101: \newcommand{\Gst}{G_{\mathbf{s},\mathbf{t}}}
102: \newcommand{\Hst}{H_{\mathbf{s},\mathbf{t}}}
103: \newcommand{\Ga}{G_\mathbf{a}}\newcommand{\Gb}{G_\mathbf{b}}
104: \newcommand{\Gab}{G_{\mathbf{a},\mathbf{b}}}
105: \newcommand{\opi}{\overline{\pi}}
106: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
107: %%%For Title Page %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
108: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
109: \title{On the field intersection problem of generic polynomials: a survey}
110: %\TitleHead{}          %optional
111: %\dedicatory{Here is a Dedication}           %optional
112: \author{Akinari \textsc{Hoshi}\footnote{Department of Mathematics, Rikkyo University, 
113: 3--34--1 Nishi-Ikebukuro Toshima-ku, Tokyo, 171-8501, Japan.}
114: ~and Katsuya \textsc{Miyake}\footnote{Department of Mathematics, 
115: School of Fundamental Science and Engineering, Waseda University, 
116: 3--4--1 Ohkubo Shinjuku-ku, Tokyo, 169-8555, Japan.}}
117: \AuthorHead{Akinari Hoshi and Katsuya Miyake}         %optional but recommended
118: \classification{11R16, 11R20, 12E25, 12F10, 12F12.}
119: \support{This work was partially supported by Grant-in-Aid for Scientific 
120: Research (C) 19540057 of Japan Society for the Promotion of Science.}  %optional
121: %\keywords{\textit{Key Words}:}         %optional
122: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
123: %11Rxx Algebraic number theory: global fields
124: %11R16 Cubic and quartic extensions
125: %11R20 Other abelian and metabelian extensions
126: %11R32 Galois theory
127: %12Exx General field theory
128: %12E25 Hilbertian fields; Hilbert's irreducibility theorem
129: %12Fxx Field extensions
130: %12F10 Separable extensions, Galois theory
131: %12F12 Inverse Galois theory
132: %12F20 Transcendental extensions
133: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
134: % Fill in the above with the appropriate information.
135: %
136: %
137: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
138: %%%For Header %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
139: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
140: \VolumeNo{4x}           %editors must define this.
141: \YearNo{200x}           %editors must define this.
142: \PagesNo{000--000}      %editors must define this.
143: %\setcounter{page}{000} %editors must define this.
144: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
145: % The above lines are for the editor's use only.  Please do not delete them.
146: %
147: %
148: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
149: %%%For Foot of Title Page %%%%%%%%%%%%%%%%%%%%%%%%
150: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
151: %\communication{KK, June 8, 2000
152: %	Revised September 11, 2000.}      %editors must define this.
153: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
154: % The above lines are for the editor's use only.  Please do not delete them.
155: %
156: %
157: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
158: %
159: \begin{document}
160: \maketitle
161: \begin{abstract}
162: Let $k$ be a field of characteristic $\neq 2$. 
163: We survey a general method of the field intersection problem of generic polynomials via 
164: formal Tschirnhausen transformation. 
165: We announce some of our recent results of cubic, quartic and quintic cases the details of 
166: which are to appear elsewhere. 
167: In this note, we give an explicit answer to the problem in the cases of cubic and dihedral 
168: quintic by using multi-resolvent polynomials. 
169: \end{abstract}
170: %
171: %
172: \section{Introduction}\label{seIntro}
173: %
174: %
175: Let $G$ be a finite group, $k$ a field of characteristic $\neq 2$, $M$ a field containing $k$ 
176: with $\# M=\infty$, and $k(\bt)$ the rational function field over $k$ with $n$ indeterminates 
177: $\bt=(t_1,\ldots,t_n)$. 
178: Our main interest in this note is a $k$-generic polynomial for $G$ 
179: (cf. \cite{DeM83}, \cite{Kem01}, \cite{JLY02}). 
180: 
181: \begin{definition}
182: A polynomial $f_\bt(X)\in k(\bt)[X]$ is called $k$-generic for $G$ if it has the following 
183: property: the Galois group of $f_\bt(X)$ over $k(\bt)$ is isomorphic to $G$ and every 
184: $G$-Galois extension $L/M$ over an arbitrary infinite field $M\supset k$ can be obtained as 
185: $L=\mathrm{Spl}_M f_\ba(X)$, the splitting field of $f_\ba(X)$ over $M$, for some 
186: $\ba=(a_1,\ldots,a_n)\in M^n$. 
187: \end{definition}
188: 
189: Let $f_\bt^G(X)\in k(\bt)[X]$ be a $k$-generic polynomial for $G$. 
190: Examples of $k$-generic polynomials for $G$ are known for various pairs of $(k,G)$ 
191: (for example, see \cite{Kem94}, \cite{KM00}, \cite{JLY02}, \cite{Rik04}). 
192: Since a $k$-generic polynomial $f_\bt^G(X)$ for $G$ covers all $G$-Galois extensions over 
193: $M\supset k$ by specializing parameters, it is natural to ask the following problem: 
194: %
195: %%%%%%%%%%%%%%%%%%%%%%%%%%%
196: %Field Isomorphism Problem%
197: %%%%%%%%%%%%%%%%%%%%%%%%%%%
198: \begin{isom-prob}
199: For a field $M\supset k$ and $\ba, \bb\in M^n$, 
200: determine whether $\mathrm{Spl}_M f_\ba^G(X)$ and $\mathrm{Spl}_M f_\bb^G(X)$ are 
201: isomorphic over $M$ or not. 
202: \end{isom-prob}
203: %
204: It would be desired to give an answer to the problem within the base field $M$ 
205: by using the data $\ba,\bb\in M^n$. 
206: Throughout this paper, we assume that $f_\ba^G(X)$ is separable for $\ba\in M^n$. 
207: 
208: 
209: Let $S_n$ (resp. $D_n$, $C_n$) be the symmetric (resp. the dihedral, the cyclic) group 
210: of degree $n$. 
211: We take $k$-generic polynomials 
212: \begin{align*}
213: f_t^{C_3}(X)&:=X^3-tX^2-(t+3)X-1\,\in k(t)[X],\\
214: f_t^{S_3}(X)&:=X^3+tX+t\,\in k(t)[X],\\
215: f_{s,t}^{D_4}(X)&:=X^4+sX^2+t\,\in k(s,t)[X]
216: \end{align*}
217: for $G=C_3$, $S_3$, $D_4$, respectively. 
218: By using formal Tschirnhausen transformation, 
219: we showed the following theorem which is an analogue to 
220: the results of Morton \cite{Mor94} and Chapman \cite{Cha96}. 
221: %
222: \begin{theorem}[\cite{Mor94}, \cite{Cha96}, \cite{HM}]\label{thC3}
223: For $m,n\in M$, the splitting fields of $f_m^{C_3}(X)$ and 
224: of $f_n^{C_3}(X)$ over $M$ coincide if and only if there exists $z\in M$ such that either 
225: \[
226: n\,=\,\frac{m(z^3-3z-1)-9z(z+1)}{mz(z+1)+z^3+3z^2-1}\ \ \mathit{or}\ \ 
227: n\,=\,-\frac{m(z^3+3z^2-1)+3(z^3-3z-1)}{mz(z+1)+z^3+3z^2-1}.
228: \]
229: \end{theorem}
230: %
231: We also have analogues to the above theorem for two non-abelian groups $S_3$ and $D_4$ 
232: via formal Tschirnhausen transformation. 
233: %
234: \begin{theorem}[\cite{HM07}]\label{thS3isom}
235: Assume that char $k\neq 3$. 
236: For $a,b\in M$ with $a\neq b$, the splitting fields of $f_a^{S_3}(X)$ and of $f_b^{S_3}(X)$ 
237: over $M$ coincide if and only if there exists $u\in M$ such that 
238: \[
239: b=\frac{a(u^2+9u-3a)^3}{(u^3-2au^2-9au-2a^2-27a)^2}.
240: \]
241: \end{theorem}
242: %
243: \begin{theorem}[\cite{HM-2}]
244: For $a,b\in M$, we assume that $\mathrm{Gal}(f_{a,b}^{D_4}/M)=D_4$. 
245: Then for $a,b,a',b'\in M$, the splitting fields of $f_{a,b}^{D_4}(X)$ and of 
246: $f_{a',b'}^{D_4}(X)$ over $M$ coincide if and only if 
247: there exist $p,q\in M$ such that either 
248: \begin{align*}
249: {\rm (i)}\quad a'&=ap^2-4bpq+abq^2,\quad b'=b(p^2-apq+bq^2)^2\ \ \mathit{or}\\
250: {\rm (ii)}\quad a'&=2(ap^2-4bpq+abq^2),\quad b'=(a^2-4b)(p^2-bq^2)^2.
251: \end{align*}
252: \end{theorem}
253: %
254: By applying Hilbert's irreducibility theorem (cf. for example \cite[Chapter 3]{JLY02}) 
255: and Siegel's theorem for curves of genus $0$ (cf. for example \cite[Theorem 6.1]{Lan78}) 
256: to the theorems above respectively, we get the following corollaries: 
257: %
258: \begin{corollary}\label{cor0}
259: Let $f_\ba^G(X)=f_m^{C_3}(X)$ $($resp. $f_a^{S_3}(X)$, $f_{a,b}^{D_4}(X))$ be as above 
260: in Theorem {\rm 1.1} $($resp. Theorem {\rm 1.2}, Theorem {\rm 1.3}\,$)$ with given $\ba\in M^n$, 
261: and suppose that $M\supset k$ is Hilbertian $($e.g. a number field\,$)$. Then 
262: there exist infinitely many $\bb\in M^n$ such that 
263: $\mathrm{Spl}_{M} f_\ba^G(X)=\mathrm{Spl}_M f_\bb^G(X)$. 
264: \end{corollary}
265: %
266: \begin{corollary}\label{cor01}
267: Let $M$ be a number field and $\mathcal{O}_M$ the ring of integers in $M$. 
268: For $f_a^G(X)=f_m^{C_3}(X)$ $($resp. $f_a^{S_3}(X))$ as above in Theorem {\rm 1.1} 
269: $($resp. Theorem {\rm 1.2}\,$)$ with a given integer $a\in \mathcal{O}_M$, 
270: there exist only finitely many integers $b\in\mathcal{O}_M$ such that 
271: $\mathrm{Spl}_{M} f_a^G(X)=\mathrm{Spl}_M f_b^G(X)$. 
272: \end{corollary}
273: %
274: Indeed integers $b\in\mathcal{O}_M$ as in Corollary \ref{cor01} are 
275: derived from some integer solutions of (finitely many) cubic Thue equations 
276: $aXY(X+Y)+X^3+3XY^2-Y^3=\lambda$ (resp. $X^3-2aX^2Y-9aXY^2-2aY^3-27aY^3=\lambda$) 
277: which are curves of genus $1$ (see also the proof of \cite[Theorem 6.1]{Lan78}). 
278: \vspace*{2mm}
279: 
280: Kemper \cite{Kem01}, furthermore, showed that for a subgroup $H$ of $G$ every $H$-Galois 
281: extension over $M$ is also given by a specialization $f_\ba^G(X)$, $\ba \in M^n$, 
282: of a generic polynomial $f_\bt^G(X)$ for $G$. 
283: Hence a problem naturally arises. 
284: %
285: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
286: %Field intersection Problem%
287: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
288: %
289: \begin{int-prob}
290: For a field $M\supset k$ and $\ba,\bb\in M^n$, determine 
291: the intersection of $\mathrm{Spl}_M f_\ba^G(X)$ and $\mathrm{Spl}_M f_\bb^G(X)$. 
292: \end{int-prob}
293: %
294: Clearly if we get an answer to the field intersection problem of a $k$-generic polynomial, 
295: we also obtain an answer to the corresponding field isomorphism problem. 
296: 
297: The aim of this note is to survey a method to give an answer to the isomorphism problem and 
298: the intersection problem of $k$-generic polynomials via formal Tschirnhausen transformation 
299: and multi-resolvent polynomials. 
300: In Section \ref{seResolv}, we review known results about resolvent polynomials. 
301: In Section \ref{seTschin}, we recall a formal Tschirnhausen transformation which is given 
302: in \cite{HM}. 
303: In Section \ref{seInt}, 
304: we give a general method to solve the intersection problem of $k$-generic polynomials. 
305: In Section \ref{seCubic}, we obtain an explicit answer to the problems in the cubic case. 
306: We give a proof of Theorem \ref{thS3isom} as a special case of the intersection problem 
307: of $f_t^{S_3}(X)=X^3+tX+t
308: $ via formal Tschirnhausen transformation. 
309: In Section \ref{seQuin}, we take the $k$-generic polynomial 
310: \begin{align*}
311: f_{s,t}^{D_5}(X)&=X^5+(t-3)X^4+(s-t+3)X^3+(t^2-t-2s-1)X^2+sX+t\in k(s,t)[X]
312: \end{align*}
313: for $D_5$ which is called Brumer's quintic (cf. \cite{JLY02}). 
314: Based on the general result, we illustrate how to give an answer to the problem 
315: for $f_{s,t}^{D_5}(X)$ by multi-resolvent polynomials. 
316: We also give some numerical examples. 
317: 
318: %
319: %
320: \section{Resolvent polynomials}\label{seResolv}
321: %
322: %
323: 
324: In this section we review some known results in the computational aspects of Galois theory 
325: (cf. the text books \cite{Coh93}, \cite{Ade01}). 
326: One of the fundamental tools in the determination of Galois group of a polynomial is resolvent 
327: polynomials; an absolute resolvent polynomial was introduced by Lagrange \cite{Lag1770} and 
328: a relative one by Stauduhar \cite{Sta73}. 
329: Several kinds of methods to compute resolvent polynomials have been developed by many 
330: mathematicians (see, for example, \cite{Sta73}, \cite{Gir83}, \cite{SM85}, \cite{Yok97}, 
331: \cite{MM97}, \cite{AV00}, \cite{GK00} and the references therein). 
332: 
333: Let $M(\supset k)$ be an infinite field and $\overline{M}$ a fixed algebraic closure of $M$.
334: Let $f(X):=\prod_{i=1}^m(X-\alpha_i) \in M[X]$ be a separable polynomial of degree $m$ with 
335: some fixed order of the roots $\alpha_1,\ldots,\alpha_m\in \overline{M}$. 
336: The Galois group of the splitting field $\mathrm{Spl}_M f(X)$ of $f(X)$ over $M$ 
337: may be obtained by using suitable resolvent polynomials. 
338: 
339: Let $k[\bx]:=k[x_1,\ldots,x_m]$ be the polynomial ring over $k$ with $m$ indeterminates 
340: $x_1,\ldots,x_m$. 
341: Put $R:=k[\bx, 1/\Delta_\bx]$, where $\Delta_\bx:=\prod_{1\leq i<j\leq m}(x_j-x_i)$. 
342: We take a surjective evaluation homomorphism $\omega_f : R \rightarrow 
343: k(\alpha_1,\ldots,\alpha_m),\, \Theta(x_1,\ldots,x_m)\mapsto 
344: \Theta(\alpha_1,\ldots,\alpha_m)$ for $\Theta \in R$.
345: We note that $\omega_f(\Delta_\bx)\neq 0$ from the assumption that $f(X)$ is separable. 
346: The kernel of the map $\omega_f$ is the ideal 
347: $I_f=\{\Theta(x_1,\ldots,x_m)\in R \mid \Theta(\alpha_1,\ldots,\alpha_m)=0\}$. 
348: 
349: Let $S_m$ be the symmetric group of degree $m$. 
350: We extend the action of $S_m$ on $m$ letters $\{1,\ldots,m\}$ 
351: to that on $R$ by $\pi(\Theta(x_1,\ldots,x_m)):=\Theta(x_{\pi(1)},\ldots,x_{\pi(m)})$. 
352: We define the Galois group of a polynomial $f(X)$ over $M$ by 
353: $\mathrm{Gal}(f/M):={\{\pi\in S_m \mid \pi(I_f)\subseteq I_f\}}$. 
354: We write $\mathrm{Gal}(f):=\mathrm{Gal}(f/M)$ for simplicity. 
355: The Galois group of the splitting field $\mathrm{Spl}_M f(X)$ of the polynomial $f(X)$ over $M$ 
356: is isomorphic to $\mathrm{Gal}(f)$. If we take another ordering of roots 
357: $\alpha_{\pi(1)},\ldots,\alpha_{\pi(m)}$ of $f(X)$ for some $\pi\in S_m$, the
358: corresponding realization of $\mathrm{Gal}(f)$ is conjugate in $S_m$. 
359: Hence, for arbitrary ordering of the roots of $f(X)$, $\mathrm{Gal}(f)$ 
360: is determined up to conjugacy in $S_m$. 
361: 
362: \begin{definition}
363: For $H\leq G\leq S_m$, an element $\Theta\in R$ is called a $G$-primitive $H$-invariant if 
364: $H=\mathrm{Stab}_G(\Theta)$ $:=$ $\{\pi\in G\ |\ \pi(\Theta)=\Theta\}$. 
365: For a $G$-primitive $H$-invariant $\Theta$, the polynomial 
366: \[
367: \mathcal{RP}_{\Theta,G}(X)=\prod_{\opi\in G/H}(X-\pi(\Theta))\in R^G[X], 
368: \]
369: where $\opi$ runs through the left cosets of $H$ in $G$, 
370: is called the {\it formal} $G$-relative $H$-invariant resolvent by $\Theta$, and the polynomial
371: \[
372: \mathcal{RP}_{\Theta,G,f}(X)=\prod_{\opi\in G/H}\bigl(X-\omega_f(\pi(\Theta))\bigr)\in M[X]
373: \]
374: is called the $G$-relative $H$-invariant resolvent of $f$ by $\Theta$. 
375: \end{definition}
376: 
377: The following theorem is fundamental in the theory of resolvent polynomials 
378: (cf. for example \cite[p.95]{Ade01}). 
379: 
380: \begin{theorem}\label{thfun}
381: Let $H\leq G\leq S_m$ be a tower of finite groups and $\Theta$ a $G$-primitive  $H$-invariant. 
382: Assume $\mathrm{Gal}(f)\leq G$. 
383: Suppose that $\mathcal{RP}_{\Theta,G,f}(X)=\prod_{i=1}^l h_i^{e_i}(X)$ gives the decomposition 
384: of $\mathcal{RP}_{\Theta,G,f}(X)$ into a product of powers of distinct irreducible 
385: polynomials $h_i(X), i=1,\ldots, l,$ in $M[X]$. Then we have a bijection 
386: \begin{align*}
387: \mathrm{Gal}(f)\backslash G/H\quad &\longrightarrow \quad \{h_1^{e_1}(X),\ldots,h_l^{e_l}(X)\}\\
388: \mathrm{Gal}(f)\, \pi\, H\quad &\longmapsto\quad h_\pi(X)
389: =\prod_{\tau H\subseteq \mathrm{Gal}(f)\,\pi\,H}\bigl(X-\omega_{f}(\tau(\Theta))\bigr)
390: \end{align*}
391: where the product runs through the left cosets $\tau H$ of $H$ in $G$ contained in 
392: $\mathrm{Gal}(f)\, \pi\, H$, that is, through $\tau=\pi_\sigma \pi$ where $\pi_\sigma$ runs 
393: a system of representative of the left cosets of $\mathrm{Gal}(f) \cap \pi H\pi^{-1};$ each 
394: $h_\pi(X)$ is irreducible or a power of an irreducible polynomial with $\mathrm{deg}(h_\pi(X))$ 
395: $=$ $|\mathrm{Gal}(f)\, \pi\, H|/|H|$ $=$ $|\mathrm{Gal}(f)|/|\mathrm{Gal}(f)\cap \pi H\pi^{-1}|$. 
396: \end{theorem}
397: 
398: \begin{corollary} 
399: If $\mathrm{Gal}(f)\leq \pi H\pi^{-1}$ for some $\pi\in G$ then 
400: $\mathcal{RP}_{\Theta,G,f}(X)$ has a linear factor over $M$. 
401: Conversely, if $\mathcal{RP}_{\Theta,G,f}(X)$ has a non-repeated linear factor over $M$ 
402: then there exists $\pi\in G$ such that $\mathrm{Gal}(f)\leq \pi H\pi^{-1}$. 
403: \end{corollary}
404: 
405: \begin{remark}\label{remGir}
406: When the resolvent polynomial $\mathcal{RP}_{\Theta,G,f}(X)$ has a repeated factor, there 
407: always exists a suitable Tschirnhausen transformation $\hat{f}$ of $f$ (cf. \S 3) over $M$ 
408: (resp. $X-\hat{\Theta}$ of $X-\Theta$ over $k$) such that $\mathcal{RP}_{\Theta,G,\hat{f}}(X)$ 
409: (resp. $\mathcal{RP}_{\hat{\Theta},G,f}(X)$) has no repeated factors (cf. \cite{Gir83}, 
410: \cite[Alg. 6.3.4]{Coh93}, \cite{Col95}). 
411: \end{remark}
412: 
413: In the case where $\mathcal{RP}_{\Theta,G,f}(X)$ has no repeated factors, 
414: we have the followings: \\
415: {\rm (i)} For $\pi\in G$, the fixed group of the field $M\bigl(\omega_{f}(\pi(\Theta))\bigr)$ 
416: corresponds to $\mathrm{Gal}(f)\cap \pi H\pi^{-1}$. 
417: In particular, we have $\mathrm{Gal}(\mathcal{RP}_{\Theta,G,f})\cong \mathrm{Gal}(f)/N$ with 
418: $N=\mathrm{Gal}(f)\cap \bigcap_{\pi\in G}\pi H\pi^{-1}$; \\
419: {\rm (ii)} let $\varphi : G\rightarrow S_{[G:H]}$ denote the permutation representation of 
420: $G$ on the set of left cosets $G/H$ given by the left multiplication. Then we 
421: have a realization of the Galois group of $\mathrm{Spl}_M \mathcal{RP}_{\Theta,G,f}(X)$ 
422: as a subgroup of the symmetric group $S_{[G:H]}$ by $\varphi(\mathrm{Gal}(f))$. 
423: 
424: %
425: %
426: \section{Formal Tschirnhausen transformation}\label{seTschin}
427: %
428: %
429: 
430: We recall the geometric interpretation of a Tschirnhausen transformation 
431: which is given in \cite{HM} (see also \cite{HM-3}). 
432: Let $f(X)$ be a monic separable polynomial of degree $n$ in $M[X]$ 
433: with a fixed order of the roots $\alpha_1,\ldots,\alpha_n$ of $f(X)$ in $\overline{M}$. 
434: A Tschirnhausen transformation of $f(X)$ over $M$ is a polynomial of the form 
435: \[
436: g(X)=\prod_{i=1}^n 
437: \bigl(X-(c_0+c_1\alpha_i+\cdots+c_{n-1}\alpha_i^{n-1})\bigr),\ c_i \in M.
438: \]
439: Two polynomials $f(X)$ and $g(X)$ in $M[X]$ are Tschirnhausen equivalent over $M$ if they are 
440: Tschirnhausen transformations over $M$ of each other. 
441: For two irreducible separable polynomials $f(X)$ and $g(X)$ in $M[X]$, 
442: $f(X)$ and $g(X)$ are Tschirnhausen equivalent over $M$ if and only if 
443: the quotient fields $M[X]/(f(X))$ and $M[X]/(g(X))$ are $M$-isomorphic. 
444: 
445: In order to obtain an answer to the field intersection problem of $k$-generic polynomials 
446: via multi-resolvent polynomials, we first treat a general polynomial 
447: whose roots are $n$ indeterminates $x_1,\ldots,x_n$: 
448: \[
449: f_\bs(X)\, =\, \prod_{i=1}^n(X-x_i)\, 
450: =\, X^n-s_1X^{n-1}+s_2X^{n-2}+\cdots+(-1)^n s_n\ \in k[\bs][X]
451: \]
452: where $k[x_1,\ldots,x_n]^{S_n}=k[\bs]:=k[s_1,\ldots,s_n], \bs=(s_1,\ldots,s_n),$ and $s_i$ is the 
453: $i$-th elementary symmetric function in $n$ variables $\bx=(x_1,\ldots,x_n)$. 
454: 
455: Put $R_{\bx,\by}:=k[\bx,\by,1/\Delta_\bx,1/\Delta_\by]$, where $\by=(y_1,\ldots,y_n)$ consists of 
456: $n$ indeterminates, 
457: $\Delta_\bx:=\prod_{1\leq i<j\leq n}(x_j-x_i)$ and $\Delta_\by:=\prod_{1\leq i<j\leq n}(y_j-y_i)$. 
458: We define the interchanging involution $\iota_{\bx,\by}$ which exchanges the 
459: $x_i$'s and the $y_i$'s: 
460: \begin{align}
461: \iota_{\bx,\by}\ :\ R_{\bx,\by}\longrightarrow R_{\bx,\by},\ 
462: x_i\longmapsto y_i,\ y_i\longmapsto x_i,\quad (i=1,\ldots,n).\label{defiota}
463: \end{align}
464: We take another general polynomial $f_\bt(X):=\iota_{\bx,\by}(f_\bs(X))\in k[\bt][X], 
465: \bt=(t_1,\ldots, t_n)$, whose roots are $n$ indeterminates $y_1,\ldots,y_n$ where 
466: $t_i=\iota_{\bx,\by}(s_i)$ is the $i$-th elementary symmetric function in $\by=(y_1,\ldots,y_n)$. 
467: We put $K:=k(\bs,\bt)$ and $f_{\bs,\bt}(X):=f_\bs(X)f_\bt(X)$. 
468: The polynomial $f_{\bs,\bt}(X)$ of degree $2n$ is defined over $K$. We denote 
469: \[
470: \Gs\, :=\, \mathrm{Gal}(f_\bs/K),\quad \Gt\, :=\, \mathrm{Gal}(f_\bt/K),\quad 
471: \Gst\, :=\, \mathrm{Gal}(f_{\bs,\bt}/K). 
472: \]
473: %
474: Then we have $\Gst=\Gs\times\Gt, \Gs\cong \Gt\cong S_n$ and $k(\bx,\by)^{\Gst}=K$. 
475: 
476: We intend to apply the results of the previous section for 
477: $m=2n$, $G=\Gst\leq S_{2n}$ and $f=f_{\bs,\bt}$. 
478: 
479: Note that in the splitting field $\mathrm{Spl}_K f_{\bs,\bt}(X)=k(\bx,\by)$, there exist $n!$ 
480: Tschirnhausen transformations from $f_\bs(X)$ to $f_\bt(X)$ with respect to 
481: $y_{\pi(1)},\ldots,y_{\pi(n)}$ for $\pi\in S_n$. 
482: We shall study the field of definition of each Tschirnhausen transformation from 
483: $f_\bs(X)$ to $f_\bt(X)$. 
484: Let $D:=[x_i^{j-1}]_{1\leq i,j\leq n}$ be the Vandermonde matrix of size $n$. 
485: The matrix $D\in M_n(k(\bx))$ is invertible because ${\rm det}\,D=\Delta_\bx$. 
486: The field $k(\bs)(\Delta_\bx)$ is a quadratic extension of $k(\bs)$ 
487: which corresponds to the fixed field of the alternating group of degree $n$. 
488: %%%%%%%%%%%%%%%%%%%%%%%
489: %Definition of $u_i$'s%
490: %%%%%%%%%%%%%%%%%%%%%%%
491: We define the $n$-tuple $(u_0(\mathbf{x},\mathbf{y}),\ldots,
492: u_{n-1}(\mathbf{x},\mathbf{y}))\in (R_{\bx,\by})^n$ by 
493: \[
494: \left[\begin{array}{c}u_0(\mathbf{x},\mathbf{y})\\ u_1(\mathbf{x},
495: \mathbf{y})\\ \vdots \\ u_{n-1}(\mathbf{x},\mathbf{y})\end{array}\right]
496: :=D^{-1}\left[\begin{array}{c}y_1\\ y_2\\ \vdots \\ 
497: y_n\end{array}\right]. 
498: \]
499: Cramer's rule shows us 
500: \[
501: u_i(\mathbf{x},\mathbf{y})=\Delta_\bx^{-1}\cdot\mathrm{det}
502: \left[\begin{array}{cccccccc}
503: 1 & x_1 & \cdots & x_1^{i-1} & y_1 & x_1^{i+1} & \cdots & x_1^{n-1}\\ 
504: 1 & x_2 & \cdots & x_2^{i-1} & y_2 & x_2^{i+1} & \cdots & x_2^{n-1}\\
505: \vdots & \vdots & & \vdots & \vdots & \vdots & & \vdots\\
506: 1 & x_n & \cdots & x_n^{i-1} & y_n & x_n^{i+1} & \cdots & x_n^{n-1}
507: \end{array}\right].
508: \]
509: We write $u_i:=u_i(\mathbf{x},\mathbf{y}), (i=0,\ldots,n-1)$. 
510: The Galois group $\Gst$ acts on the orbit $\{\pi(u_i)\ |\ \pi\in \Gst \}$ 
511: via regular representation from the left. 
512: However this action is not faithful. 
513: %
514: We put 
515: \[
516: \Hst:=\{(\pi_\bx, \pi_\by)\in \Gst\ |\ \pi_\bx(i)=\pi_\by(i)\ \mathrm{for}\ 
517: i=1,\ldots,n \}\cong S_n. 
518: \]
519: If $\pi \in \Hst$ then we have $\pi(u_i)=u_i$ for $i=0,\ldots,n-1$. 
520: Indeed we see the following: 
521: %
522: \begin{lemma}\label{stabil}
523: For $i$, $0\leq i\leq n-1$, $u_i$ is a $\Gst$-primitive $\Hst$-invariant. 
524: \end{lemma}
525: %
526: 
527: Let $\Theta:=\Theta(\bx,\by)$ be a $\Gst$-primitive $\Hst$-invariant. 
528: Let $\opi=\pi\Hst$ be a left coset of $\Hst$ in $\Gst$. 
529: The group $\Gst$ acts on the set $\{ \pi(\Theta)\ |\ \opi\in \Gst/\Hst\}$ 
530: transitively from the left through the action on the set $\Gst/\Hst$ of left cosets. 
531: Each of the sets $\{ \overline{(1,\pi_\by)}\ |\ (1,\pi_\by)\in \Gst\}$ 
532: and $\{ \overline{(\pi_\bx,1)}\ |\ (\pi_\bx,1)\in \Gst\}$ forms a complete residue 
533: system of $\Gst/\Hst$, and hence the subgroups $\Gs$ and $\Gt$ of $\Gst$ act on the set 
534: $\{ \pi(\Theta)\ |\ \opi\in \Gst/\Hst\}$ transitively. 
535: For $\opi=\overline{(1,\pi_\by)}\in \Gst/\Hst$, we obtain the following equality: 
536: \[
537: y_{\pi_\by(i)} = \pi_\by(u_0)+\pi_\by(u_1) x_i+\cdots+\pi_\by(u_{n-1})x_i^{n-1}\ 
538: \mathrm{for}\ i=1,\ldots,n. 
539: \]
540: Hence the set $\{(\pi(u_0),\ldots,\pi(u_{n-1}))\ |\ \opi\in \Gst/\Hst\}$ 
541: gives coefficients of $n!$ different Tschirnhausen transformations from $f_\bs(X)$ 
542: to $f_\bt(X)$ each of which is defined over $K(\pi(u_0),\ldots,\pi(u_{n-1}))$ respectively. 
543: %
544: \begin{definition}
545: We call $K(\pi(u_0),\ldots,\pi(u_{n-1})), (\opi\in \Gst/\Hst)$, a field of formal 
546: Tschirnhausen coefficients from $f_\bs(X)$ to $f_\bt(X)$. 
547: \end{definition}
548: %
549: Put $v_i:=\iota_{\bx,\by}(u_i)$, for $i=0,\ldots,n-1$. 
550: Then $v_i$ is also a $\Gst$-primitive $\Hst$-invariant, and $K(\pi(v_0),\ldots,\pi(v_{n-1}))$ 
551: gives a field of formal Tschirnhausen coefficients from $f_\bt(X)$ to $f_\bs(X)$. 
552: %
553: \begin{proposition}\label{prop1}
554: For every  $\Gst$-primitive $\Hst$-invariant $\Theta$, we have $K(\pi(\Theta))$ $=$ 
555: $k(\bx,\by)^{\pi\Hst \pi^{-1}}\!\!$ $=$ $K(\pi(u_0),\ldots,\pi(u_{n-1}))$, 
556: and $[K(\pi(\Theta)) : K]=n!$ for each $\opi\in \Gst/\Hst$. 
557: \end{proposition}
558: %
559: Hence, for each of $n!$ fields $K(\pi(\Theta))$, 
560: we have $\mathrm{Spl}_{K(\pi(\Theta))} f_\bs(X)=\mathrm{Spl}_{K(\pi(\Theta))} f_\bt(X)$. 
561: %
562: \begin{proposition}\label{propLL}
563: Let $\Theta$ be a $\Gst$-primitive $\Hst$-invariant. 
564: Then we have 
565: \begin{align*}
566: &{\rm (i)}\ K(\bx)\cap K(\pi(\Theta))=K(\by)\cap K(\pi(\Theta))=K\quad \textrm{for}\quad 
567: \opi\in \Gst/\Hst\,{\rm ;}\\
568: &{\rm (ii)}\ K(\bx,\by)=K(\bx,\pi(\Theta))=K(\by,\pi(\Theta))\quad \textrm{for}\quad 
569: \opi\in \Gst/\Hst\,{\rm ;}\\
570: &{\rm (iii)}\ K(\bx,\by)=K(\pi(\Theta)\ |\ \opi\in \Gst/\Hst). 
571: \end{align*}
572: \end{proposition}
573: 
574: %
575: %
576: \section{Field intersection problem}\label{seInt}
577: %
578: %
579: 
580: In this section, we explain how to get an answer to the field intersection problem of 
581: generic polynomials via multi-resolvent polynomial (cf. \cite{HM}, \cite{HM-3}). 
582: For $\ba=(a_1,\ldots,a_n)$, $\bb=(b_1,\ldots,b_n)\in M^n$, 
583: we take some fixed order of the roots $\alpha_1,\ldots,\alpha_n$ of $f_\ba(X)$ 
584: and $\beta_1,\ldots,\beta_n$ of $f_\bb(X)$ in $\overline{M}$, respectively, and denote 
585: $L_{\ba} :=M(\alpha_1,\ldots,\alpha_n)$ and $L_{\bb}:=M(\beta_1,\ldots,\beta_n)$. 
586: We put $f_{\ba,\bb}(X):=f_\ba(X)f_\bb(X)\in M[X]$ and define a specialization homomorphism 
587: $\omega_{f_{\ba,\bb}}$ by 
588: %
589: \[
590: \omega_{f_{\ba,\bb}}\, :\ R_{\bx,\by}\,\longrightarrow\,
591: k(\alpha_1,\ldots,\alpha_n,\beta_1,\ldots,\beta_n),\quad 
592: \Theta(\bx,\by)\,\longmapsto\,\Theta(\alpha_1,\ldots,\alpha_n,\beta_1,\ldots,\beta_n). 
593: \]
594: %
595: Put $\Delta_\ba:=\omega_{f_{\ba,\bb}}(\Delta_\bx)$ and 
596: $\Delta_\bb:=\omega_{f_{\ba,\bb}}(\Delta_\by)$. 
597: We assume that both of the polynomials $f_\ba(X)$ and $f_\bb(X)$ are separable over $M$, 
598: i.e. $\Delta_\ba\cdot\Delta_\bb\neq 0$. 
599: Put $G_\ba:=\mathrm{Gal}(f_{\ba}/M)$, $G_\bb:=\mathrm{Gal}(f_{\bb}/M)$ and 
600: $G_{\ba,\bb}:=\mathrm{Gal}(f_{\ba,\bb}/M)$. 
601: Then we may naturally regard $G_{\ba,\bb}$ as a subgroup of $\Gst$. 
602: For $\opi \in \Gst/\Hst$, we put $c_{j,\pi}:=\omega_{f_{\ba,\bb}}(\pi(u_j)), 
603: d_{j,\pi}:=\omega_{f_{\ba,\bb}}\bigl(\pi(\iota_{\bx,\by}(u_j))\bigr)$, ($j=0,\ldots,n-1$). 
604: Then for each $i = 1, \ldots, n$, we have 
605: \begin{align*}
606: \beta_{\pi_\by(i)}\,&=\, c_{0,\pi} + c_{1,\pi}\,\alpha_{\pi_\bx(i)}+ \cdots 
607: + c_{n-1,\pi}\,\alpha_{\pi_\bx(i)}^{n-1}, \\
608: \alpha_{\pi_\bx(i)}\,&=\, d_{0,\pi} + d_{1,\pi}\,\beta_{\pi_\by(i)}+ \cdots 
609: + d_{n-1,\pi}\,\beta_{\pi_\by(i)}^{n-1}.
610: \end{align*}
611: %
612: For each $\opi\in\Gst/\Hst$, there exists a Tschirnhausen transformation from $f_\ba(X)$ 
613: to $f_\bb(X)$ over the field of Tschirnhausen coefficients $M(c_{0,\pi},\ldots,c_{n-1,\pi})$. 
614: From the assumption $\Delta_\ba\cdot \Delta_\bb\neq 0$, we first see the following lemma 
615: (cf. \cite[p. 141]{JLY02}, \cite{HM}). 
616: %
617: \begin{lemma}\label{lemM}
618: Let $M'/M$ be a field extension. 
619: If $f_\bb(X)$ is a Tschirnhausen transformation of $f_\ba(X)$ over $M'$, then $f_\ba(X)$ 
620: is a Tschirnhausen transformation of $f_\bb(X)$ over $M'$. 
621: In particular, we have $M(c_{0,\pi},\ldots,c_{n-1,\pi})=M(d_{0,\pi},\ldots,d_{n-1,\pi})$ 
622: for every $\opi\in\Gst/\Hst$. 
623: \end{lemma}
624: 
625: To obtain an answer to the field intersection problem of $f_\bs(X)$ we study the 
626: $n!$ fields 
627: $M(c_{0,\pi},\ldots,c_{n-1,\pi})$ of Tschirnhausen coefficients from $f_\ba(X)$ to 
628: $f_\bb(X)$ over $M$. 
629: %
630: \begin{proposition}\label{propc}
631: Under the assumption $\Delta_\ba\cdot\Delta_\bb\neq 0$, we have\\
632: {\rm (i)}\ \ $\mathrm{Spl}_{M(c_{0,\pi},\ldots,c_{n-1,\pi})} f_\ba(X)
633: =\mathrm{Spl}_{M(c_{0,\pi},\ldots,c_{n-1,\pi})} f_\bb(X)\ \,\textit{for each}\ \,
634: \opi\in\Gst/\Hst$\,{\rm ;}\\
635: {\rm (ii)}\ $L_\ba L_\bb=L_\ba\, M(c_{0,\pi},\ldots,c_{n-1,\pi})=L_\bb\, M(c_{0,\pi},\ldots,c_{n-1,\pi})$\ \,for each\ \,$\opi\in\Gst/\Hst$.
636: \end{proposition}
637: %
638: Applying the specialization $\omega_{f_{\ba,\bb}}$, 
639: we take the $\Gst$-relative $\Hst$-invariant resolvent polynomial of $f_{\ba,\bb}$ 
640: by a $\Gst$-primitive $\Hst$-invariant $\Theta$: 
641: \[
642: \mathcal{RP}_{\Theta,\Gst,f_{\ba,\bb}}(X)\, =\, 
643: \prod_{\opi\in \Gst/\Hst} \bigl(X-\omega_{f_{\ba,\bb}}(\pi(\Theta))\bigr)\in M[X]. 
644: \]
645: The resolvent polynomial $\mathcal{RP}_{\Theta,\Gst,f_{\ba,\bb}}(X)$ is also called 
646: an (absolute) multi-resolvent (cf. \cite{GLV88}, \cite{RV99}, \cite{Val}). 
647: %
648: \begin{proposition}\label{prop12}
649: For $\ba,\bb \in M^n$ with $\Delta_\ba\cdot\Delta_\bb\neq 0$, suppose that the polynomial 
650: $\mathcal{RP}_{\Theta,\Gst,f_{\ba,\bb}}(X)$ has no repeated factors. 
651: Then the following two assertions hold\,{\rm :}\\
652: $(\mathrm{i})$\ $M(c_{0,\pi},\ldots,c_{n-1,\pi})=M\bigl(\omega_{f_{\ba,\bb}}(\pi(\Theta))\bigr)$\, 
653: for each $\,\opi\in\Gst/\Hst$\,{\rm ;}\\
654: $(\mathrm{ii})$\ ${\rm Spl}_M f_{\ba,\bb}(X)
655: =M(\omega_{f_{\ba,\bb}}(\pi(\Theta))\ |\ \opi\in\Gst/\Hst)$. 
656: \end{proposition}
657: 
658: \begin{definition}
659: For a separable polynomial $f(X)\in M[X]$ of degree $d$, the decomposition type of $f(X)$ 
660: over $M$, denoted by {\rm DT}$(f/M)$, is defined as the partition of $d$ induced by the 
661: degrees of the irreducible factors of $f(X)$ over $M$. 
662: We define the decomposition type {\rm DT}$(\mathcal{RP}_{\Theta,G,f}/M)$ of 
663: $\mathcal{RP}_{\Theta,G,f}(X)$ over $M$ by {\rm DT}$(\mathcal{RP}_{\Theta,G,\hat{f}}/M)$ where 
664: $\hat{f}(X)$ is a Tschirnhausen transformation of $f(X)$ over $M$ which satisfies that 
665: $\mathcal{RP}_{\Theta,G,\hat{f}}(X)$ has no repeated factors (cf. Remark \ref{remGir}). 
666: \end{definition}
667: 
668: We write $\mathrm{DT}(f):=\mathrm{DT}(f/M)$ for simplicity. 
669: From Theorem \ref{thfun}, the decomposition type 
670: $\mathrm{DT}(\mathcal{RP}_{\Theta,\Gst,f_{\ba,\bb}})$ coincides with the partition of $n!$ 
671: induced by the lengths of the orbits of $\Gst/\Hst$ under the action of 
672: $\mathrm{Gal}(f_{\ba,\bb})$. 
673: Hence, by Proposition \ref{prop12}, $\mathrm{DT}(\mathcal{RP}_{\Theta,\Gst,f_{\ba,\bb}})$ 
674: gives the degrees of $n!$ fields of Tschirnhausen coefficients 
675: from $f_\ba(X)$ to $f_\bb(X)$ over $M$.  
676: 
677: We conclude that $\mathrm{DT}(\mathcal{RP}_{\Theta,\Gst,f_{\ba,\bb}})$ gives us information 
678: about the field intersection problem for $f_\bs(X)$ through the degrees of the fields 
679: of Tschirnhausen coefficients $M(c_{0,\pi},\ldots,c_{n-1,\pi})$ over $M$ 
680: and is determined by the degeneration of the Galois group $\mathrm{Gal}(f_{\ba,\bb})$ 
681: under the specialization $(\bs, \bt) \mapsto (\ba, \bb)$. 
682: %
683: As a special case of the field intersection problem, we get the followings: 
684: %
685: \begin{theorem}[\cite{HM-3}]\label{throotf}
686: For $\ba,\bb \in M^n$ with $\Delta_\ba\cdot\Delta_\bb\neq 0$, 
687: the quotient fields $M[X]/(f_\ba(X))$ and $M[X]/(f_\bb(X))$ are $M$-isomorphic if and only if 
688: the decomposition type ${\rm  DT}(\mathcal{RP}_{\Theta,\Gst,f_{\ba,\bb}})$ over $M$ includes $1$.
689: \end{theorem}
690: %
691: \begin{corollary}[The field isomorphism problem]\label{cor1}
692: For $\ba,\bb \in M^n$ with $\Delta_\ba\cdot\Delta_\bb\neq 0$, we assume that 
693: both of $G_\ba$ and $G_\bb$ are isomorphic to a transitive subgroup $G\leq S_n$ 
694: and that all subgroups of $G$ with index $n$ are conjugate in $G$. 
695: Then $\mathrm{DT}(\mathcal{RP}_{\Theta,\Gst,f_{\ba,\bb}})$ over $M$ includes $1$ if and only if 
696: $\mathrm{Spl}_M f_\ba(X)$ and $\mathrm{Spl}_M f_\bb(X)$ coincide. 
697: \end{corollary}
698: 
699: %
700: %
701: \section{Field intersection problem: the case of $S_3$}\label{seCubic}
702: %
703: %
704: 
705: We take $f_s^{S_3}(X)=X^3+sX+s\in k(s)[X]$. 
706: For $a,b\in M$, we put $L_a:=\mathrm{Spl}_M f_a^{S_3}(X)$ and 
707: $G_a:=\mathrm{Gal}(f_a^{S_3}/M), G_{a,b}:=\mathrm{Gal}(f_a^{S_3}f_b^{S_3}/M)$. 
708: For $a,b\in M$, we assume that $ab(4a+27)(4b+27)\neq 0$ because the discriminant of 
709: $f_s^{S_3}(X)$ equals $-s^2(4s+27)$. 
710: 
711: In the case of char $k\neq 3$, we take a $\Gst$-primitive $\Hst$-invariant 
712: $\Theta:=3u_1/u_2$ where 
713: $u_1$ and $u_2$ are formal Tschirnhausen coefficients which are given in Section \ref{seTschin}. 
714: Then we may evaluate 
715: \begin{align*}
716: F_{s,t}(X)&:=(s-t)\cdot\mathcal{RP}_{\Theta,\Gst,f_s^{S_3}f_t^{S_3}}(X)
717: =(s-t)\cdot\!\!\prod_{\opi\in\Gst/\Hst}\!\!(X-\pi(\Theta))\, \in k(s,t)[X]\\
718: &\,=\,s(X^2+9X-3s)^3-t(X^3-2sX^2-9sX-2s^2-27s)^2. 
719: \end{align*}
720: The discriminant of $F_{s,t}(X)$ is $s^{10}t^4(4s+27)^{15}(4t+27)^3$. 
721: Note that, for $a,b\in M$ with $ab(4a+27)(4b+27)\neq 0$, 
722: $F_{a,b}(X)$ has no repeated factors. 
723: In the case of char $k=3$, we may take $\Theta=u_0$, 
724: the constant term of a formal Tschirnhausen transformation, as a suitable $\Gst$-primitive 
725: $\Hst$-invariant. 
726: Then we get squarefree $\mathcal{RP}_{u_0,\Gst,f_s^{S_3}f_t^{S_3}}(X)$ 
727: as we mentioned in Remark \ref{remGir} (see \cite{HM}). 
728: %
729: \begin{theorem}[\cite{HM}]\label{thS3}
730: Assume that char $k\neq 3$. 
731: For $a,b\in M$, we also assume that $a\neq b$ and $\#G_a\geq \#G_b>1$. 
732: Then an answer to the field intersection problem for $f_s^{S_3}(X)$ 
733: is given by {\rm DT}$(F_{a,b})$ as Table $1$ shows. 
734: \end{theorem}
735: %
736: \begin{center}
737: {\small 
738: {\rm Table} $1$\vspace*{2mm}\\
739: }
740: {\small 
741: \begin{tabular}{|c|c|c|l|l|}\hline
742: $G_a$& $G_b$ & $G_{a,b}$ & & ${\rm DT}(F_{a,b})$\\ \hline 
743: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
744: & & $S_3\times S_3$ & $L_a\cap L_b=M$ & $6$ \\ \cline{3-5} 
745: & $S_3$ & $(C_3\times C_3)\rtimes C_2$ & $[L_a\cap L_b : M]=2$ & $3,3$\\ \cline{3-5}
746: \raisebox{-1.6ex}[0cm][0cm]{$S_3$} & & $S_3$ & $L_a=L_b$ & $3,2,1$ \\ \cline{2-5}
747: & $C_3$ & $S_3\times C_3$ & $L_a\cap L_b=M$ & $6$ \\ \cline{2-5}
748: & \raisebox{-1.6ex}[0cm][0cm]{$C_2$} & $D_6$ & $L_a\not\supset L_b$ & $6$\\ \cline{3-5}
749: & & $S_3$ & $L_a\supset L_b$ & $3,3$ \\ \hline
750: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
751: & \raisebox{-1.6ex}[0cm][0cm]{$C_3$} & $C_3\times C_3$ & $L_a\neq L_b$ & $3,3$ \\ \cline{3-5}
752: $C_3$ & & $C_3$ & $L_a=L_b$ & $3,1,1,1$\\ \cline{2-5}
753: & $C_2$ & $C_6$ & $L_a\cap L_b=M$ & $6$ \\ \hline
754: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
755: \raisebox{-1.6ex}[0cm][0cm]{$C_2$} & \raisebox{-1.6ex}[0cm][0cm]{$C_2$} & $C_2\times C_2$ 
756: & $L_a\neq L_b$ & $4,2$ \\ \cline{3-5}
757: & & $C_2$ & $L_a=L_b$ & $2,2,1,1$ \\ \hline
758: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
759: \end{tabular}
760: }\vspace*{1mm}
761: \end{center}
762: %
763: As a special case of Theorem \ref{thS3} (cf. Theorem \ref{throotf} and Corollary 
764: \ref{cor1}), we get Theorem \ref{thS3isom} which we introduced in Section \ref{seIntro}. 
765: \vspace*{2mm}
766: 
767: {\bf Proof of Theorem \ref{thS3isom}.}
768: Because the polynomial 
769: $
770: F_{s,t}(X)=s(X^2+9X-3s)^3-t(X^3-2sX^2-9sX-2s^2-27s)^2
771: $
772: is linear in $t$, we see that $F_{a,b}(X)$ has a root in $M$ if and only if there exists 
773: $u\in M$ such that
774: \[
775: b=\frac{a(u^2+9u-3a)^3}{(u^3-2au^2-9au-2a^2-27a)^2}.
776: \]
777: 
778: %
779: %
780: \section{Field intersection problem: the case of $D_5$}\label{seQuin}
781: %
782: %
783: Let $\sigma:=(12345)$, $\rho:=(1243)$, $\tau:=\rho^2$, $\omega:=(12)\in S_5$ act on 
784: $k(x_1,\ldots,x_5)$ by $\pi(x_i)=x_{\pi(i)}, (\pi \in S_5)$. 
785: For simplicity, we write 
786: \[
787: C_5=\langle\sigma\rangle,\quad D_5=\langle\sigma,\tau\rangle,\quad 
788: F_{20}=\langle\sigma,\rho\rangle,\quad S_5=\langle\sigma,\omega\rangle.
789: \]
790: We take the cross-ratios 
791: %
792: \begin{align*}
793: x:=\xi(x_1,\ldots,x_5)=\frac{x_1-x_4}{x_1-x_3}\bigg{/}\frac{x_2-x_4}{x_2-x_3},\quad 
794: y:=\eta(x_1,\ldots,x_5)=\frac{x_2-x_5}{x_2-x_4}\bigg{/}\frac{x_3-x_5}{x_3-x_4}.
795: \end{align*}
796: %
797: Then $S_5$ acts faithfully on $k(x,y)$ as 
798: \begin{align*}
799: \sigma\ &:\ x\longmapsto y,\ y\longmapsto -(y-1)/x,&
800: \rho\ &:\ x\longmapsto x/(x-1),\ y\longmapsto (y-1)/(x+y-1),\\
801: \tau\ &:\ x\longmapsto x,\ y\longmapsto -(x-1)/y,& 
802: \omega\ &:\ x\longmapsto 1/x,\ y\longmapsto (x+y-1)/x.
803: \end{align*}
804: We take a $k$-generic polynomial $f_{s,t}^{D_5}(X)\in k(s,t)[X]$ 
805: as the formal $D_5$-relative $\langle\tau\rangle$-invariant resolvent polynomial by $x$: 
806: \begin{align*}
807: f_{s,t}^{D_5}(X)&:=\mathcal{RP}_{x,D_5}(X)
808: =\bigl(X-x\bigr)\bigl(X-y\bigr)\bigl(X-\frac{1-y}{x}\bigr)
809: \bigl(X-\frac{x+y-1}{xy}\bigr)\bigl(X-\frac{1-x}{y}\bigr)\\
810: &=X^5+(t-3)X^4+(s-t+3)X^3+(t^2-t-2s-1)X^2+sX+t.
811: \end{align*}
812: Note that $k(s,t)=k(x,y)^{D_5}$. 
813: We take two fields $k(\bx)=k(x,y)$ and $k(\bx')=k(x',y')$ where 
814: $x':=\xi(y_1,\ldots,y_5)$, $y':=\eta(y_1,\ldots,y_5)$ and the interchanging involution
815: \[
816: \iota \ :\ k(\bx,\bx')\longrightarrow k(\bx,\bx'),\quad
817: (x,y,x',y')\longmapsto (x',y',x,y)
818: \]
819: which is induced by the $\iota_{\bx,\by}$ of (\ref{defiota}).
820: %
821: We put $(s',t',d'):=\iota(s,t,d)$, $\bs':=(s',t')$ and $(\sigma',\tau',\rho')$ $:=$ 
822: $(\iota^{-1}\sigma\iota,\iota^{-1}\tau\iota$, $\iota^{-1}\rho\iota)$ $\in$ 
823: $\mathrm{Aut}_k(k(x',y'))$ and write 
824: %
825: \[
826: {D_5}'=\langle\sigma',\tau'\rangle,\quad 
827: {F_{20}}'=\langle\sigma',\rho'\rangle\quad \mathrm{and}\quad 
828: {D_5}''=\langle\sigma\sigma',\tau\tau'\rangle.
829: \]
830: %
831: We now take the $D_5\times D_5'$-primitive $D_5''$-invariant 
832: %
833: \begin{align*}
834: P:=xx'+yy'+\frac{(y-1)(y'-1)}{xx'}+\frac{(x+y-1)(x'+y'-1)}{xx'yy'}+\frac{(x-1)(x'-1)}{yy'}
835: \end{align*}
836: %
837: and the formal $D_5\times D_5'$-relative $D_5''$-invariant resolvent polynomial by $P$: 
838: %
839: \begin{align}
840: F_{\bs,\bs'}^1(X)&:=\mathcal{RP}_{P,D_5\times D_5'}(X)
841: =\prod_{\opi\in (D_5\times D_5')/D_5''}(X-\pi(P))\label{exform}\\
842: &\ =\Bigl(X^5-(t-3)({t'}-3)X^4+c_3X^3+\frac{c_2}{2}X^2+\frac{c_1}{2}X+\frac{c_0}{2}\Bigr)^2
843: \nonumber\\
844: &\qquad -\frac{d^2d'^2}{4}\Bigl(X^2+(t+{t'}-1)X+(s-t+{s'}-{t'}+t{t'}+2)\Bigr)^2;\nonumber
845: \end{align}
846: %
847: here $c_3,c_2,c_1,c_0\in k(s,t,{s'},{t'})$ are given by 
848: {\small 
849: \begin{align*}
850: c_3&=\bigl[2s-21t+3t^2-2t{s'}+t^2{\!}{s'}-t^2{t'}\bigr]+31-3s{s'}+5t{t'},\\
851: c_2&=\bigl[-20s+112t+8st-32t^2+2t^3+5t{s'}-13st{s'}-12t^2{\!}{s'}+4t^3{\!}{s'}-15st{t'}\\
852: &+14t^2{t'}+2t^3{t'}+8t^2{\!}{s'}{\!}{t'}-2t^3{t'}^2\bigr]-102+27s{s'}-119t{t'}
853: -st{s'}{t'}+6t^2{t'}^2,\\
854: c_1&=\bigl[32s+2s^2-128t-26st+60t^2+4st^2-8t^3-6s^2{s'}-7t{s'}+38st{s'}+9t^2{\!}{s'}
855: -5st^2{\!}{s'}\\
856: &-12t^3{\!}{s'}+2t^4{\!}{s'}-20t{s'}^2-8st{s'}^2+6t^2{\!}{s'}^2+2t^3{\!}{s'}^2+2st{t'}-77t^2{t'}
857: +3st^2{t'}+8t^3{t'}-29t^2{\!}{s'}{t'}\\
858: &+st^2{\!}{s'}{t'}+18t^3{\!}{s'}{t'}-2st^2{t'}^2+10t^3{t'}^2\bigr]+80-37s{s'}+145t{t'}
859: -45st{s'}{t'}+24t^2{t'}^2-8t^3{t'}^3,\\
860: c_0&=\bigl[-16s-2s^2+56t+24st+2s^2t-38t^2-8st^2+8t^3+5s^2{s'}-2t{s'}-38st{s'}-7s^2t{s'}\\
861: &+5t^2{\!}{s'}+13st^2{\!}{s'}+8t^3{\!}{s'}+2st^3{\!}{s'}-4t^4{\!}{s'}-21t{s'}^2-11st{s'}^2
862: -2t^2{\!}{s'}^2+2st^2{\!}{s'}^2+4t^3{\!}{s'}^2\\
863: &-104st{t'}-33s^2t{t'}+105t^2{t'}+35st^2{t'}+4t^3{t'}+16st^3{t'}-6t^4{t'}-2t^5{t'}
864: -s^2t{s'}{t'}+36t^2{\!}{s'}{t'}\\
865: &-14st^2{\!}{s'}{t'}-6t^3{\!}{s'}{t'}+6t^4{\!}{s'}{t'}+8t^2{\!}{s'}^2{t'}-37st^2{t'}^2
866: +22t^3{t'}^2-2st^3{t'}^2+8t^4{t'}^2+8t^3{\!}{s'}{t'}^2\\
867: &-2t^4{t'}^3\bigr]-24+14s{s'}-8s^2{s'}^2-224t{t'}+st{s'}{t'}
868: -101t^2{t'}^2-st^2{s'}{t'}^2-8t^3{t'}^3
869: \end{align*}
870: }
871: \noindent 
872: where $\bigl[a\bigr]:=a+\iota(a)$ for $a\in k(s,t,s',t')$, and 
873: $d^2\in k(s,t)$ is given by the formula 
874: \[
875: d^2=s^2-4s^3+4t-14st-30s^2t-91t^2-34st^2+s^2t^2+40t^3+24st^3+4t^4-4t^5. 
876: \]
877: 
878: We also take $\rho(P)\in k(s,t)$ which is conjugate of $P$ under the action of 
879: $F_{20}\times F_{20}'$ but not so under the action of $D_5\times D_5'$. 
880: Put 
881: \begin{align*}
882: F_{\bs,\bs'}^2(X)&:=\mathcal{RP}_{\rho(P),D_5\times D_5'}(X)
883: =F_{\rho(s),\rho(t),s',t'}^1(X)=\rho\bigl(F_{\bs,\bs'}^1(X)\bigr). 
884: \end{align*}
885: 
886: For $\ba=(a_1,a_2)\in M^2$, we denote $L_\ba:=\mathrm{Spl}_M f_\ba^{D_5}(X)$, 
887: $G_\ba:=\mathrm{Gal}(f_\ba^{D_5})$ and $G_{\ba,\ba'}:=\mathrm{Gal}(f_\ba^{D_5}f_{\ba'}^{D_5})$. 
888: We now state the result of the dihedral quintic case. 
889: %
890: \begin{theorem}[\cite{HM-3}]\label{thD5}
891: For $\ba=(a_1,a_2)$, $\ba'=(a_1',a_2')\in M^2$, we assume $\#G_\ba\geq \#G_{\ba'}>1$. 
892: An answer to the field intersection problem for $f_{s,t}^{D_5}(X)$ 
893: is given by the decomposition types {\rm DT}$(F_{\ba,\ba'}^1)$ and {\rm DT}$(F_{\ba,\ba'}^2)$ 
894: over $M$ as Table $2$ shows. 
895: \end{theorem}
896: %
897: \begin{center}
898: {\small
899: {\rm Table} $2$\vspace*{2mm}\\
900: }
901: {\footnotesize 
902: \begin{tabular}{|c|c|c|l|l|l|}\hline
903: $G_\ba$& $G_{\ba'}$ & $G_{\ba,{\ba'}}$ & & ${\rm DT}(F_{\ba,\ba'}^1)$ 
904: & ${\rm DT}(F_{\ba,\ba'}^2)$ \\ \hline 
905: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
906: & & $D_5\times D_5$ & $L_\ba\cap L_{\ba'}=M$ & $10$ & $10$ \\ \cline{3-6} 
907: & \raisebox{-1.6ex}[0cm][0cm]{$D_5$}  & $(C_5\times C_5)\rtimes C_2$ & 
908: $[L_\ba\cap L_{\ba'} : M]=2$ & $5,5$ & $5,5$\\ \cline{3-6}
909: & & \raisebox{-1.6ex}[0cm][0cm]{$D_5$} & \raisebox{-1.6ex}[0cm][0cm]{$L_\ba=L_{\ba'}$} 
910: & $5,2,2,1$ & $5,5$ \\ \cline{5-6}
911: $D_5$ & & & & $5,5$ & $5,2,2,1$ \\ \cline{2-6}
912: & $C_5$ & $D_5\times C_5$ & $L_\ba\cap L_{\ba'}=M$ & $10$ & $10$ \\ \cline{2-6}
913: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
914: & \raisebox{-1.6ex}[0cm][0cm]{$C_2$} & $D_{10}$ 
915: & $L_\ba\not\supset L_{\ba'}$ & $10$ & $10$ \\ \cline{3-6} 
916: & & $D_5$ & $L_\ba\supset L_{\ba'}$ & $5,5$ & $5,5$ \\ \cline{1-6}
917: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
918: & & $C_5\times C_5$ & $L_\ba\neq L_{\ba'}$ & $5,5$ & $5,5$\\ \cline{3-6}
919: \raisebox{-1.6ex}[0cm][0cm]{$C_5$} & $C_5$  & \raisebox{-1.6ex}[0cm][0cm]{$C_5$} 
920: & \raisebox{-1.6ex}[0cm][0cm]{$L_\ba=L_{\ba'}$} 
921: & $5,1,1,1,1,1$ & $5,5$ \\ \cline{5-6}
922: & & & & $5,5$ & $5,1,1,1,1,1$ \\ \cline{2-6}
923: & $C_2$ & $C_{10}$ & $L_\ba\cap L_{\ba'}=M$ & $10$ & $10$ \\ \hline
924: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
925: \raisebox{-1.6ex}[0cm][0cm]{$C_2$} & \raisebox{-1.6ex}[0cm][0cm]{$C_2$} & $C_2\times C_2$ 
926: & $L_\ba\neq L_{\ba'}$ & $4,4,2$ & $4,4,2$\\ \cline{3-6}
927: & & $C_2$ & $L_\ba=L_{\ba'}$ & $2,2,2,2,1,1$ & $2,2,2,2,1,1$\\ \hline
928: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
929: \end{tabular}
930: }\vspace*{0mm}
931: \end{center}
932: %
933: We checked the decomposition types on Table $2$ using the computer algebra system 
934: GAP \cite{GAP}.
935: %
936: \begin{remark}
937: In the cases of $G_\ba=D_5$, $C_2$, 
938: the quadratic subextension of $L_{s_1,t_1}$ over $M$ for $(s_1,t_1)\in M^2$ is given by 
939: {\small 
940: \begin{align*}
941: M\Bigl(\sqrt{s_1^2-4s_1^3+4t_1-14s_1t_1-30s_1^2t_1-91t_1^2-34s_1t_1^2+s_1^2t_1^2
942: +40t_1^3+24s_1t_1^3+4t_1^4-4t_1^5}\Bigr). 
943: \end{align*}
944: }
945: \end{remark}
946: %
947: \begin{remark}
948: As is well known, for a suitable integer $l$, we may have an answer to the field intersection 
949: problem via the formal $D_5\times D_5'$-relative 
950: $\langle\tau\rangle\times\langle\tau'\rangle$-invariant resolvent polynomial 
951: $G_{\bs,\bs'}(X):=\mathcal{RP}_{x+lx'\!,D_5\times D_5'}(X)$ by $x+lx'$ 
952: (cf. \cite{Coh93}, \cite{Coh00}). 
953: Although the degree of the multi-resolvent polynomials $F_{\bs,\bs'}^1(X)$ and 
954: $F_{\bs,\bs'}^1(X)$ with respect to $X$ is $10$, the degree of $G_{\bs,\bs'}(X)$ is $25$ and 
955: an explicit formula of $G_{\bs,\bs'}(X)$ in terms of $\bs$ and $\bs'$ is very complicated. 
956: We remark that to construct a suitable explicit formula is significant to investigate the 
957: structure of all $G$-Galois extensions over $M$ (cf. Section \ref{seIntro} and Corollaries 
958: \ref{cor0} and \ref{cor01}). 
959: \end{remark}
960: %
961: \begin{example}
962: Take $M=\mathbb{Q}$, $\ba=(0,3)$ and $\ba'=(10,3)$. 
963: Then we have 
964: {\small 
965: \begin{align*}
966: F_{\ba,\ba'}^1(X)&=(X+5)^3(X^2-15X+150)(X^5-625X^2-9375),\\
967: F_{\ba,\ba'}^2(X)&=\Bigl(X^5-\frac{125}{9}X^3+\frac{6250}{81}X-\frac{3125}{27}\Bigr)
968: \Bigl(X^5-\frac{125}{9}X^3+\frac{625}{9}X^2+\frac{15625}{81}X+\frac{15625}{27}\Bigr).
969: \end{align*}
970: }
971: \hspace*{-3.2mm} 
972: The decomposition type of $F_{\ba,\ba'}^1(X)$ over $\mathbb{Q}$ should be $5,2,2,1$ 
973: (cf. also Theorem \ref{thfun}); 
974: and hence we conclude that $\mathrm{Spl}_\mathbb{Q} f_\ba^{D_5}(X)=\mathrm{Spl}_\mathbb{Q} 
975: f_{\ba'}^{D_5}(X)$ and $G_\ba=D_5$. 
976: \end{example}
977: %
978: From the viewpoint of Diophantine geometry, we give some numerical examples 
979: of the field isomorphism problem of $f_{s,t}^{D_5}(X)$ over $M=\mathbb{Q}$ 
980: and for integral points $\ba,\ba'\in\mathbb{Z}^2$ using Theorem \ref{thD5} and 
981: the explicit formula (\ref{exform}). 
982: We do not know, however, for a given $\ba\in\mathbb{Z}^2$ 
983: whether there exist only finitely many $\ba'\in\mathbb{Z}^2$ such that 
984: $\mathrm{Spl}_\mathbb{Q} f_\ba^{D_5}(X)=\mathrm{Spl}_\mathbb{Q} f_{\ba'}^{D_5}(X)$ or not 
985: (cf. Corollary \ref{cor01}). 
986: 
987: %
988: \begin{example}
989: Take $M=\mathbb{Q}$ and $t:=1$. 
990: Then we have $f_{s,1}^{D_5}(X)=X^5-2X^4+(s+2)X^3-(2s+1)X^2+sX+1$. 
991: %
992: For $s_1,s_1'\in \mathbb{Z}$ in the range $-10000\leq s_1< s_1'\leq 10000$, 
993: $\mathrm{Spl}_\mathbb{Q} f_{s_1,1}^{D_5}(X)=\mathrm{Spl}_\mathbb{Q} f_{s_1',1}^{D_5}(X)$ if and 
994: only if $(s_1,s_1')\in X_1\cup X_2$ where
995: \begin{align*}
996: \vspace*{-1mm}
997: X_1&=\{(-6,0),(-1,41),(-94,-10)\},\\
998: X_2&=\{(-1,0),(-6,-1),(-18,-7),(1,34),(0,41),(-6,41),(-167,-8)\}.
999: \end{align*}
1000: It was directly checked by Theorem \ref{thD5} that, in the range $-10000\leq s_1<s_1'\leq 10000$, 
1001: $(s_1,s_1')\in X_i$ if and only if $\mathrm{DT}(F_{s_1,1,s_1',1}^i/\mathbb{Q})$ 
1002: includes $1$, for each of $i=1,2$. 
1003: \end{example}
1004: %
1005: \begin{example}
1006: Kida-Renault-Yokoyama \cite{KRY} showed that there exist infinitely many $b\in \mathbb{Q}$ 
1007: such that $\mathrm{Spl}_\mathbb{Q} f_{0,1}^{D_5}(X)=\mathrm{Spl}_\mathbb{Q} f_{b,1}^{D_5}(X)$. 
1008: Their method enables us to construct such $b$'s explicitly via rational points of an 
1009: associated elliptic curve (cf. \cite{KRY}). 
1010: They also pointed out that in the range $-400\leq s_1, t_1\leq 400$ there are $25$ pairs 
1011: $(s_1,t_1)\in \mathbb{Z}^2$ such that $\mathrm{Spl}_\mathbb{Q} f_{0,1}^{D_5}(X)
1012: =\mathrm{Spl}_\mathbb{Q} f_{s_1,t_1}^{D_5}(X)$. 
1013: We may classify the $25$ pairs by the polynomials $F_{0,1,s_1,t_1}^1(X)$ and 
1014: $F_{0,1,s_1,t_1}^2(X)$. 
1015: In the range above, for $i=1,2$, 
1016: $\mathrm{DT}(F_{0,1,s_1,t_1}^i/\mathbb{Q})$ includes $1$ if and only if 
1017: $(s_1,t_1)\in X_i$ where 
1018: \vspace*{-1mm}
1019: \begin{align*}
1020: X_1=\{&(0,1),(4,-1),(4,5),(-6,1),(-24,19),(34,11),(36,-5),\\
1021: &(46,-1),(-188,23),(264,31),(372,-5),(378,43)\},\\
1022: X_2=\{&(-1,-1),(-1,1),(5,-1),(41,1),(-43,5),(47,13),(59,-5),\\
1023: &(59,19),(101,19),(125,-23),(149,11),(155,25),(-169,55)\}.
1024: \end{align*}
1025: \vspace*{-6mm}\\
1026: By Theorem \ref{thD5}, we checked such pairs in the range $-20000\leq s_1, t_1\leq 20000$; 
1027: and added six pairs $(526,41)$, $(952,113)$, $(2302,95)$, $(6466,311)$, $(7180,143)$ and 
1028: $(7480,-169)$ to $X_1$ and just four pairs $(785,-25)$, $(3881,29)$, $(-11215,299)$ and 
1029: $(19739,-281)$ to $X_2$. 
1030: \end{example}
1031: %
1032: \begin{example}
1033: Take the $k$-generic polynomial $g_{A,B}^{C_5}(X)$ $\in$ $k(A,B)[X]$ for $C_5$ 
1034: which is constructed by Hashimoto-Tsunogai \cite{HT03}: 
1035: \vspace*{-1mm}
1036: \begin{align*}
1037: g_{A,B}^{C_5}(X)=X^5-\frac{P}{Q^2}(A^2-2A+15B^2+2)X^3+\frac{P^2}{Q^3}(2BX^2-(A-1)X-2B) 
1038: \end{align*}
1039: \vspace*{-5mm}\\
1040: where $P=(A^2-A-1)^2+25(A^2+1)B^2+125B^4$, $Q=(A+7)B^2-A+1$. 
1041: We may apply Theorem \ref{thD5} to $g_{A,B}^{C_5}(X)$ since 
1042: there exist $s_1,t_1\in k(A,B)$ such that 
1043: $\mathrm{Spl}_{k(A,B)} f_{s_1,t_1}^{D_5}(X)=\mathrm{Spl}_{k(A,B)} g_{A,B}^{C_5}(X)$ 
1044: (cf. \cite{HT03}, \cite{HM-3}). 
1045: For $\ba=(a,b)$, $\ba'=(a',b')\in \mathbb{Z}^2$, 
1046: if $\ba'=(a,\pm b)$ or $\{\ba,\ba'\}=\{(-1,\pm b),(1,\pm b)\}$ then 
1047: $\mathrm{Spl}_\mathbb{Q} g_{a,b}^{C_5}(X)=\mathrm{Spl}_\mathbb{Q} g_{a',b'}^{C_5}(X)$ 
1048: (cf. \cite{HM-3}). 
1049: For $\ba, \ba'\in \mathbb{Z}^2$ in the range $-50\leq a, a'\leq 50$, $0\leq b\leq b'\leq 50$ 
1050: with $\ba\neq \ba'$, $\{\ba,\ba'\}\neq\{(-1,b),(1,b)\}$, we see 
1051: that $\mathrm{Spl}_\mathbb{Q} g_{a,b}^{C_5}(X)=\mathrm{Spl}_\mathbb{Q} g_{a',b'}^{C_5}(X)$ 
1052: if and only if $(a,b,a',b')\in X_1\cup X_2$ where
1053: \vspace*{-1mm}
1054: \begin{align*}
1055: X_1&=\{(-3,1,-3,11),(3,3,23,3),(23,3,3,3),(7,3,27,9),\\
1056: &\qquad (2,2,-28,14),(8,11,33,14),(23,5,35,7),(41,11,-15,17)\},\\
1057: X_2&=\{(-2,1,3,2),(4,1,-6,2),(3,1,13,7),(16,2,-12,5),(-2,2,18,4),(31,1,-19,7),\\
1058: &\qquad (-3,3,-33,3),(-33,3,-3,3),(-16,13,34,19),(-2,3,43,6),(12,4,46,10)\}.
1059: \end{align*}
1060: By Theorem \ref{thD5}, it can be checked, in the range above and for each of $i=1,2$, that 
1061: $(a,b,a',b')\in X_i$ if and only if the decomposition type of $F_{\ba,\ba'}^i(X)$ over 
1062: $\mathbb{Q}$ includes $1$. 
1063: \end{example}
1064: %
1065: \begin{example}
1066: Let $h_n(X)$ be Lehmer's simplest quintic polynomial 
1067: \vspace*{-1mm}
1068: \begin{align*}
1069: h_n(X)=X^5&+n^2X^4-(2n^3+6n^2+10n+10)X^3\\
1070: &+(n^4+5n^3+11n^2+15n+5)X^2+(n^3+4n^2+10n+10)X+1
1071: \end{align*}
1072: \vspace*{-6mm}\\
1073: (cf. \cite{Leh88}), and take $M=\mathbb{Q}$. We regard $n$ as an 
1074: independent parameter over $\mathbb{Q}$. 
1075: By the result in \cite{HR}, for Brumer's quintic $f_{s,t}^{D_5}(X)$, we see that 
1076: $\mathrm{Spl}_{\mathbb{Q}(n)} h_n(X)=\mathrm{Spl}_{\mathbb{Q}(n)}f_{s,t}^{D_5}(X)$ 
1077: where $s=-20-5n+10n^2+12n^3+5n^4+n^5$, $t=-7-10n-5n^2-n^3$. 
1078: By Theorem \ref{thD5}, we checked pairs $(n, n')\in\mathbb{Z}^2$ in the range 
1079: $-10000\leq n<n'\leq 10000$ to confirm that $\mathrm{Spl}_\mathbb{Q} h_n(X)
1080: =\mathrm{Spl}_\mathbb{Q} h_{n'}(X)$ if and only if $(n,n')=(-2,-1)$. 
1081: \end{example}
1082: %
1083: \begin{acknowledgments}
1084: We thank Professor Kazuhiro Yokoyama for drawing our attention to multi-resolvent 
1085: polynomials and also for his useful suggestions. 
1086: We also thank the referee for giving fruitful comments and for careful reading of the 
1087: manuscript. 
1088: \end{acknowledgments}
1089: 
1090: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1091: 
1092: \begin{thebibliography}{Lag1770}
1093: 
1094: \bibitem[Ade01]{Ade01} C. Adelmann, {\it The decomposition of primes in torsion point fields}, 
1095: Lecture Notes in Mathematics, 1761, Springer, 2001. 
1096: 
1097: \bibitem[AV00]{AV00} P. Aubry, A. Valibouze, {\it Using Galois ideals for computing relative 
1098: resolvents}, Algorithmic methods in Galois theory. J. Symbolic Comput. \textbf{30} (2000), 
1099: 635--651. 
1100: 
1101: \bibitem[Cha96]{Cha96} R. J. Chapman, {\it Automorphism polynomials in cyclic cubic 
1102: extensions}, J. Number Theory \textbf{61} (1996), 283--291. 
1103: 
1104: \bibitem[Coh93]{Coh93} H. Cohen, {\it A course in computational algebraic number theory}, 
1105: Graduate Texts in Mathematics, 138, Springer, 1993. 
1106: 
1107: \bibitem[Coh00]{Coh00} H. Cohen, {\it Advanced topics in computational number theory}, 
1108: Graduate Texts in Mathematics, 193, Springer, 2000. 
1109: 
1110: \bibitem[Col95]{Col95} A. Colin, {\it Formal computation of Galois groups with relative 
1111: resolvents}, 169--182, Lecture Notes in Comput. Sci., 948, Springer, 1995. 
1112: 
1113: \bibitem[DeM83]{DeM83} F. R. DeMeyer, {\it Generic Polynomials}, 
1114: J. Algebra \textbf{84} (1983), 441--448. 
1115: 
1116: \bibitem[GAP]{GAP} The GAP Group, GAP --- Groups, Algorithms, and Programming, 
1117: Version 4.4.10; 2007 (http://www.gap-system.org). 
1118: 
1119: \bibitem[GK00]{GK00} K. Geissler, J. Kl\"uners, {\it Galois group computation for rational 
1120: polynomials}, Algorithmic methods in Galois theory. J. Symbolic Comput. \textbf{30} (2000), 
1121: 653--674. 
1122: 
1123: \bibitem[Gir83]{Gir83} K. Girstmair, {\it On the computation of resolvents and Galois groups}, 
1124: Manuscripta Math. \textbf{43} (1983), 289--307. 
1125: 
1126: \bibitem[GLV88]{GLV88} M. Giusti, D. Lazard, A. Valibouze, {\it Algebraic transformations 
1127: of polynomial equations, symmetric polynomials and elimination}, 
1128: Symbolic and algebraic computation (Rome, 1988), 
1129: 309--314, Lecture Notes in Comput. Sci., 358, Springer, 1989. 
1130: 
1131: \bibitem[Has00]{Has00} K. Hashimoto, {\it On Brumer's family of RM-curves of genus two}, 
1132: Tohoku Math. J. \textbf{52} (2000), 475--488. 
1133: 
1134: \bibitem[HH05]{HH05} K. Hashimoto and A. Hoshi, {\it Families of cyclic polynomials 
1135: obtained from geometric generalization of Gaussian period relations}, Math. Comp. 
1136: \textbf{74} (2005), 1519--1530. 
1137: 
1138: \bibitem[HH05-2]{HH05-2} K. Hashimoto and A. Hoshi, {\it Geometric generalization of 
1139: Gaussian period relations with application to Noether's problem for meta-cyclic groups}, 
1140: Tokyo J. Math. \textbf{28} (2005), 13--32.
1141: 
1142: \bibitem[HT03]{HT03} K. Hashimoto and H. Tsunogai, {\it Generic Polynomials over 
1143: $\mathbf{Q}$ with two parameters for the transitive groups of degree five}, 
1144: Proc. Japan. Acad. Ser. A \textbf{79} (2003), 142--145. 
1145: 
1146: \bibitem[HM07]{HM07} A. Hoshi, K. Miyake, {\it Tschirnhausen transformation of a cubic 
1147: generic polynomial and a $2$-dimensional involutive Cremona transformation}, 
1148: Proc. Japan Acad. Ser. A \textbf{83} (2007), 21--26.
1149: 
1150: \bibitem[HM]{HM} A. Hoshi, K. Miyake, {\it A geometric framework for the subfield problem 
1151: of generic polynomials via Tschirnhausen transformation}, preprint. arXiv:0710.0287v1 [math.NT]
1152: 
1153: \bibitem[HM-2]{HM-2} A. Hoshi, K. Miyake, {\it On the field intersection problem of 
1154: quartic generic polynomials via formal Tschirnhausen transformation}, preprint. 
1155: arXiv:0812.4807v1 [math.NT]
1156: 
1157: \bibitem[HM-3]{HM-3} A. Hoshi, K. Miyake, {\it On the field intersection problem of 
1158: solvable quintic generic polynomials}, preprint. arXiv:0804.4875v1 [math.NT] 
1159: 
1160: \bibitem[HR]{HR} A. Hoshi, Y. Rikuna, {\it On a transformation from dihedral quintic 
1161: polynomials into Brumer's form}, in preparation. 
1162: 
1163: \bibitem[JLY02]{JLY02} C. Jensen, A. Ledet, N. Yui, {\it Generic polynomials. Constructive 
1164: aspects of the inverse Galois problem}, Mathematical Sciences Research Institute Publications, 
1165: 45, Cambridge, 2002. 
1166: 
1167: \bibitem[Kem94]{Kem94} G. Kemper, {\it Das Noethersche Problem und generische Polynome}, 
1168: Dissertation, University of Heidelberg, 1994, also available as: IWR Preprint 94-49, Heidelberg. 
1169: 
1170: \bibitem[KM00]{KM00} G. Kemper, E. Mattig, {\it Generic polynomials with few parameters}, 
1171: Algorithmic methods in Galois theory. J. Symbolic Comput. \textbf{30} (2000), 843--857. 
1172: 
1173: \bibitem[Kem01]{Kem01} G. Kemper, {\it Generic polynomials are descent-generic}, 
1174: Manuscripta Math. \textbf{105} (2001), 139--141. 
1175: 
1176: \bibitem[Kid05]{Kid05} M. Kida, {\it Kummer theory for norm algebraic tori}, 
1177: J. Algebra \textbf{293} (2005), 427--447. 
1178: 
1179: \bibitem[Kid06]{Kid06} M. Kida, {\it Cyclic polynomials arising from Kummer theory of norm 
1180: algebraic tori}, Algorithmic number theory, 102--113, 
1181: Lecture Notes in Comput. Sci., 4076, Springer, 2006. 
1182: 
1183: \bibitem[KRY]{KRY} M. Kida, G. Renault, K. Yokoyama, {\it Quintic polynomials of 
1184: Hashimoto-Tsunogai, Brumer, and Kummer}, to appear in Int. J. Number Theory. 
1185: 
1186: \bibitem[Kom]{Kom} T. Komatsu, {\it Generic sextic polynomial related to the subfield problem 
1187: of a cubic polynomial}, preprint. {\tt http://www.math.kyushu-u.ac.jp/coe/report/pdf/2006-9.pdf}
1188: 
1189: \bibitem[Lag1770]{Lag1770} J. Lagrange, {\it R\'eflexions sur la r\'esolution alg\'ebrique 
1190: des \'equations}, M\'emoires de l'Acad\'emie de Berlin (1770), 205--421. 
1191: 
1192: \bibitem[Lan78]{Lan78} S. Lang, {\it Elliptic curves: Diophantine analysis}, 
1193: Grundlehren der Mathematischen Wissenschaften 231, Springer, 1978.
1194: 
1195: \bibitem[Leh88]{Leh88} E. Lehmer, {\it Connection between Gaussian periods and cyclic units}, 
1196: Math. Comp. \textbf{50} (1988), 535--541. 
1197: 
1198: \bibitem[MM97]{MM97} T. Mattman, J. McKay, {\it Computation of Galois groups over function 
1199: fields}, Math. Comp. \textbf{66} (1997), 823--831. 
1200: 
1201: \bibitem[Mor94]{Mor94} P. Morton, {\it Characterizing cyclic cubic extensions by automorphism 
1202: polynomials}, J. Number Theory \textbf{49} (1994), 183--208. 
1203: 
1204: \bibitem[RV99]{RV99} N. Rennert, A. Valibouze, {\it Calcul de r\'esolvantes avec les modules 
1205: de Cauchy}, Experiment. Math. \textbf{8} (1999), 351--366. 
1206: 
1207: \bibitem[Rik04]{Rik04} Y. Rikuna, {\it Explicit constructions of generic polynomials for some 
1208: elementary groups}, Galois theory and modular forms, 173--194, Dev. Math., 11, Kluwer Acad. 
1209: Publ., Boston, MA, 2004.
1210: 
1211: \bibitem[SM85]{SM85} L. Soicher, J. McKay, {\it Computing Galois groups over the rationals}, 
1212: J. Number Theory \textbf{20} (1985), 273--281. 
1213: 
1214: \bibitem[Sta73]{Sta73} R. P. Stauduhar, {\it The determination of Galois groups}, 
1215: Math. Comp. \textbf{27} (1973), 981--996. 
1216: 
1217: \bibitem[Val95]{Val95} A. Valibouze, {\it Computation of the Galois groups of the resolvent 
1218: factors for the direct and inverse Galois problems}, 
1219: 456--468, Lecture Notes in Comput. Sci., 948, Springer, 1995. 
1220: 
1221: \bibitem[Val]{Val} A. Valibouze, {\it Galois theory and reducible polynomials}. 
1222: Available from {\tt http://} 
1223: {\tt www.lix.polytechnique.fr/\verb+~+max/publications/valibouze/jnt99.dvi}
1224: 
1225: \bibitem[Wol03]{Wol03} S. Wolfram, {\it The Mathematica Book, Fifth Edition}, Wolfram Media, 
1226: Inc., Cambridge University Press, 2003.
1227: 
1228: \bibitem[Yok97]{Yok97} K. Yokoyama, {\it A modular method for computing the Galois groups 
1229: of polynomials}, J. Pure Appl. Algebra 117/118 (1997), 617--636. 
1230: 
1231: \end{thebibliography}
1232: \end{document}
1233: 
1234: %%
1235: {\small 
1236: \hspace*{-0.5cm}\\ 
1237: \begin{tabular}{ll}
1238: Akinari HOSHI & Katsuya MIYAKE\\ 
1239: Department of Mathematics & Department of Mathematics\\ 
1240: Rikkyo University & School of Fundamental Science and Engineering\\ 
1241: 3--34--1 Nishi-Ikebukuro Toshima-ku & Waseda University\\
1242: Tokyo, 171--8501, Japan & 3--4--1 Ohkubo Shinjuku-ku\\
1243: E-mail: \texttt{hoshi@rikkyo.ac.jp} & Tokyo, 169--8555, Japan\\
1244: & E-mail: \texttt{miyakek@aoni.waseda.jp}
1245: \end{tabular}
1246: }
1247: