<h4>3.1 <span class="Heading">Notation and definitions</span></h4>
<p>Let <span class="SimpleMath">\(G\)</span> be a group and <span class="SimpleMath">\(F\)</span> a field. Then the <em>group algebra</em> <span class="SimpleMath">\(FG\)</span> consists of the set of formal linear combinations of the form</p>
<p class="center">\[
\sum_{g \in G}\alpha_g g,\qquad \alpha_g \in F
\]</p>
<p>where all but finitely many of the <span class="SimpleMath">\(\alpha_g\)</span> are zero. The group algebra <span class="SimpleMath">\(FG\)</span> is an <span class="SimpleMath">\(F\)</span>-algebra with the obvious operations. Clearly, <span class="SimpleMath">\(\dim FG=|G|\)</span>.</p>
<p>The <em>augmentation homomorphism</em> <span class="SimpleMath">\( \chi : FG \rightarrow F\)</span> is defined by</p>
<p>It is easy to see that <span class="SimpleMath">\(\chi\)</span> is indeed a homomorphism onto <span class="SimpleMath">\(F\)</span>. The kernel of <span class="SimpleMath">\(\chi\)</span> is called the <em>augmentation ideal</em> of <span class="SimpleMath">\(FG\)</span>. The augmentation ideal is denoted <span class="SimpleMath">\(A(FG)\)</span>, or simply <span class="SimpleMath">\(A\)</span> when there is no danger of confusion. It follows from the isomorphism theorems that <span class="SimpleMath">\(\dim A(FG)=\dim FG-1=|G|-1\)</span>. Another way to write the augmentation ideal is</p>
<p>An invertible element of <span class="SimpleMath">\(FG\)</span> is said to be a <em>unit</em>. Clearly the elements of <span class="SimpleMath">\(G\)</span> and the non-zero elements of <span class="SimpleMath">\(F\)</span> are units. The set of units in <span class="SimpleMath">\(FG\)</span> is a group with respect to the multiplication of <span class="SimpleMath">\(FG\)</span>. The <em>unit group</em> of <span class="SimpleMath">\(FG\)</span> is denoted <span class="SimpleMath">\(U(FG)\)</span> or simply <span class="SimpleMath">\(U\)</span> when there is no risk of confusion. A unit <span class="SimpleMath">\(u\)</span> is said to be <em>normalised</em> if <span class="SimpleMath">\(\chi(u)=1\)</span>. The set of normalised units forms a subgroup of the unit group, and is referred to as the <em>normalised unit group</em>. The normalised unit group of <span class="SimpleMath">\(FG\)</span> is denoted <span class="SimpleMath">\(V(FG)\)</span>, or simply <span class="SimpleMath">\(V\)</span>. It is easy to prove that <span class="SimpleMath">\(U(FG) = F^* \times V(FG)\)</span> where <span class="SimpleMath">\(F^*\)</span> denotes the multiplicative group of <spanclass="SimpleMath">\(F\)</span>.</p>
<h4>3.2 <span class="Heading"><span class="SimpleMath">\(p\)</span>-modular group algebras</span></h4>
<p>A group algebra <span class="SimpleMath">\(FG\)</span> is said to be <span class="SimpleMath">\(p\)</span>-modular if <span class="SimpleMath">\(F\)</span> is the field of characteristic <spanclass="SimpleMath">\(p\)</span>, and <span class="SimpleMath">\(G\)</span> is a finite <span class="SimpleMath">\(p\)</span>-group. A lot of information about the structure of <span class="SimpleMath">\(p\)</span>-modular group algebras can be found in <a href="chapBib_mj.html#biBHB">[HB82, Chapter VIII]</a>. In a <span class="SimpleMath">\(p\)</span>-modular group algebra we have that an element <span class="SimpleMath">\(u\)</span> is a unit if and only if <span class="SimpleMath">\(\chi(u)\neq 0\)</span>. Hence the normalised unit group <span class="SimpleMath">\(V\)</span> consists of all elements of <span class="SimpleMath">\(FG\)</span> with augmentation <span class="SimpleMath">\(1\)</span>. In other words <span class="SimpleMath">\(V\)</span> is a coset of the augmentation ideal, namely <span class="SimpleMath">\(V=1+A\)</span>. This also implies that <span class="SimpleMath">\(|V|=|A|=|F|^{|G|-1}\)</span>, and so <span class="SimpleMath">\(V\)</span> is a finite <span class="SimpleMath">\(p\)</span>-group.</p>
<p>One of the aims of the <strong class="pkg">LAGUNA</strong> package is to compute a power-commutator presentation for the normalised unit group in the case when <span class="SimpleMath">\(G\)</span> is a finite <span class="SimpleMath">\(p\)</span>-group and <span class="SimpleMath">\(F\)</span> is a field of <span class="SimpleMath">\(p\)</span> elements. Such a presentation is given by generators <span class="SimpleMath">\(y_1, \ldots, y_{|G|-1} \)</span> and two types of relations: <span class="SimpleMath">\(y_i^p=(y_{i+1})^{\alpha_{i,i+1}} \cdots (y_{|G|-1})^{\alpha_{i,|G|-1}}\)</span> for <span class="SimpleMath">\( 1 \leq i \leq |G|-1 \)</span>, and <span class="SimpleMath">\( [y_j,y_i]=(y_{j+1})^{\alpha_{j,i,j+1}} \cdots (y_{|G|-1})^{\alpha_{j,i,|G|-1}} \)</span> for <span class="SimpleMath">\( 1 \leq i < j \leq |G|-1\)</span>, where the exponents <span class="SimpleMath">\(\alpha_{i,k}\)</span> and <span class="SimpleMath">\(\alpha_{i,j,k}\)</span> are elements of the set <span class="SimpleMath">\(\{0,\ldots,p-1\}\)</span>. Having such a presentation, it is possible to carry out efficient computations in the finite <span class="SimpleMath">\(p\)</span>-group <span class="SimpleMath">\(V\)</span>; see <a href="chapBib_mj.html#biBSims">[Sim94, Chapter 9]</a>.</p>
<h4>3.3 <span class="Heading">Polycyclic generating set for <span class="SimpleMath">\(V\)</span></span></h4>
<p>Let <span class="SimpleMath">\(G\)</span> be a finite <span class="SimpleMath">\(p\)</span>-group and <span class="SimpleMath">\(F\)</span> the field of <span class="SimpleMath">\(p\)</span> elements. Our aim is to construct a power-commutator presentation for <span class="SimpleMath">\(V=V(FG)\)</span>. We noted earlier that <span class="SimpleMath">\(V=1+A\)</span>, where <spanclass="SimpleMath">\(A\)</span> is the augmentation ideal. We use this piece of information and construct a polycyclic generating set for <span class="SimpleMath">\(V\)</span> using a suitable basis for <span class="SimpleMath">\(A\)</span>. Before constructing this generating set, we note that <span class="SimpleMath">\(A\)</span> is a nilpotent ideal in <span class="SimpleMath">\(FG\)</span>. In other words there is some <span class="SimpleMath">\(c\)</span> such that <span class="SimpleMath">\(A^c\neq 0\)</span> but <span class="SimpleMath">\(A^{c+1}=0\)</span>. Hence we can consider the following series of ideals in <span class="SimpleMath">\(A\)</span>:</p>
<p>It is clear that a quotient <span class="SimpleMath">\(A^i/A^{i+1}\)</span>of this chain has trivial multiplication, that is, such a quotient is a nil-ring. The chain <span class="SimpleMath">\(A^i\)</span> gives rise to a series of normal subgroups in <span class="SimpleMath">\(V\)</span>:</p>
<p>It is easy to see that the chain <span class="SimpleMath">\(1+A^i\)</span> is central, that is, <span class="SimpleMath">\((1+A^i)/(1+A^{i+1})\leq Z((1+A)/(1+A^{i+1}))\)</span>.</p>
<p>Now we show how to compute a basis for <span class="SimpleMath">\(A^i\)</span> that gives a polycyclic generating set for <span class="SimpleMath">\(1+A^i\)</span>. Let</p>
<p>be the <em>Jennings series</em> of <span class="SimpleMath">\(G\)</span>. That is, <span class="SimpleMath">\(G_{i+1}=[G_i,G]G_{j^p}\)</span> where <span class="SimpleMath">\(j\)</span> is the smallest non-negative integer such that <span class="SimpleMath">\(j\geq i/p\)</span>. For all <span class="SimpleMath">\(i\leq k\)</span> select elements <span class="SimpleMath">\(x_{i,1},\ldots,x_{i,l_i}\)</span> of <span class="SimpleMath">\(G_i\)</span> such that <span class="SimpleMath">\(\{x_{i,1}G_{i+1},\ldots,x_{i,l_i}G_{i+1}\}\)</span> is a minimal generating set for the elementary abelian group <span class="SimpleMath">\(G_i/G_{i+1}\)</span>. For the Jennings series it may happen that <span class="SimpleMath">\(G_i=G_{i+1}\)</span> for some <span class="SimpleMath">\(i\)</span>. In this case we choose an empty generating set for the quotient <span class="SimpleMath">\(G_i/G_{i+1}\)</span> and <span class="SimpleMath">\(l_i=0\)</span>. Then the set <span class="SimpleMath">\(x_{1,1},\ldots,x_{1,l_1},\ldots,x_{k,1},\ldots,x_{k,l_k}\)</span> is said to be a <em>dimension basis</em> for <span class="SimpleMath">\(G\)</span>. The <em>weight</em> of a dimension basis element <span class="SimpleMath">\(x_{i,j}\)</span> is <span class="SimpleMath">\(i\)</span>.</p>
<p>where <span class="SimpleMath">\(0\leq \alpha_{i,j}\leq p-1\)</span> is said to be <em>standard</em>. Clearly, a standard product is an element of the augmentation ideal <span class="SimpleMath">\(A\)</span>. The weight of the standard product <span class="SimpleMath">\(u\)</span> is</p>
<p>The total number of standard products is <span class="SimpleMath">\(|G|-1\)</span> .</p>
<p><strong class="button">Lemma (</strong><a href="chapBib_mj.html#biBHB">[HB82, Theorem VIII.2.6]</a><strong class="button">).</strong> For <span class="SimpleMath">\(i\leq c\)</span>, the set <span class="SimpleMath">\(S_i\)</span> of standard products of weight at least <span class="SimpleMath">\(i\)</span> forms a basis for <span class="SimpleMath">\(A^i\)</span>. Moreover, the set <span class="SimpleMath">\(1+S_i=\{1+s\ |\ s \in S_i\}\)</span> is a polycyclic generating set for <span class="SimpleMath">\(1+A^i\)</span>. In particular <span class="SimpleMath">\(1+S_1\)</span> is a polycyclic generating set for <span class="SimpleMath">\(V\)</span>.</p>
<p>A basis for <span class="SimpleMath">\(A\)</span> consisting of the standard products is referred to as a <em>weighted basis</em>. Note that a weighted basis is a basis for the augmentation ideal, and not for the whole group algebra.</p>
<p>Let <span class="SimpleMath">\(x_1,\ldots,x_{{|G|}-1}\)</span> denote the standard products where we choose the indices so that the weight of <span class="SimpleMath">\(x_i\)</span> is not larger than the weight of <span class="SimpleMath">\(x_{i+1}\)</span> for all <span class="SimpleMath">\(i\)</span>, and set <span class="SimpleMath">\(y_i=1+x_i\)</span>. Then every element <span class="SimpleMath">\(v\)</span> of <span class="SimpleMath">\(V\)</span> can be uniquely written in the form</p>
<p>This expression is called the <em>canonical form</em> of <span class="SimpleMath">\(v\)</span>. We note that by adding a generator of <span class="SimpleMath">\(F^*\)</span> to the set <span class="SimpleMath">\(y_1,\ldots,y_{|G|-1|}\)</span> we can obtain a polycyclic generating set for the unit group <span class="SimpleMath">\(U\)</span>.</p>
<h4>3.4 <span class="Heading">Computing the canonical form</span></h4>
<p>We show how to compute the canonical form of a normalised unit with respect to the polycyclic generating set <span class="SimpleMath">\(y_1,\ldots,y_{|G|-1}\)</span>. We use the following elementary lemma.</p>
<p><strong class="button">Lemma.</strong> Let <span class="SimpleMath">\(i\leq c\)</span> and suppose that <span class="SimpleMath">\(w \in A^i\)</span>. Assume that <span class="SimpleMath">\(x_{s_i},x_{s_i+1}\ldots,x_{r_i}\)</span> are the standard products with weight <span class="SimpleMath">\(i\)</span> and for <span class="SimpleMath">\(s_i\leq j\leq r_i\)</span> set <span class="SimpleMath">\(y_j=1+x_j\)</span>. Then for all <span class="SimpleMath">\(\alpha_{s_i},\ldots,\alpha_{r_i}\in\{0,\ldots,p-1\}\)</span> we have that</p>
<p>Suppose that <span class="SimpleMath">\(w\)</span> is an element of the augmentation ideal <span class="SimpleMath">\(A\)</span> and <span class="SimpleMath">\(1+w\)</span> is a normalised unit. Let <span class="SimpleMath">\(x_1,\ldots,x_{r_1}\)</span> be the standard products of weight 1, and let <span class="SimpleMath">\(y_1,\ldots,y_{r_1}\)</span> be the corresponding elements in the polycyclic generating set. Then using the previous lemma, we find <span class="SimpleMath">\(\alpha_1,\ldots,\alpha_{r_1}\)</span> such that</p>
<p>Now we have that <span class="SimpleMath">\(1+w=(y_{1})^{\alpha_{1}}\cdots (y_{r_1})^{\alpha_{r_1}}(1+w_2)\)</span> for some <span class="SimpleMath">\(w_2 \in A^2\)</span>. Then suppose that <span class="SimpleMath">\(x_{s_2},x_{s_2+1},\ldots,x_{r_2}\)</span> are the standard products of weight 2. We find <span class="SimpleMath">\(\alpha_{s_2},\ldots,\alpha_{r_2}\)</span> such that</p>
<p>Thus <span class="SimpleMath">\(1+w_2=(y_{s_2})^{\alpha_{s_2}}\cdots (y_{r_2})^{\alpha_{r_2}}(1+w_3)\)</span> for some <span class="SimpleMath">\(w_3 \in A^3\)</span>, and so <span class="SimpleMath">\(1+w=(y_{1})^{\alpha_{1}}\cdots (y_{r_1})^{\alpha_{r_1}}(y_{s_2})^{\alpha_{s_2}}\cdots (y_{r_2})^{\alpha_{r_2}}(1+w_3)\)</span>. We repeat this process, and after <span class="SimpleMath">\(c\)</span> steps we obtain the canonical form for the element <span class="SimpleMath">\(1+w\)</span>.</p>
<h4>3.5 <span class="Heading">Computing a power commutator presentation for <span class="SimpleMath">\(V\)</span></span></h4>
<p>Using the procedure in the previous section, it is easy to compute a power commutator presentation for the normalized unit group <span class="SimpleMath">\(V\)</span> of a <span class="SimpleMath">\(p\)</span>-modular group algebra over the field of <span class="SimpleMath">\(p\)</span> elements. First we compute the polycyclic generating sequence <span class="SimpleMath">\(y_1,\ldots,y_{|G|-1}\)</span> as in Section <a href="chap3_mj.html#X7B74767A849D921E"><span class="RefLink">3.3</span></a>. Then for each <span class="SimpleMath">\(y_i\)</span> and for each <span class="SimpleMath">\(y_j,\ y_i\)</span> such that <span class="SimpleMath">\(i<j\)</span> we compute the canonical form for <span class="SimpleMath">\(y_i^p\)</span> and <span class="SimpleMath">\([y_j,y_i]\)</span> as described in Section <a href="chap3_mj.html#X7DCB12F48752C17E"><span class="RefLink">3.4</span></a>.</p>
<p>Once a power-commutator presentation for <span class="SimpleMath">\(V\)</span> is constructed, it is easy to obtain a polycyclic presentation for the unit group <span class="SimpleMath">\(U\)</span> by adding an extra central generator <span class="SimpleMath">\(y\)</span> corresponding to a generator of the cyclic group <span class="SimpleMath">\(F^*\)</span> and enforcing that <span class="SimpleMath">\(y^{p-1}=1\)</span>.</p>
<h4>3.6 <span class="Heading">Verifying Lie properties of <span class="SimpleMath">\(FG\)</span></span></h4>
<p>If <span class="SimpleMath">\(FG\)</span> is a group algebra then one can consider the Lie bracket operation defined by <span class="SimpleMath">\([a,b]=ab-ba\)</span>. Then it is well-known that <span class="SimpleMath">\(FG\)</span> with respect to the scalar multiplication, the addition, and the bracket operation becomes a Lie algebra over <span class="SimpleMath">\(F\)</span>. This Lie algebra is also denoted <span class="SimpleMath">\(FG\)</span>. Some Lie properties of such Lie algebras can be computed very efficiently. In particular, it can be verified whether the Lie algebra <span class="SimpleMath">\(FG\)</span> is nilpotent, soluble, metabelian, centre-by-metabelian. Fast algorithms that achieve these goals are described in <a href="chapBib_mj.html#biBLR86">[LR86]</a>, <a href="chapBib_mj.html#biBPPS73">[PPS73]</a>, and <a href="chapBib_mj.html#biBRos00">[Ros00]</a>.</p>
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.