Difference between revisions of "ApCoCoA-1:NC.SetOrdering"

From ApCoCoAWiki
Line 8: Line 8:
 
</syntax>
 
</syntax>
 
<description>
 
<description>
Note that the default ordering is <quotes>LLEX</quotes> (length-lexicographic ordering).
+
Note that a <em>word ordering</em> is a well-ordering which is compatible with multiplication. The default ordering is <quotes>LLEX</quotes> (the length-lexicographic ordering).
 
<itemize>
 
<itemize>
<item>@param <em>Ordering</em>: a string which indicates an (admissible) ordering. For the time being, the package supports <quotes>LLEX</quotes> (length-lexicographic ordering), <quotes>ELIM</quotes> (elimination ordering) and <quotes>LRLEX</quotes> (length-reverse-lexicographic ordering).</item>
+
<item>@param <em>Ordering</em>: a string which indicates a word ordering. For the time being, the package supports <quotes>LLEX</quotes> (the length-lexicographic ordering), <quotes>ELIM</quotes> (an elimination ordering),<quotes>LRLEX</quotes> (the length-reverse-lexicographic ordering), and <quotes>DEGREVLEX</quotes> (the degree-reverse-lexicographic ordering).</item>
 
</itemize>
 
</itemize>
Let <tt>X=x_{1}x_{2}...x_{n}</tt>. We define the (left-to-right) lexicographic ordering <quotes>LEX</quotes> on <tt>&lt;X&gt;</tt> as follows. For two words <tt>W1, W2</tt> in <tt>&lt;X&gt;</tt>, we say <tt>W1&gt;_{Lex}W2</tt> if we have <tt>W1=W2*W</tt> for some non-empty word <tt>W</tt> in <tt>&lt;X&gt;</tt>, or if we have <tt>W1=W*x_{i}*W3, W2=W*x_{j}*W4</tt> for some words <tt>W,W3,W4</tt> in <tt>&lt;X&gt;</tt> and some letters <tt>x_{i},x_{j}</tt> in <tt>X</tt> such that <tt>i&lt;j</tt>. Thus, we have <tt>x_{1}&gt;_{LEX}x_{2}&gt;_{LEX}...&gt;_{LEX}x_{n}</tt>. Note that <quotes>LEX</quotes> is not an admissible ordering on <tt>&lt;X&gt;</tt>. We define admissible orderings <quotes>LLEX</quotes>, <quotes>ELIM</quotes> and <quotes>LRLEX</quotes> on <tt>&lt;X&gt;</tt> as follows.
+
<par/>
 +
In the following, we let <tt>Wn</tt> be the monoid of all words generated by <tt>{x[1],x[2],...,x[n]}</tt>. We define the (left-to-right) lexicographic ordering <quotes>LEX</quotes> on <tt>Wn</tt> as follows. For two words <tt>W, W'</tt> in <tt>Wn</tt>, we say <tt>W&gt;_{Lex}W'</tt> if we have <tt>W=W'W_{1}</tt> for some non-empty word <tt>W_{1}</tt> in <tt>Wn</tt>, or if we have <tt>W=W_{1}x[i]W_{2}, W'=W_{1}x[j]W_{3}</tt> for some words <tt>W_{1},W_{2},W_{3}</tt> in <tt>Wn</tt> and <tt>i&lt;j</tt>. Thus, we have <tt>x[1]&gt;_{LEX}x[2]&gt;_{LEX}...&gt;_{LEX}x[n]</tt>. Note that <quotes>LEX</quotes> is not a word ordering on <tt>Wn</tt>. Given two words <tt>W, W'</tt> in <tt>Wn</tt>, we define word orderings <quotes>LLEX</quotes>, <quotes>ELIM</quotes>, <quotes>LRLEX</quotes>, and <quotes>DEGREVLEX</quotes> on <tt>Wn</tt> as follows.
 
<itemize>
 
<itemize>
<item><quotes>LLEX</quotes>: for two words <tt>W1, W2</tt> in <tt>&lt;X&gt;</tt>, we say <tt>W1&gt;_{LLEX}W2</tt> if <tt>len(W1)&gt;len(W2)</tt>, or <tt>len(W1)=len(W2)</tt> and <tt>W1</tt> is lexicographically larger than <tt>W2</tt>.</item>
+
<item><quotes>LLEX</quotes>: we say <tt>W&gt;_{LLEX}W'</tt> if <tt>len(W)&gt;len(W')</tt>, or <tt>len(W)=len(W')</tt> and <tt>W</tt> is lexicographically larger than <tt>W'</tt>.</item>
<item><quotes>ELIM</quotes>: it first compares the associated commutative terms lexicographically and then breaks ties using the non-commutative lexicographic orderin with respect to <tt>x_{1}&gt;...&gt;x_{n}</tt>. That is, for two words <tt>W1, W2</tt> in <tt>&lt;X&gt;</tt>, we say <tt>W1&gt;_{ELIM}W2</tt> if <tt>W1</tt> is lexicographically larger than <tt>W2</tt> by considering <tt>W1, W2</tt> as two terms in the commutative case, or <tt>W1=W2</tt> by considering <tt>W1, W2</tt> as two terms in the commutative case and <tt>W1&gt;_{Lex}W2</tt> (<tt>W1</tt> is left-to-right lexicographically larger than <tt>W2</tt> by considering <tt>W1, W2</tt> as two words in the non-commutative case). Thus, the elimination ordering <quotes>ELIM</quotes> first eliminates the letter <tt>x_{1}</tt>, and then <tt>x_{2}</tt>, and then <tt>x_{3}</tt>, and so on and so forth.</item>
+
 
<item><quotes>LRLEX</quotes>: for two words <tt>W1, W2</tt> in <tt>&lt;X&gt;</tt>, we say <tt>W1&gt;_{LRLEX}W2</tt> if <tt>len(W1)&gt;len(W2)</tt>, or <tt>len(W1)=len(W2)</tt> and <tt>W1</tt> is larger than <tt>W2</tt> by right-to-left lexicographic ordering.</item>
+
<item><quotes>ELIM</quotes>: it first compares the associated commutative terms lexicographically and then breaks ties using the non-commutative lexicographic ordering with respect to <tt>x[1]&gt;_{LEX}...&gt;_{LEX}x[n]</tt>. That is, we say <tt>W&gt;_{ELIM}W'</tt> if <tt>W</tt> is lexicographically larger than <tt>W'</tt> by considering them as two terms in the commutative case, or <tt>W=W'</tt> as two commutative terms and <tt>W&gt;_{Lex}W'</tt> (<tt>W</tt> is lexicographically larger than <tt>W'</tt> by considering them as two words in the non-commutative case). Thus, the elimination ordering <quotes>ELIM</quotes> first eliminates the indeterminate <tt>x[1]</tt>, and then <tt>x[2]</tt>, and then <tt>x[3]</tt>, and so on and so forth.</item>
 +
 
 +
<item><quotes>LRLEX</quotes>: we say <tt>W&gt;_{LRLEX}W'</tt> if <tt>len(W)&gt;len(W')</tt>, or <tt>len(W)=len(W')</tt> and <tt>W</tt> is larger than <tt>W'</tt> by the right-to-left lexicographic ordering.</item>
 
</itemize>
 
</itemize>
An admissible ordering on is called <em>length compatible</em> if <tt>len(W1)>len(W2)</tt> implies <tt>W1</tt> is larger than <tt>W2</tt> for all <tt>W1, W2</tt> in <tt>&lt;X&gt;</tt>. For instance, <quotes>LLEX</quotes> and <quotes>LRLEX</quotes> are length compatible while <quotes>ELIM</quotes> is not.
+
A word ordering on is said to be <em>length compatible</em> if <tt>len(W)>len(W')</tt> implies <tt>W</tt> is larger than <tt>W'</tt> for all <tt>W, W'</tt> in <tt>Wn</tt>. For instance, <quotes>LLEX</quotes> and <quotes>LRLEX</quotes> are length compatible while <quotes>ELIM</quotes> is not.
 
<example>
 
<example>
 
NC.RingEnv();
 
NC.RingEnv();

Revision as of 19:02, 25 April 2013

NC.SetOrdering

Set a word ordering on the monoid of all words in a non-commutative polynomial ring.

Syntax

NC.SetOrdering(Ordering:STRING)

Description

Note that a word ordering is a well-ordering which is compatible with multiplication. The default ordering is "LLEX" (the length-lexicographic ordering).

  • @param Ordering: a string which indicates a word ordering. For the time being, the package supports "LLEX" (the length-lexicographic ordering), "ELIM" (an elimination ordering),"LRLEX" (the length-reverse-lexicographic ordering), and "DEGREVLEX" (the degree-reverse-lexicographic ordering).

In the following, we let Wn be the monoid of all words generated by {x[1],x[2],...,x[n]}. We define the (left-to-right) lexicographic ordering "LEX" on Wn as follows. For two words W, W' in Wn, we say W>_{Lex}W' if we have W=W'W_{1} for some non-empty word W_{1} in Wn, or if we have W=W_{1}x[i]W_{2}, W'=W_{1}x[j]W_{3} for some words W_{1},W_{2},W_{3} in Wn and i<j. Thus, we have x[1]>_{LEX}x[2]>_{LEX}...>_{LEX}x[n]. Note that "LEX" is not a word ordering on Wn. Given two words W, W' in Wn, we define word orderings "LLEX", "ELIM", "LRLEX", and "DEGREVLEX" on Wn as follows.

  • "LLEX": we say W>_{LLEX}W' if len(W)>len(W'), or len(W)=len(W') and W is lexicographically larger than W'.

  • "ELIM": it first compares the associated commutative terms lexicographically and then breaks ties using the non-commutative lexicographic ordering with respect to x[1]>_{LEX}...>_{LEX}x[n]. That is, we say W>_{ELIM}W' if W is lexicographically larger than W' by considering them as two terms in the commutative case, or W=W' as two commutative terms and W>_{Lex}W' (W is lexicographically larger than W' by considering them as two words in the non-commutative case). Thus, the elimination ordering "ELIM" first eliminates the indeterminate x[1], and then x[2], and then x[3], and so on and so forth.

  • "LRLEX": we say W>_{LRLEX}W' if len(W)>len(W'), or len(W)=len(W') and W is larger than W' by the right-to-left lexicographic ordering.

A word ordering on is said to be length compatible if len(W)>len(W') implies W is larger than W' for all W, W' in Wn. For instance, "LLEX" and "LRLEX" are length compatible while "ELIM" is not.

Example

NC.RingEnv();
Coefficient ring : Q
Ordering : LLEX
-------------------------------
NC.SetOrdering(<quotes>ELIM</quotes>);
NC.RingEnv();
Coefficient ring : Q
Ordering : ELIM
-------------------------------