[BACK]Return to groebner.texi CVS log [TXT][DIR] Up to [local] / OpenXM / src / asir-doc / parts

Diff for /OpenXM/src/asir-doc/parts/groebner.texi between version 1.11 and 1.15

version 1.11, 2003/04/28 06:43:10 version 1.15, 2004/09/14 02:28:20
Line 1 
Line 1 
 @comment $OpenXM: OpenXM/src/asir-doc/parts/groebner.texi,v 1.10 2003/04/28 03:09:23 noro Exp $  @comment $OpenXM: OpenXM/src/asir-doc/parts/groebner.texi,v 1.14 2004/09/14 01:32:34 noro Exp $
 \BJP  \BJP
 @node $B%0%l%V%J4pDl$N7W;;(B,,, Top  @node $B%0%l%V%J4pDl$N7W;;(B,,, Top
 @chapter $B%0%l%V%J4pDl$N7W;;(B  @chapter $B%0%l%V%J4pDl$N7W;;(B
Line 15 
Line 15 
 * $B4pK\E*$JH!?t(B::  * $B4pK\E*$JH!?t(B::
 * $B7W;;$*$h$SI=<($N@)8f(B::  * $B7W;;$*$h$SI=<($N@)8f(B::
 * $B9`=g=x$N@_Dj(B::  * $B9`=g=x$N@_Dj(B::
   * Weight::
 * $BM-M}<0$r78?t$H$9$k%0%l%V%J4pDl7W;;(B::  * $BM-M}<0$r78?t$H$9$k%0%l%V%J4pDl7W;;(B::
 * $B4pDlJQ49(B::  * $B4pDlJQ49(B::
 * Weyl $BBe?t(B::  * Weyl $BBe?t(B::
Line 26 
Line 27 
 * Fundamental functions::  * Fundamental functions::
 * Controlling Groebner basis computations::  * Controlling Groebner basis computations::
 * Setting term orderings::  * Setting term orderings::
   * Weight::
 * Groebner basis computation with rational function coefficients::  * Groebner basis computation with rational function coefficients::
 * Change of ordering::  * Change of ordering::
 * Weyl algebra::  * Weyl algebra::
Line 1055  beforehand, and some heuristic trial may be inevitable
Line 1057  beforehand, and some heuristic trial may be inevitable
 \E  \E
   
 \BJP  \BJP
   @node Weight ,,, $B%0%l%V%J4pDl$N7W;;(B
   @section Weight
   \E
   \BEG
   @node Weight,,, Groebner basis computation
   @section Weight
   \E
   \BJP
   $BA0@a$G>R2p$7$?9`=g=x$O(B, $B3FJQ?t$K(B weight ($B=E$_(B) $B$r@_Dj$9$k$3$H$G(B
   $B$h$j0lHLE*$J$b$N$H$J$k(B.
   \E
   \BEG
   Term orderings introduced in the previous section can be generalized
   by setting a weight for each variable.
   \E
   @example
   [0] dp_td(<<1,1,1>>);
   3
   [1] dp_set_weight([1,2,3])$
   [2] dp_td(<<1,1,1>>);
   6
   @end example
   \BJP
   $BC19`<0$NA4<!?t$r7W;;$9$k:](B, $B%G%U%)%k%H$G$O(B
   $B3FJQ?t$N;X?t$NOB$rA4<!?t$H$9$k(B. $B$3$l$O3FJQ?t$N(B weight $B$r(B 1 $B$H(B
   $B9M$($F$$$k$3$H$KAjEv$9$k(B. $B$3$NNc$G$O(B, $BBh0l(B, $BBhFs(B, $BBh;0JQ?t$N(B
   weight $B$r$=$l$>$l(B 1,2,3 $B$H;XDj$7$F$$$k(B. $B$3$N$?$a(B, @code{<<1,1,1>>}
   $B$NA4<!?t(B ($B0J2<$G$O$3$l$rC19`<0$N(B weight $B$H8F$V(B) $B$,(B @code{1*1+1*2+1*3=6} $B$H$J$k(B.
   weight $B$r@_Dj$9$k$3$H$G(B, $BF1$89`=g=x7?$N$b$H$G0[$J$k9`=g=x$,Dj5A$G$-$k(B.
   $BNc$($P(B, weight $B$r$&$^$/@_Dj$9$k$3$H$G(B, $BB?9`<0$r(B weighted homogeneous
   $B$K$9$k$3$H$,$G$-$k>l9g$,$"$k(B.
   \E
   \BEG
   By default, the total degree of a monomial is equal to
   the sum of all exponents. This means that the weight for each variable
   is set to 1.
   In this example, the weights for the first, the second and the third
   variable are set to 1, 2 and 3 respectively.
   Therefore the total degree of @code{<<1,1,1>>} under this weight,
   which is called the weight of the monomial, is @code{1*1+1*2+1*3=6}.
   By setting weights, different term orderings can be set under a type of
   term ordeing. In some case a polynomial can
   be made weighted homogeneous by setting an appropriate weight.
   \E
   
   \BJP
   $B3FJQ?t$KBP$9$k(B weight $B$r$^$H$a$?$b$N$r(B weight vector $B$H8F$V(B.
   $B$9$Y$F$N@.J,$,@5$G$"$j(B, $B%0%l%V%J4pDl7W;;$K$*$$$F(B, $BA4<!?t$N(B
   $BBe$o$j$KMQ$$$i$l$k$b$N$rFC$K(B sugar weight $B$H8F$V$3$H$K$9$k(B.
   sugar strategy $B$K$*$$$F(B, $BA4<!?t$NBe$o$j$K;H$o$l$k$+$i$G$"$k(B.
   $B0lJ}$G(B, $B3F@.J,$,I,$:$7$b@5$H$O8B$i$J$$(B weight vector $B$O(B,
   sugar weight $B$H$7$F@_Dj$9$k$3$H$O$G$-$J$$$,(B, $B9`=g=x$N0lHL2=$K$O(B
   $BM-MQ$G$"$k(B. $B$3$l$i$O(B, $B9TNs$K$h$k9`=g=x$N@_Dj$K$9$G$K8=$l$F(B
   $B$$$k(B. $B$9$J$o$A(B, $B9`=g=x$rDj5A$9$k9TNs$N3F9T$,(B, $B0l$D$N(B weight vector
   $B$H8+$J$5$l$k(B. $B$^$?(B, $B%V%m%C%/=g=x$O(B, $B3F%V%m%C%/$N(B
   $BJQ?t$KBP1~$9$k@.J,$N$_(B 1 $B$GB>$O(B 0 $B$N(B weight vector $B$K$h$kHf3S$r(B
   $B:G=i$K9T$C$F$+$i(B, $B3F%V%m%C%/Kh$N(B tie breaking $B$r9T$&$3$H$KAjEv$9$k(B.
   \E
   
   \BEG
   A list of weights for all variables is called a weight vector.
   A weight vector is called a sugar weight vector if
   its elements are all positive and it is used for computing
   a weighted total degree of a monomial, because such a weight
   is used instead of total degree in sugar strategy.
   On the other hand, a weight vector whose elements are not necessarily
   positive cannot be set as a sugar weight, but it is useful for
   generalizing term order. In fact, such a weight vector already
   appeared in a matrix order. That is, each row of a matrix defining
   a term order is regarded as a weight vector. A block order
   is also considered as a refinement of comparison by weight vectors.
   It compares two terms by using a weight vector whose elements
   corresponding to variables in a block is 1 and 0 otherwise,
   then it applies a tie breaker.
   \E
   
   \BJP
   weight vector $B$N@_Dj$O(B @code{dp_set_weight()} $B$G9T$&$3$H$,$G$-$k(B
   $B$,(B, $B9`=g=x$r;XDj$9$k:]$NB>$N%Q%i%a%?(B ($B9`=g=x7?(B, $BJQ?t=g=x(B) $B$H(B
   $B$^$H$a$F@_Dj$G$-$k$3$H$,K>$^$7$$(B. $B$3$N$?$a(B, $B<!$N$h$&$J7A$G$b(B
   $B9`=g=x$,;XDj$G$-$k(B.
   \E
   \BEG
   A weight vector can be set by using @code{dp_set_weight()}.
   However it is more preferable if a weight vector can be set
   together with other parapmeters such as a type of term ordering
   and a variable order. This is realized as follows.
   \E
   
   @example
   [64] B=[x+y+z-6,x*y+y*z+z*x-11,x*y*z-6]$
   [65] dp_gr_main(B|v=[x,y,z],sugarweight=[3,2,1],order=0);
   [z^3-6*z^2+11*z-6,x+y+z-6,-y^2+(-z+6)*y-z^2+6*z-11]
   [66] dp_gr_main(B|v=[y,z,x],order=[[1,1,0],[0,1,0],[0,0,1]]);
   [x^3-6*x^2+11*x-6,x+y+z-6,-x^2+(-y+6)*x-y^2+6*y-11]
   [67] dp_gr_main(B|v=[y,z,x],order=[[x,1,y,2,z,3]]);
   [x+y+z-6,x^3-6*x^2+11*x-6,-x^2+(-y+6)*x-y^2+6*y-11]
   @end example
   
   \BJP
   $B$$$:$l$NNc$K$*$$$F$b(B, $B9`=g=x$O(B option $B$H$7$F;XDj$5$l$F$$$k(B.
   $B:G=i$NNc$G$O(B @code{v} $B$K$h$jJQ?t=g=x$r(B, @code{sugarweight} $B$K$h$j(B
   sugar weight vector $B$r(B, @code{order}$B$K$h$j9`=g=x7?$r;XDj$7$F$$$k(B.
   $BFs$DL\$NNc$K$*$1$k(B @code{order} $B$N;XDj$O(B matrix order $B$HF1MM$G$"$k(B.
   $B$9$J$o$A(B, $B;XDj$5$l$?(B weight vector $B$r:8$+$i=g$K;H$C$F(B weight $B$NHf3S(B
   $B$r9T$&(B. $B;0$DL\$NNc$bF1MM$G$"$k$,(B, $B$3$3$G$O(B weight vector $B$NMWAG$r(B
   $BJQ?tKh$K;XDj$7$F$$$k(B. $B;XDj$,$J$$$b$N$O(B 0 $B$H$J$k(B. $B;0$DL\$NNc$G$O(B,
   @code{order} $B$K$h$k;XDj$G$O9`=g=x$,7hDj$7$J$$(B. $B$3$N>l9g$K$O(B,
   tie breaker $B$H$7$FA4<!?t5U<-=q<0=g=x$,<+F0E*$K@_Dj$5$l$k(B.
   $B$3$N;XDjJ}K!$O(B, @code{dp_gr_main}, @code{dp_gr_mod_main} $B$J$I(B
   $B$NAH$_9~$_4X?t$G$N$_2DG=$G$"$j(B, @code{gr} $B$J$I$N%f!<%6Dj5A4X?t(B
   $B$G$OL$BP1~$G$"$k(B.
   \E
   \BEG
   In each example, a term ordering is specified as options.
   In the first example, a variable order, a sugar weight vector
   and a type of term ordering are specified by options @code{v},
   @code{sugarweight} and @code{order} respectively.
   In the second example, an option @code{order} is used
   to set a matrix ordering. That is, the specified weight vectors
   are used from left to right for comparing terms.
   The third example shows a variant of specifying a weight vector,
   where each component of a weight vector is specified variable by variable,
   and unspecified components are set to zero. In this example,
   a term order is not determined only by the specified weight vector.
   In such a case a tie breaker by the graded reverse lexicographic ordering
   is set automatically.
   This type of a term ordering specification can be applied only to builtin
   functions such as @code{dp_gr_main()}, @code{dp_gr_mod_main()}, not to
   user defined functions such as @code{gr()}.
   \E
   
   \BJP
 @node $BM-M}<0$r78?t$H$9$k%0%l%V%J4pDl7W;;(B,,, $B%0%l%V%J4pDl$N7W;;(B  @node $BM-M}<0$r78?t$H$9$k%0%l%V%J4pDl7W;;(B,,, $B%0%l%V%J4pDl$N7W;;(B
 @section $BM-M}<0$r78?t$H$9$k%0%l%V%J4pDl7W;;(B  @section $BM-M}<0$r78?t$H$9$k%0%l%V%J4pDl7W;;(B
 \E  \E
Line 1329  Computation of the global b function is implemented as
Line 1464  Computation of the global b function is implemented as
 * tolexm minipolym::  * tolexm minipolym::
 * dp_gr_main dp_gr_mod_main dp_gr_f_main dp_weyl_gr_main dp_weyl_gr_mod_main dp_weyl_gr_f_main::  * dp_gr_main dp_gr_mod_main dp_gr_f_main dp_weyl_gr_main dp_weyl_gr_mod_main dp_weyl_gr_f_main::
 * dp_f4_main dp_f4_mod_main dp_weyl_f4_main dp_weyl_f4_mod_main::  * dp_f4_main dp_f4_mod_main dp_weyl_f4_main dp_weyl_f4_mod_main::
   * nd_gr nd_gr_trace nd_f4 nd_weyl_gr nd_weyl_gr_trace::
 * dp_gr_flags dp_gr_print::  * dp_gr_flags dp_gr_print::
 * dp_ord::  * dp_ord::
 * dp_ptod::  * dp_ptod::
Line 1412  strategy $B$K$h$k7W;;(B, @code{hgr()} $B$O(B trace
Line 1548  strategy $B$K$h$k7W;;(B, @code{hgr()} $B$O(B trace
 @item  @item
 @code{dgr()} $B$GI=<($5$l$k;~4V$O(B, $B$3$NH!?t$,<B9T$5$l$F$$$k%W%m%;%9$G$N(B  @code{dgr()} $B$GI=<($5$l$k;~4V$O(B, $B$3$NH!?t$,<B9T$5$l$F$$$k%W%m%;%9$G$N(B
 CPU $B;~4V$G$"$j(B, $B$3$NH!?t$N>l9g$O$[$H$s$IDL?.$N$?$a$N;~4V$G$"$k(B.  CPU $B;~4V$G$"$j(B, $B$3$NH!?t$N>l9g$O$[$H$s$IDL?.$N$?$a$N;~4V$G$"$k(B.
   @item
   $BB?9`<0%j%9%H(B @var{plist} $B$NMWAG$,J,;6I=8=B?9`<0$N>l9g$O(B
   $B7k2L$bJ,;6I=8=B?9`<0$N%j%9%H$G$"$k(B.
   $B$3$N>l9g(B, $B0z?t$NJ,;6B?9`<0$OM?$($i$l$?=g=x$K=>$$(B @code{dp_sort} $B$G(B
   $B%=!<%H$5$l$F$+$i7W;;$5$l$k(B.
   $BB?9`<0%j%9%H$NMWAG$,J,;6I=8=B?9`<0$N>l9g$b(B
   $BJQ?t$N?tJ,$NITDj85$N%j%9%H$r(B @var{vlist} $B0z?t$H$7$FM?$($J$$$H$$$1$J$$(B
   ($B%@%_!<(B).
 \E  \E
 \BEG  \BEG
 @item  @item
Line 1440  Therefore this function is useful to reduce the actual
Line 1584  Therefore this function is useful to reduce the actual
 The CPU time shown after an exection of @code{dgr()} indicates  The CPU time shown after an exection of @code{dgr()} indicates
 that of the master process, and most of the time corresponds to the time  that of the master process, and most of the time corresponds to the time
 for communication.  for communication.
   @item
   When the elements of @var{plist} are distributed polynomials,
   the result is also a list of distributed polynomials.
   In this case, firstly  the elements of @var{plist} is sorted by @code{dp_sort}
   and the Grobner basis computation is started.
   Variables must be given in @var{vlist} even in this case
   (these variables are dummy).
 \E  \E
 @end itemize  @end itemize
   
Line 2149  except for lack of the argument for controlling homoge
Line 2300  except for lack of the argument for controlling homoge
 @fref{dp_ord},  @fref{dp_ord},
 @fref{dp_gr_flags dp_gr_print},  @fref{dp_gr_flags dp_gr_print},
 @fref{gr hgr gr_mod},  @fref{gr hgr gr_mod},
   \JP @fref{$B7W;;$*$h$SI=<($N@)8f(B}.
   \EG @fref{Controlling Groebner basis computations}
   @end table
   
   \JP @node nd_gr nd_gr_trace nd_f4 nd_weyl_gr nd_weyl_gr_trace,,, $B%0%l%V%J4pDl$K4X$9$kH!?t(B
   \EG @node nd_gr nd_gr_trace nd_f4 nd_weyl_gr nd_weyl_gr_trace,,, Functions for Groebner basis computation
   @subsection @code{nd_gr}, @code{nd_gr_trace}, @code{nd_f4}, @code{nd_weyl_gr}, @code{nd_weyl_gr_trace}
   @findex nd_gr
   @findex nd_gr_trace
   @findex nd_f4
   @findex nd_weyl_gr
   @findex nd_weyl_gr_trace
   
   @table @t
   @item nd_gr(@var{plist},@var{vlist},@var{p},@var{order})
   @itemx nd_gr_trace(@var{plist},@var{vlist},@var{homo},@var{p},@var{order})
   @itemx nd_f4(@var{plist},@var{vlist},@var{modular},@var{order})
   @item nd_weyl_gr(@var{plist},@var{vlist},@var{p},@var{order})
   @itemx nd_weyl_gr_trace(@var{plist},@var{vlist},@var{homo},@var{p},@var{order})
   \JP :: $B%0%l%V%J4pDl$N7W;;(B ($BAH$_9~$_H!?t(B)
   \EG :: Groebner basis computation (built-in functions)
   @end table
   
   @table @var
   @item return
   \JP $B%j%9%H(B
   \EG list
   @item plist  vlist
   \JP $B%j%9%H(B
   \EG list
   @item order
   \JP $B?t(B, $B%j%9%H$^$?$O9TNs(B
   \EG number, list or matrix
   @item homo
   \JP $B%U%i%0(B
   \EG flag
   @item modular
   \JP $B%U%i%0$^$?$OAG?t(B
   \EG flag or prime
   @end table
   
   \BJP
   @itemize @bullet
   @item
   $B$3$l$i$NH!?t$O(B, $B%0%l%V%J4pDl7W;;AH$_9~$_4X?t$N?7<BAu$G$"$k(B.
   @item @code{nd_gr} $B$O(B, @code{p} $B$,(B 0 $B$N$H$-M-M}?tBN>e$N(B Buchberger
   $B%"%k%4%j%:%`$r<B9T$9$k(B. @code{p} $B$,(B 2 $B0J>e$N<+A3?t$N$H$-(B, GF(p) $B>e$N(B
   Buchberger $B%"%k%4%j%:%`$r<B9T$9$k(B.
   @item @code{nd_gr_trace} $B$OM-M}?tBN>e$G(B trace $B%"%k%4%j%:%`$r<B9T$9$k(B.
   @code{p} $B$,(B 0 $B$^$?$O(B 1 $B$N$H$-(B, $B<+F0E*$KA*$P$l$?AG?t$rMQ$$$F(B, $B@.8y$9$k(B
   $B$^$G(B trace $B%"%k%4%j%:%`$r<B9T$9$k(B.
   @code{p} $B$,(B 2 $B0J>e$N$H$-(B, trace $B$O(BGF(p) $B>e$G7W;;$5$l$k(B. trace $B%"%k%4%j%:%`(B
   $B$,<:GT$7$?>l9g(B 0 $B$,JV$5$l$k(B. @code{p} $B$,Ii$N>l9g(B, $B%0%l%V%J4pDl%A%'%C%/$O(B
   $B9T$o$J$$(B. $B$3$N>l9g(B, @code{p} $B$,(B -1 $B$J$i$P<+F0E*$KA*$P$l$?AG?t$,(B,
   $B$=$l0J30$O;XDj$5$l$?AG?t$rMQ$$$F%0%l%V%J4pDl8uJd$N7W;;$,9T$o$l$k(B.
   @item
   @code{nd_f4} $B$O(B, $BM-8BBN>e$N(B F4 $B%"%k%4%j%:%`$r<B9T$9$k(B.
   @item
   @code{nd_weyl_gr}, @code{nd_weyl_gr_trace} $B$O(B Weyl $BBe?tMQ$G$"$k(B.
   @item
   $B$$$:$l$N4X?t$b(B, $BM-M}4X?tBN>e$N7W;;$OL$BP1~$G$"$k(B.
   @item
   $B0lHL$K(B @code{dp_gr_main}, @code{dp_gr_mod_main} $B$h$j9bB.$G$"$k$,(B,
   $BFC$KM-8BBN>e$N>l9g82Cx$G$"$k(B.
   @end itemize
   \E
   
   \BEG
   @itemize @bullet
   @item
   These functions are new implementations for computing Groebner bases.
   @item @code{nd_gr} executes Buchberger algorithm over the rationals
   if  @code{p} is 0, and that over GF(p) if @code{p} is a prime.
   @item @code{nd_gr_trace} executes the trace algorithm over the rationals.
   If @code{p} is 0 or 1, the trace algorithm is executed until it succeeds
   by using automatically chosen primes.
   If @code{p} a positive prime,
   the trace is comuted over GF(p).
   If the trace algorithm fails 0 is returned.
   If @code{p} is negative,
   the Groebner basis check and ideal-membership check are omitted.
   In this case, an automatically chosen prime if @code{p} is 1,
   otherwise the specified prime is used to compute a Groebner basis
   candidate.
   @item
   @code{nd_f4} executes F4 algorithm over a finite field.
   @item
   @code{nd_weyl_gr}, @code{nd_weyl_gr_trace} are for Weyl algebra computation.
   @item
   Each function cannot handle rational function coefficient cases.
   @item
   In general these functions are more efficient than
   @code{dp_gr_main}, @code{dp_gr_mod_main}, especially over finite fields.
   @end itemize
   \E
   
   @example
   [38] load("cyclic")$
   [49] C=cyclic(7)$
   [50] V=vars(C)$
   [51] cputime(1)$
   [52] dp_gr_mod_main(C,V,0,31991,0)$
   26.06sec + gc : 0.313sec(26.4sec)
   [53] nd_gr(C,V,31991,0)$
   ndv_alloc=1477188
   5.737sec + gc : 0.1837sec(5.921sec)
   [54] dp_f4_mod_main(C,V,31991,0)$
   3.51sec + gc : 0.7109sec(4.221sec)
   [55] nd_f4(C,V,31991,0)$
   1.906sec + gc : 0.126sec(2.032sec)
   @end example
   
   @table @t
   \JP @item $B;2>H(B
   \EG @item References
   @fref{dp_ord},
   @fref{dp_gr_flags dp_gr_print},
 \JP @fref{$B7W;;$*$h$SI=<($N@)8f(B}.  \JP @fref{$B7W;;$*$h$SI=<($N@)8f(B}.
 \EG @fref{Controlling Groebner basis computations}  \EG @fref{Controlling Groebner basis computations}
 @end table  @end table
Legend:
Removed from v.1.11  
changed lines
  Added in v.1.15
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>