===================================================================
RCS file: /home/cvs/OpenXM_contrib2/asir2000/engine/distm.c,v
retrieving revision 1.1
retrieving revision 1.8
diff -u -p -r1.1 -r1.8
--- OpenXM_contrib2/asir2000/engine/distm.c	1999/12/03 07:39:08	1.1
+++ OpenXM_contrib2/asir2000/engine/distm.c	2000/12/05 08:29:44	1.8
@@ -1,19 +1,70 @@
-/* $OpenXM: OpenXM/src/asir99/engine/distm.c,v 1.1.1.1 1999/11/10 08:12:26 noro Exp $ */
+/*
+ * Copyright (c) 1994-2000 FUJITSU LABORATORIES LIMITED 
+ * All rights reserved.
+ * 
+ * FUJITSU LABORATORIES LIMITED ("FLL") hereby grants you a limited,
+ * non-exclusive and royalty-free license to use, copy, modify and
+ * redistribute, solely for non-commercial and non-profit purposes, the
+ * computer program, "Risa/Asir" ("SOFTWARE"), subject to the terms and
+ * conditions of this Agreement. For the avoidance of doubt, you acquire
+ * only a limited right to use the SOFTWARE hereunder, and FLL or any
+ * third party developer retains all rights, including but not limited to
+ * copyrights, in and to the SOFTWARE.
+ * 
+ * (1) FLL does not grant you a license in any way for commercial
+ * purposes. You may use the SOFTWARE only for non-commercial and
+ * non-profit purposes only, such as academic, research and internal
+ * business use.
+ * (2) The SOFTWARE is protected by the Copyright Law of Japan and
+ * international copyright treaties. If you make copies of the SOFTWARE,
+ * with or without modification, as permitted hereunder, you shall affix
+ * to all such copies of the SOFTWARE the above copyright notice.
+ * (3) An explicit reference to this SOFTWARE and its copyright owner
+ * shall be made on your publication or presentation in any form of the
+ * results obtained by use of the SOFTWARE.
+ * (4) In the event that you modify the SOFTWARE, you shall notify FLL by
+ * e-mail at risa-admin@sec.flab.fujitsu.co.jp of the detailed specification
+ * for such modification or the source code of the modified part of the
+ * SOFTWARE.
+ * 
+ * THE SOFTWARE IS PROVIDED AS IS WITHOUT ANY WARRANTY OF ANY KIND. FLL
+ * MAKES ABSOLUTELY NO WARRANTIES, EXPRESSED, IMPLIED OR STATUTORY, AND
+ * EXPRESSLY DISCLAIMS ANY IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS
+ * FOR A PARTICULAR PURPOSE OR NONINFRINGEMENT OF THIRD PARTIES'
+ * RIGHTS. NO FLL DEALER, AGENT, EMPLOYEES IS AUTHORIZED TO MAKE ANY
+ * MODIFICATIONS, EXTENSIONS, OR ADDITIONS TO THIS WARRANTY.
+ * UNDER NO CIRCUMSTANCES AND UNDER NO LEGAL THEORY, TORT, CONTRACT,
+ * OR OTHERWISE, SHALL FLL BE LIABLE TO YOU OR ANY OTHER PERSON FOR ANY
+ * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, PUNITIVE OR CONSEQUENTIAL
+ * DAMAGES OF ANY CHARACTER, INCLUDING, WITHOUT LIMITATION, DAMAGES
+ * ARISING OUT OF OR RELATING TO THE SOFTWARE OR THIS AGREEMENT, DAMAGES
+ * FOR LOSS OF GOODWILL, WORK STOPPAGE, OR LOSS OF DATA, OR FOR ANY
+ * DAMAGES, EVEN IF FLL SHALL HAVE BEEN INFORMED OF THE POSSIBILITY OF
+ * SUCH DAMAGES, OR FOR ANY CLAIM BY ANY OTHER PARTY. EVEN IF A PART
+ * OF THE SOFTWARE HAS BEEN DEVELOPED BY A THIRD PARTY, THE THIRD PARTY
+ * DEVELOPER SHALL HAVE NO LIABILITY IN CONNECTION WITH THE USE,
+ * PERFORMANCE OR NON-PERFORMANCE OF THE SOFTWARE.
+ *
+ * $OpenXM: OpenXM_contrib2/asir2000/engine/distm.c,v 1.7 2000/12/05 06:59:16 noro Exp $ 
+*/
 #include "ca.h"
 #include "inline.h"
 
-#define NV(p) ((p)->nv)
-#define C(p) ((p)->c)
-#if 0
-#define ITOS(p) (((unsigned int)(p))>>1)
-#define STOI(i) ((P)((((unsigned int)(i))<<1)|1))
-#else
-#define ITOS(p) (((unsigned int)(p))&0x7fffffff)
-#define STOI(i) ((P)((unsigned int)(i)|0x80000000))
-#endif
-
 extern int (*cmpdl)();
+extern int do_weyl;
 
+void comm_mulmd();
+void weyl_mulmd();
+void weyl_mulmdm();
+void weyl_mulmmm();
+
+void mulmdm_dup();
+void comm_mulmd_dup();
+void weyl_mulmd_dup();
+void weyl_mulmdm_dup();
+void weyl_mulmmm_dup();
+void comm_mulmd_tab_destructive();
+void adddl_dup();
 void ptomd(vl,mod,dvl,p,pr)
 VL vl,dvl;
 int mod;
@@ -62,7 +113,7 @@ DP *pr;
 					mptomd(vl,mod,dvl,COEF(dc),&t);
 					NEWDL(d,n); d->td = QTOS(DEG(dc)); d->d[i] = d->td;
 					NEWMP(m); m->dl = d; C(m) = (P)ONEM; NEXT(m) = 0; MKDP(n,m,u);
-					mulmd(vl,mod,t,u,&r); addmd(vl,mod,r,s,&t); s = t;
+					comm_mulmd(vl,mod,t,u,&r); addmd(vl,mod,r,s,&t); s = t;
 				}
 				*pr = s;
 			}
@@ -196,8 +247,22 @@ VL vl;
 int mod;
 DP p1,p2,*pr;
 {
+	if ( !do_weyl )
+		comm_mulmd(vl,mod,p1,p2,pr);
+	else
+		weyl_mulmd(vl,mod,p1,p2,pr);
+}
+
+void comm_mulmd(vl,mod,p1,p2,pr)
+VL vl;
+int mod;
+DP p1,p2,*pr;
+{
 	MP m;
 	DP s,t,u;
+	int i,l,l1;
+	static MP *w;
+	static int wlen;
 
 	if ( !p1 || !p2 )
 		*pr = 0;
@@ -206,13 +271,61 @@ DP p1,p2,*pr;
 	else if ( OID(p2) <= O_P )
 		mulmdc(vl,mod,p1,(P)p2,pr);
 	else {
-		for ( m = BDY(p2), s = 0; m; m = NEXT(m) ) {
-			mulmdm(vl,mod,p1,m,&t); addmd(vl,mod,s,t,&u); s = u;
+		for ( m = BDY(p1), l1 = 0; m; m = NEXT(m), l1++ );
+		for ( m = BDY(p2), l = 0; m; m = NEXT(m), l++ );
+		if ( l1 < l ) {
+			t = p1; p1 = p2; p2 = t;
+			l = l1;
 		}
+		if ( l > wlen ) {
+			if ( w ) GC_free(w);
+			w = (MP *)MALLOC(l*sizeof(MP));
+			wlen = l;
+		}
+		for ( m = BDY(p2), i = 0; i < l; m = NEXT(m), i++ )
+			w[i] = m;
+		for ( s = 0, i = l-1; i >= 0; i-- ) {
+			mulmdm(vl,mod,p1,w[i],&t); addmd(vl,mod,s,t,&u); s = u;
+		}
+		bzero(w,l*sizeof(MP));
 		*pr = s;
 	}
 }
 
+void weyl_mulmd(vl,mod,p1,p2,pr)
+VL vl;
+int mod;
+DP p1,p2,*pr;
+{
+	MP m;
+	DP s,t,u;
+	int i,l,l1;
+	static MP *w;
+	static int wlen;
+
+	if ( !p1 || !p2 )
+		*pr = 0;
+	else if ( OID(p1) <= O_P )
+		mulmdc(vl,mod,p2,(P)p1,pr);
+	else if ( OID(p2) <= O_P )
+		mulmdc(vl,mod,p1,(P)p2,pr);
+	else {
+		for ( m = BDY(p2), l = 0; m; m = NEXT(m), l++ );
+		if ( l > wlen ) {
+			if ( w ) GC_free(w);
+			w = (MP *)MALLOC(l*sizeof(MP));
+			wlen = l;
+		}
+		for ( m = BDY(p2), i = 0; i < l; m = NEXT(m), i++ )
+			w[i] = m;
+		for ( s = 0, i = l-1; i >= 0; i-- ) {
+			weyl_mulmdm(vl,mod,p1,w[i],&t); addmd(vl,mod,s,t,&u); s = u;
+		}
+		bzero(w,l*sizeof(MP));
+		*pr = s;
+	}
+}
+
 void mulmdm(vl,mod,p,m0,pr)
 VL vl;
 int mod;
@@ -245,6 +358,117 @@ DP *pr;
 	}
 }
 
+void weyl_mulmdm(vl,mod,p,m0,pr)
+VL vl;
+int mod;
+DP p;
+MP m0;
+DP *pr;
+{
+	DP r,t,t1;
+	MP m;
+	int n,l,i;
+	static MP *w;
+	static int wlen;
+
+	if ( !p )
+		*pr = 0;
+	else {
+		for ( m = BDY(p), l = 0; m; m = NEXT(m), l++ );
+		if ( l > wlen ) {
+			if ( w ) GC_free(w);
+			w = (MP *)MALLOC(l*sizeof(MP));
+			wlen = l;
+		}
+		for ( m = BDY(p), i = 0; i < l; m = NEXT(m), i++ )
+			w[i] = m;
+		for ( r = 0, i = l-1, n = NV(p); i >= 0; i-- ) {
+			weyl_mulmmm(vl,mod,w[i],m0,n,&t);
+			addmd(vl,mod,r,t,&t1); r = t1;
+		}
+		bzero(w,l*sizeof(MP));
+		if ( r )
+			r->sugar = p->sugar + m0->dl->td;
+		*pr = r;
+	}
+}
+
+/* m0 = x0^d0*x1^d1*... * dx0^d(n/2)*dx1^d(n/2+1)*... */
+
+void weyl_mulmmm(vl,mod,m0,m1,n,pr)
+VL vl;
+int mod;
+MP m0,m1;
+int n;
+DP *pr;
+{
+	MP m,mr,mr0;
+	MQ mq;
+	DP r,t,t1;
+	P c,c0,c1,cc;
+	DL d,d0,d1;
+	int i,j,a,b,k,l,n2,s,min;
+	static int *tab;
+	static int tablen;
+
+	if ( !m0 || !m1 )
+		*pr = 0;
+	else {
+		c0 = C(m0); c1 = C(m1);
+		mulmp(vl,mod,c0,c1,&c);
+		d0 = m0->dl; d1 = m1->dl;
+		n2 = n>>1;
+		if ( n & 1 ) { 
+			/* homogenized computation; dx-xd=h^2 */
+			/* offset of h-degree */
+			NEWDL(d,n);
+			d->td = d->d[n-1] = d0->d[n-1]+d1->d[n-1];
+			NEWMP(mr); mr->c = (P)ONEM; mr->dl = d; NEXT(mr) = 0;
+			MKDP(n,mr,r); r->sugar = d->td;
+		} else
+			r = (DP)ONEM;
+		for ( i = 0; i < n2; i++ ) {
+			a = d0->d[i]; b = d1->d[n2+i];
+			k = d0->d[n2+i]; l = d1->d[i];
+			/* degree of xi^a*(Di^k*xi^l)*Di^b */
+			s = a+k+l+b;
+			/* compute xi^a*(Di^k*xi^l)*Di^b */
+			min = MIN(k,l);
+
+			if ( min+1 > tablen ) {
+				if ( tab ) GC_free(tab);
+				tab = (int *)MALLOC((min+1)*sizeof(int));
+				tablen = min+1;
+			}
+			mkwcm(k,l,mod,tab);
+			if ( n & 1 )
+				for ( mr0 = 0, j = 0; j <= min; j++ ) {
+					NEXTMP(mr0,mr); NEWDL(d,n);
+					d->d[i] = l-j+a; d->d[n2+i] = k-j+b;
+					d->td = s;
+					d->d[n-1] = s-(d->d[i]+d->d[n2+i]); 
+					STOMQ(tab[j],mq); mr->c = (P)mq; mr->dl = d;
+				}
+			else
+				for ( mr0 = 0, s = 0, j = 0; j <= min; j++ ) {
+					NEXTMP(mr0,mr); NEWDL(d,n);
+					d->d[i] = l-j+a; d->d[n2+i] = k-j+b;
+					d->td = d->d[i]+d->d[n2+i]; /* XXX */
+					s = MAX(s,d->td); /* XXX */
+					STOMQ(tab[j],mq); mr->c = (P)mq; mr->dl = d;
+				}
+			bzero(tab,(min+1)*sizeof(int));
+			if ( mr0 )
+				NEXT(mr) = 0;
+			MKDP(n,mr0,t);
+			if ( t )
+				t->sugar = s;
+			comm_mulmd(vl,mod,r,t,&t1); r = t1;
+		}
+		mulmdc(vl,mod,r,c,pr);
+	}
+}
+
 void mulmdc(vl,mod,p,c,pr)
 VL vl;
 int mod;
@@ -297,11 +521,8 @@ DP *pr;
 	}
 }
 
-#define MKDPM(n,m,d) (NEWDP(d),(d)->nv=(n),BDY(d)=(m))
-
-void _mdtop(vl,mod,dvl,p,pr)
+void _dtop_mod(vl,dvl,p,pr)
 VL vl,dvl;
-int mod;
 DP p;
 P *pr;
 {
@@ -310,32 +531,104 @@ P *pr;
 	MP m;
 	P r,s,t,u,w;
 	Q q;
-	MQ tq;
 	VL tvl;
 
 	if ( !p )
 		*pr = 0;
 	else {
 		for ( n = p->nv, m = BDY(p), s = 0; m; m = NEXT(m) ) {
-			STOMQ(ITOS(C(m)),tq); t = (P)tq;
+			i = ITOS(m->c); STOQ(i,q); t = (P)q;
 			for ( i = 0, d = m->dl, tvl = dvl;
 				i < n; tvl = NEXT(tvl), i++ ) {
-				MKV(tvl->v,r); STOQ(d->d[i],q); pwrmp(vl,mod,r,q,&u);
-				mulmp(vl,mod,t,u,&w); t = w;
+				MKV(tvl->v,r); STOQ(d->d[i],q); pwrp(vl,r,q,&u);
+				mulp(vl,t,u,&w); t = w;
 			}
-			addmp(vl,mod,s,t,&u); s = u;
+			addp(vl,s,t,&u); s = u;
 		}
-		mptop(s,pr);
+		*pr = s;
 	}
 }
 
-void _addmd(vl,mod,p1,p2,pr)
-VL vl;
+void _dp_mod(p,mod,subst,rp)
+DP p;
 int mod;
+NODE subst;
+DP *rp;
+{
+	MP m,mr,mr0;
+	P t,s;
+	NODE tn;
+
+	if ( !p )
+		*rp = 0;
+	else {
+		for ( mr0 = 0, m = BDY(p); m; m = NEXT(m) ) {
+			for ( tn = subst, s = m->c; tn; tn = NEXT(NEXT(tn)) ) {
+				substp(CO,s,BDY(tn),BDY(NEXT(tn)),&t);
+				s = t;
+			}
+			ptomp(mod,s,&t);
+			if ( t ) {
+				NEXTMP(mr0,mr); mr->c = STOI(CONT((MQ)t)); mr->dl = m->dl;
+			}
+		}
+		if ( mr0 ) {
+			NEXT(mr) = 0; MKDP(p->nv,mr0,*rp); (*rp)->sugar = p->sugar;
+		} else
+			*rp = 0;
+	}
+}
+
+void _dp_monic(p,mod,rp)
+DP p;
+int mod;
+DP *rp;
+{
+	MP m,mr,mr0;
+	int c,c1;
+	NODE tn;
+
+	if ( !p )
+		*rp = 0;
+	else {
+		c = invm(ITOS(BDY(p)->c),mod);
+		for ( mr0 = 0, m = BDY(p); m; m = NEXT(m) ) {
+			c1 = dmar(ITOS(m->c),c,0,mod);
+			NEXTMP(mr0,mr); mr->c = STOI(c1); mr->dl = m->dl;
+		}
+		NEXT(mr) = 0; MKDP(p->nv,mr0,*rp); (*rp)->sugar = p->sugar;
+	}
+}
+
+void _printdp(d)
+DP d;
+{
+	int n,i;
+	MP m;
+	DL dl;
+
+	if ( !d ) {
+		printf("0"); return;
+	}
+	for ( n = d->nv, m = BDY(d); m; m = NEXT(m) ) {
+		printf("%d*<<",ITOS(m->c));
+		for ( i = 0, dl = m->dl; i < n-1; i++ )
+			printf("%d,",dl->d[i]);
+		printf("%d",dl->d[i]);
+		printf(">>");
+		if ( NEXT(m) )
+			printf("+");
+	}
+}
+
+/* merge p1 and p2 into pr */
+
+void addmd_destructive(mod,p1,p2,pr)
+int mod;
 DP p1,p2,*pr;
 {
 	int n;
-	MP m1,m2,mr,mr0;
+	MP m1,m2,mr,mr0,s;
 	int t;
 
 	if ( !p1 )
@@ -349,16 +642,18 @@ DP p1,p2,*pr;
 					t = (ITOS(C(m1))+ITOS(C(m2))) - mod;
 					if ( t < 0 )
 						t += mod;
+					s = m1; m1 = NEXT(m1);
 					if ( t ) {
-						NEXTMP(mr0,mr); mr->dl = m1->dl; C(mr) = STOI(t);
+						NEXTMP2(mr0,mr,s); C(mr) = STOI(t);
 					}
-					m1 = NEXT(m1); m2 = NEXT(m2); break;
+					s = m2; m2 = NEXT(m2);
+					break;
 				case 1:
-					NEXTMP(mr0,mr); mr->dl = m1->dl; C(mr) = C(m1);
-					m1 = NEXT(m1); break;
+					s = m1; m1 = NEXT(m1); NEXTMP2(mr0,mr,s);
+					break;
 				case -1:
-					NEXTMP(mr0,mr); mr->dl = m2->dl; C(mr) = C(m2);
-					m2 = NEXT(m2); break;
+					s = m2; m2 = NEXT(m2); NEXTMP2(mr0,mr,s);
+					break;
 			}
 		if ( !mr0 )
 			if ( m1 )
@@ -375,72 +670,95 @@ DP p1,p2,*pr;
 			NEXT(mr) = m2;
 		else
 			NEXT(mr) = 0;
-		MKDPM(NV(p1),mr0,*pr);
+		MKDP(NV(p1),mr0,*pr);
 		if ( *pr )
 			(*pr)->sugar = MAX(p1->sugar,p2->sugar);
 	}
 }
 
-void _submd(vl,mod,p1,p2,pr)
-VL vl;
+void mulmd_dup(mod,p1,p2,pr)
 int mod;
 DP p1,p2,*pr;
 {
-	DP t;
-
-	if ( !p2 )
-		*pr = p1;
-	else {
-		_chsgnmd(mod,p2,&t); _addmd(vl,mod,p1,t,pr);
-	}
+	if ( !do_weyl )
+		comm_mulmd_dup(mod,p1,p2,pr);
+	else
+		weyl_mulmd_dup(mod,p1,p2,pr);
 }
 
-void _chsgnmd(mod,p,pr)
+void comm_mulmd_dup(mod,p1,p2,pr)
 int mod;
-DP p,*pr;
+DP p1,p2,*pr;
 {
-	MP m,mr,mr0;
+	MP m;
+	DP s,t,u;
+	int i,l,l1;
+	static MP *w;
+	static int wlen;
 
-	if ( !p )
+	if ( !p1 || !p2 )
 		*pr = 0;
 	else {
-		for ( mr0 = 0, m = BDY(p); m; m = NEXT(m) ) {
-			NEXTMP(mr0,mr); C(mr) = STOI(mod - ITOS(C(m))); mr->dl = m->dl;
+		for ( m = BDY(p1), l1 = 0; m; m = NEXT(m), l1++ );
+		for ( m = BDY(p2), l = 0; m; m = NEXT(m), l++ );
+		if ( l1 < l ) {
+			t = p1; p1 = p2; p2 = t;
+			l = l1;
 		}
-		NEXT(mr) = 0; MKDPM(NV(p),mr0,*pr);
-		if ( *pr )
-			(*pr)->sugar = p->sugar;
+		if ( l > wlen ) {
+			if ( w ) GC_free(w);
+			w = (MP *)MALLOC(l*sizeof(MP));
+			wlen = l;
+		}
+		for ( m = BDY(p2), i = 0; i < l; m = NEXT(m), i++ )
+			w[i] = m;
+		for ( s = 0, i = l-1; i >= 0; i-- ) {
+			mulmdm_dup(mod,p1,w[i],&t); addmd_destructive(mod,s,t,&u); s = u;
+		}
+		bzero(w,l*sizeof(MP));
+		*pr = s;
 	}
 }
 
-void _mulmd(vl,mod,p1,p2,pr)
-VL vl;
+void weyl_mulmd_dup(mod,p1,p2,pr)
 int mod;
 DP p1,p2,*pr;
 {
 	MP m;
 	DP s,t,u;
+	int i,l,l1;
+	static MP *w;
+	static int wlen;
 
 	if ( !p1 || !p2 )
 		*pr = 0;
 	else {
-		for ( m = BDY(p2), s = 0; m; m = NEXT(m) ) {
-			_mulmdm(vl,mod,p1,m,&t); _addmd(vl,mod,s,t,&u); s = u;
+		for ( m = BDY(p1), l = 0; m; m = NEXT(m), l++ );
+		if ( l > wlen ) {
+			if ( w ) GC_free(w);
+			w = (MP *)MALLOC(l*sizeof(MP));
+			wlen = l;
 		}
+		for ( m = BDY(p1), i = 0; i < l; m = NEXT(m), i++ )
+			w[i] = m;
+		for ( s = 0, i = l-1; i >= 0; i-- ) {
+			weyl_mulmdm_dup(mod,w[i],p2,&t); addmd_destructive(mod,s,t,&u); s = u;
+		}
+		bzero(w,l*sizeof(MP));
 		*pr = s;
 	}
 }
 
-void _mulmdm(vl,mod,p,m0,pr)
-VL vl;
+void mulmdm_dup(mod,p,m0,pr)
 int mod;
 DP p;
 MP m0;
 DP *pr;
 {
 	MP m,mr,mr0;
-	DL d;
-	int c,n,r;
+	DL d,dt,dm;
+	int c,n,r,i;
+	int *pt,*p1,*p2;
 
 	if ( !p )
 		*pr = 0;
@@ -449,167 +767,225 @@ DP *pr;
 			m; m = NEXT(m) ) {
 			NEXTMP(mr0,mr);
 			C(mr) = STOI(dmar(ITOS(C(m)),c,0,mod));
-			adddl(n,m->dl,d,&mr->dl);
+			NEWDL_NOINIT(dt,n); mr->dl = dt;
+			dm = m->dl;
+			dt->td = d->td + dm->td;
+			for ( i = 0, pt = dt->d, p1=d->d, p2 = dm->d; i < n; i++ )
+				*pt++ = *p1++ + *p2++;
 		}
-		NEXT(mr) = 0; MKDPM(NV(p),mr0,*pr);
+		NEXT(mr) = 0; MKDP(NV(p),mr0,*pr);
 		if ( *pr )
 			(*pr)->sugar = p->sugar + m0->dl->td;
 	}
 }
 
-void _dtop_mod(vl,dvl,p,pr)
-VL vl,dvl;
+void weyl_mulmdm_dup(mod,m0,p,pr)
+int mod;
+MP m0;
 DP p;
-P *pr;
+DP *pr;
 {
-	int n,i;
-	DL d;
+	DP r,t,t1;
 	MP m;
-	P r,s,t,u,w;
-	Q q;
-	VL tvl;
+	DL d0;
+	int n,n2,l,i,j,tlen;
+	static MP *w,*psum;
+	static struct cdlm *tab;
+	static int wlen;
+	static int rtlen;
 
 	if ( !p )
 		*pr = 0;
 	else {
-		for ( n = p->nv, m = BDY(p), s = 0; m; m = NEXT(m) ) {
-			i = ITOS(m->c); STOQ(i,q); t = (P)q;
-			for ( i = 0, d = m->dl, tvl = dvl;
-				i < n; tvl = NEXT(tvl), i++ ) {
-				MKV(tvl->v,r); STOQ(d->d[i],q); pwrp(vl,r,q,&u);
-				mulp(vl,t,u,&w); t = w;
+		for ( m = BDY(p), l = 0; m; m = NEXT(m), l++ );
+		if ( l > wlen ) {
+			if ( w ) GC_free(w);
+			w = (MP *)MALLOC(l*sizeof(MP));
+			wlen = l;
+		}
+		for ( m = BDY(p), i = 0; i < l; m = NEXT(m), i++ )
+			w[i] = m;
+		n = NV(p); n2 = n>>1;
+		d0 = m0->dl;
+
+		for ( i = 0, tlen = 1; i < n2; i++ )
+			tlen *= d0->d[n2+i]+1;
+		if ( tlen > rtlen ) {
+			if ( tab ) GC_free(tab);
+			if ( psum ) GC_free(psum);
+			rtlen = tlen;
+			tab = (struct cdlm *)MALLOC(rtlen*sizeof(struct cdlm));
+			psum = (MP *)MALLOC(rtlen*sizeof(MP));
+		}
+		bzero(psum,tlen*sizeof(MP));
+		for ( i = l-1; i >= 0; i-- ) {
+			bzero(tab,tlen*sizeof(struct cdlm));
+			weyl_mulmmm_dup(mod,m0,w[i],n,tab,tlen);
+			for ( j = 0; j < tlen; j++ ) {
+				if ( tab[j].c ) {
+					NEWMP(m); m->dl = tab[j].d;
+					C(m) = STOI(tab[j].c); NEXT(m) = psum[j];
+					psum[j] = m;
+				}
 			}
-			addp(vl,s,t,&u); s = u;
 		}
-		*pr = s;
+		for ( j = tlen-1, r = 0; j >= 0; j-- ) 
+			if ( psum[j] ) {
+				MKDP(n,psum[j],t); addmd_destructive(mod,r,t,&t1); r = t1;
+			}
+		if ( r )
+			r->sugar = p->sugar + m0->dl->td;
+		*pr = r;
 	}
 }
 
-void _dp_red_mod(p1,p2,mod,rp)
-DP p1,p2;
-int mod;
-DP *rp;
-{
-	int i,n;
-	DL d1,d2,d;
-	MP m;
-	DP t,s;
-	int c,c1;
+/* m0 = x0^d0*x1^d1*... * dx0^d(n/2)*dx1^d(n/2+1)*... */
 
-	n = p1->nv; d1 = BDY(p1)->dl; d2 = BDY(p2)->dl;
-	NEWDL(d,n); d->td = d1->td - d2->td;
-	for ( i = 0; i < n; i++ )
-		d->d[i] = d1->d[i]-d2->d[i];
-	c = invm(ITOS(BDY(p2)->c),mod); c1 = dmar(c,ITOS(BDY(p1)->c),0,mod);
-	NEWMP(m); m->dl = d; m->c = STOI(mod-c1); NEXT(m) = 0;
-	MKDP(n,m,s); s->sugar = d->td;
-	_mulmd(CO,mod,p2,s,&t); _addmd(CO,mod,p1,t,rp);
-}
-
-void _dp_mod(p,mod,subst,rp)
-DP p;
+void weyl_mulmmm_dup(mod,m0,m1,n,rtab,rtablen)
 int mod;
-NODE subst;
-DP *rp;
+MP m0,m1;
+int n;
+struct cdlm *rtab;
+int rtablen;
 {
 	MP m,mr,mr0;
-	P t,s;
-	NODE tn;
+	DP r,t,t1;
+	int c,c0,c1,cc;
+	DL d,d0,d1,dt;
+	int i,j,a,b,k,l,n2,s,min,h,curlen;
+	struct cdlm *p;
+	static int *ctab;
+	static struct cdlm *tab;
+	static int tablen;
+	static struct cdlm *tmptab;
+	static int tmptablen;
 
-	if ( !p )
-		*rp = 0;
-	else {
-		for ( mr0 = 0, m = BDY(p); m; m = NEXT(m) ) {
-			for ( tn = subst, s = m->c; tn; tn = NEXT(NEXT(tn)) ) {
-				substp(CO,s,BDY(tn),BDY(NEXT(tn)),&t);
-				s = t;
-			}
-			ptomp(mod,s,&t);
-			if ( t ) {
-				NEXTMP(mr0,mr); mr->c = STOI(CONT((MQ)t)); mr->dl = m->dl;
-			}
-		}
-		if ( mr0 ) {
-			NEXT(mr) = 0; MKDP(p->nv,mr0,*rp); (*rp)->sugar = p->sugar;
-		} else
-			*rp = 0;
+	if ( !m0 || !m1 ) {
+		rtab[0].c = 0;
+		rtab[0].d = 0;
+		return;
 	}
-}
+	c0 = ITOS(C(m0)); c1 = ITOS(C(m1));
+	c = dmar(c0,c1,0,mod);
+	d0 = m0->dl; d1 = m1->dl;
+	n2 = n>>1;
+	curlen = 1;
 
-void _dp_sp_mod(p1,p2,mod,rp)
-DP p1,p2;
-int mod;
-DP *rp;
-{
-	int i,n,td;
-	int *w;
-	DL d1,d2,d;
-	MP m;
-	DP t,s,u;
+	NEWDL(d,n);
+	if ( n & 1 )
+		/* offset of h-degree */
+		d->td = d->d[n-1] = d0->d[n-1]+d1->d[n-1];
+	else
+		d->td = 0;
+	rtab[0].c = c;
+	rtab[0].d = d;
 
-	n = p1->nv; d1 = BDY(p1)->dl; d2 = BDY(p2)->dl;
-	w = (int *)ALLOCA(n*sizeof(int));
-	for ( i = 0, td = 0; i < n; i++ ) {
-		w[i] = MAX(d1->d[i],d2->d[i]); td += w[i];
+	if ( rtablen > tmptablen ) {
+		if ( tmptab ) GC_free(tmptab);
+		tmptab = (struct cdlm *)MALLOC(rtablen*sizeof(struct cdlm));
+		tmptablen = rtablen; 
 	}
-	NEWDL(d,n); d->td = td - d1->td;
-	for ( i = 0; i < n; i++ )
-		d->d[i] = w[i] - d1->d[i];
-	NEWMP(m); m->dl = d; m->c = BDY(p2)->c; NEXT(m) = 0;
-	MKDP(n,m,s); s->sugar = d->td; _mulmd(CO,mod,p1,s,&t);
-	NEWDL(d,n); d->td = td - d2->td;
-	for ( i = 0; i < n; i++ )
-		d->d[i] = w[i] - d2->d[i];
-	NEWMP(m); m->dl = d; m->c = STOI(mod - ITOS(BDY(p1)->c)); NEXT(m) = 0;
-	MKDP(n,m,s); s->sugar = d->td; _mulmd(CO,mod,p2,s,&u);
-	_addmd(CO,mod,t,u,rp);
+
+	for ( i = 0; i < n2; i++ ) {
+		a = d0->d[i]; b = d1->d[n2+i];
+		k = d0->d[n2+i]; l = d1->d[i];
+		if ( !k || !l ) {
+			a += l;
+			b += k;
+			s = a+b;
+			for ( j = 0, p = rtab; j < curlen; j++, p++ ) {
+				if ( p->c ) {
+					dt = p->d;
+					dt->d[i] = a;
+					dt->d[n2+i] = b;
+					dt->td += s;
+				}
+			}
+			curlen *= k+1;
+			continue;
+		}
+		if ( k+1 > tablen ) {
+			if ( tab ) GC_free(tab);
+			if ( ctab ) GC_free(ctab);
+			tablen = k+1;
+			tab = (struct cdlm *)MALLOC(tablen*sizeof(struct cdlm));
+			ctab = (int *)MALLOC(tablen*sizeof(int));
+		}
+		/* degree of xi^a*(Di^k*xi^l)*Di^b */
+		s = a+k+l+b;
+		/* compute xi^a*(Di^k*xi^l)*Di^b */
+		min = MIN(k,l);
+		mkwcm(k,l,mod,ctab);
+		bzero(tab,(k+1)*sizeof(struct cdlm));
+		/* n&1 != 0 => homogenized computation; dx-xd=h^2 */
+		if ( n & 1 )
+			for ( j = 0; j <= min; j++ ) {
+				NEWDL(d,n);
+				d->d[i] = l-j+a; d->d[n2+i] = k-j+b;
+				d->td = s;
+				d->d[n-1] = s-(d->d[i]+d->d[n2+i]); 
+				tab[j].d = d;
+				tab[j].c = ctab[j];
+			}
+		else
+			for ( j = 0; j <= min; j++ ) {
+				NEWDL(d,n);
+				d->d[i] = l-j+a; d->d[n2+i] = k-j+b;
+				d->td = d->d[i]+d->d[n2+i]; /* XXX */
+				tab[j].d = d;
+				tab[j].c = ctab[j];
+			}
+		comm_mulmd_tab_destructive(mod,n,rtab,curlen,tab,k+1);
+		curlen *= k+1;
+	}
 }
 
-void _dp_sp_component_mod(p1,p2,mod,f1,f2)
-DP p1,p2;
+void comm_mulmd_tab_destructive(mod,nv,t,n,t1,n1)
 int mod;
-DP *f1,*f2;
+int nv;
+struct cdlm *t;
+int n;
+struct cdlm *t1;
+int n1;
 {
-	int i,n,td;
-	int *w;
-	DL d1,d2,d;
-	MP m;
-	DP t,s,u;
+	int i,j;
+	struct cdlm *p;
+	int c;
+	DL d;
 
-	n = p1->nv; d1 = BDY(p1)->dl; d2 = BDY(p2)->dl;
-	w = (int *)ALLOCA(n*sizeof(int));
-	for ( i = 0, td = 0; i < n; i++ ) {
-		w[i] = MAX(d1->d[i],d2->d[i]); td += w[i];
+	for ( j = 1, p = t+n; j < n1; j++ ) {
+		c = t1[j].c;
+		d = t1[j].d;
+		if ( !c )
+			break;
+		for ( i = 0; i < n; i++, p++ ) {
+			if ( t[i].c ) {
+				p->c = dmar(t[i].c,c,0,mod);
+				adddl_dup(nv,t[i].d,d,&p->d);
+			}
+		}
 	}
-	NEWDL(d,n); d->td = td - d1->td;
-	for ( i = 0; i < n; i++ )
-		d->d[i] = w[i] - d1->d[i];
-	NEWMP(m); m->dl = d; m->c = BDY(p2)->c; NEXT(m) = 0;
-	MKDP(n,m,s); s->sugar = d->td; _mulmd(CO,mod,p1,s,f1);
-	NEWDL(d,n); d->td = td - d2->td;
-	for ( i = 0; i < n; i++ )
-		d->d[i] = w[i] - d2->d[i];
-	NEWMP(m); m->dl = d; m->c = BDY(p1)->c; NEXT(m) = 0;
-	MKDP(n,m,s); s->sugar = d->td; _mulmd(CO,mod,p2,s,f2);
+	c = t1[0].c;
+	d = t1[0].d;
+	for ( i = 0, p = t; i < n; i++, p++ )
+		if ( t[i].c ) {
+			p->c = dmar(t[i].c,c,0,mod);
+			/* t[i].d += d */
+			adddl_destructive(nv,t[i].d,d);
+		}
 }
 
-void _printdp(d)
-DP d;
+void adddl_dup(n,d1,d2,dr)
+int n;
+DL d1,d2;
+DL *dr;
 {
-	int n,i;
-	MP m;
-	DL dl;
+	DL dt;
+	int i;
 
-	if ( !d ) {
-		printf("0"); return;
-	}
-	for ( n = d->nv, m = BDY(d); m; m = NEXT(m) ) {
-		printf("%d*<<",ITOS(m->c));
-		for ( i = 0, dl = m->dl; i < n-1; i++ )
-			printf("%d,",dl->d[i]);
-		printf("%d",dl->d[i]);
-		printf(">>");
-		if ( NEXT(m) )
-			printf("+");
-	}
+	NEWDL(dt,n);
+	*dr = dt;
+	dt->td = d1->td + d2->td;
+	for ( i = 0; i < n; i++ )
+		dt->d[i] = d1->d[i]+d2->d[i];
 }