Annotation of OpenXM_contrib2/asir2018/builtin/pdiv.c, Revision 1.2
1.1 noro 1: /*
2: * Copyright (c) 1994-2000 FUJITSU LABORATORIES LIMITED
3: * All rights reserved.
4: *
5: * FUJITSU LABORATORIES LIMITED ("FLL") hereby grants you a limited,
6: * non-exclusive and royalty-free license to use, copy, modify and
7: * redistribute, solely for non-commercial and non-profit purposes, the
8: * computer program, "Risa/Asir" ("SOFTWARE"), subject to the terms and
9: * conditions of this Agreement. For the avoidance of doubt, you acquire
10: * only a limited right to use the SOFTWARE hereunder, and FLL or any
11: * third party developer retains all rights, including but not limited to
12: * copyrights, in and to the SOFTWARE.
13: *
14: * (1) FLL does not grant you a license in any way for commercial
15: * purposes. You may use the SOFTWARE only for non-commercial and
16: * non-profit purposes only, such as academic, research and internal
17: * business use.
18: * (2) The SOFTWARE is protected by the Copyright Law of Japan and
19: * international copyright treaties. If you make copies of the SOFTWARE,
20: * with or without modification, as permitted hereunder, you shall affix
21: * to all such copies of the SOFTWARE the above copyright notice.
22: * (3) An explicit reference to this SOFTWARE and its copyright owner
23: * shall be made on your publication or presentation in any form of the
24: * results obtained by use of the SOFTWARE.
25: * (4) In the event that you modify the SOFTWARE, you shall notify FLL by
26: * e-mail at risa-admin@sec.flab.fujitsu.co.jp of the detailed specification
27: * for such modification or the source code of the modified part of the
28: * SOFTWARE.
29: *
30: * THE SOFTWARE IS PROVIDED AS IS WITHOUT ANY WARRANTY OF ANY KIND. FLL
31: * MAKES ABSOLUTELY NO WARRANTIES, EXPRESSED, IMPLIED OR STATUTORY, AND
32: * EXPRESSLY DISCLAIMS ANY IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS
33: * FOR A PARTICULAR PURPOSE OR NONINFRINGEMENT OF THIRD PARTIES'
34: * RIGHTS. NO FLL DEALER, AGENT, EMPLOYEES IS AUTHORIZED TO MAKE ANY
35: * MODIFICATIONS, EXTENSIONS, OR ADDITIONS TO THIS WARRANTY.
36: * UNDER NO CIRCUMSTANCES AND UNDER NO LEGAL THEORY, TORT, CONTRACT,
37: * OR OTHERWISE, SHALL FLL BE LIABLE TO YOU OR ANY OTHER PERSON FOR ANY
38: * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, PUNITIVE OR CONSEQUENTIAL
39: * DAMAGES OF ANY CHARACTER, INCLUDING, WITHOUT LIMITATION, DAMAGES
40: * ARISING OUT OF OR RELATING TO THE SOFTWARE OR THIS AGREEMENT, DAMAGES
41: * FOR LOSS OF GOODWILL, WORK STOPPAGE, OR LOSS OF DATA, OR FOR ANY
42: * DAMAGES, EVEN IF FLL SHALL HAVE BEEN INFORMED OF THE POSSIBILITY OF
43: * SUCH DAMAGES, OR FOR ANY CLAIM BY ANY OTHER PARTY. EVEN IF A PART
44: * OF THE SOFTWARE HAS BEEN DEVELOPED BY A THIRD PARTY, THE THIRD PARTY
45: * DEVELOPER SHALL HAVE NO LIABILITY IN CONNECTION WITH THE USE,
46: * PERFORMANCE OR NON-PERFORMANCE OF THE SOFTWARE.
47: *
1.2 ! noro 48: * $OpenXM: OpenXM_contrib2/asir2018/builtin/pdiv.c,v 1.1 2018/09/19 05:45:06 noro Exp $
1.1 noro 49: */
50: #include "ca.h"
51: #include "parse.h"
52:
53: void Psdiv(), Psrem(), Ptdiv(), Psqr(), Pinva_mod(), Pprem();
54: void Psdiv_gf2n(), Psrem_gf2n(), Pgcd_gf2n();
55: void Psdivm(), Psremm(), Psqrm();
56: void Psrem_mod();
57: void Pugcd();
58: void Purem();
59: void Pudiv();
60:
61: struct ftab pdiv_tab[] = {
62: {"sdiv",Psdiv,-3},
63: {"srem",Psrem,-3},
64: {"prem",Pprem,-3},
65: {"sdiv_gf2n",Psdiv_gf2n,2},
66: {"srem_gf2n",Psrem_gf2n,2},
67: {"gcd_gf2n",Pgcd_gf2n,2},
68: {"sqr",Psqr,-3},
69: {"tdiv",Ptdiv,-3},
70: {"udiv",Pudiv,2},
71: {"sdivm",Psdivm,-4},
72: {"sremm",Psremm,-4},
73: {"sqrm",Psqrm,-4},
74: {"inva_mod",Pinva_mod,3},
75: {"srem_mod",Psrem_mod,3},
76: {"ugcd",Pugcd,2},
77: {"urem",Purem,2},
78: {0,0,0},
79: };
80:
81: void Psdiv(arg,rp)
82: NODE arg;
83: Obj *rp;
84: {
85: P q,r,dnd,dnd1,dvr,dvr1;
86: V v;
87: VL vl;
88:
89: asir_assert(ARG0(arg),O_P,"sdiv");
90: asir_assert(ARG1(arg),O_P,"sdiv");
91: dnd = (P)ARG0(arg); dvr = (P)ARG1(arg);
92: if ( argc(arg) == 3 ) {
93: v = VR((P)ARG2(arg));
94: change_mvar(CO,dnd,v,&dnd1); change_mvar(CO,dvr,v,&dvr1);
95: reordvar(CO,v,&vl);
96: divsrp(vl,dnd1,dvr1,&q,&r);
97: restore_mvar(CO,q,v,(P *)rp);
98: } else
99: divsrp(CO,dnd,dvr,(P *)rp,&r);
100: }
101:
102: void Psrem(arg,rp)
103: NODE arg;
104: Obj *rp;
105: {
106: P q,r,dnd,dnd1,dvr,dvr1;
107: V v;
108: VL vl;
109:
110: asir_assert(ARG0(arg),O_P,"srem");
111: asir_assert(ARG1(arg),O_P,"srem");
112: dnd = (P)ARG0(arg); dvr = (P)ARG1(arg);
113: if ( argc(arg) == 3 ) {
114: v = VR((P)ARG2(arg));
115: change_mvar(CO,dnd,v,&dnd1); change_mvar(CO,dvr,v,&dvr1);
116: reordvar(CO,v,&vl);
117: divsrp(vl,dnd1,dvr1,&q,&r);
118: restore_mvar(CO,r,v,(P *)rp);
119: } else
120: divsrp(CO,dnd,dvr,&q,(P *)rp);
121: }
122:
123: void Pprem(arg,rp)
124: NODE arg;
125: P *rp;
126: {
127: P q,r,dnd,dnd1,dvr,dvr1;
128: V v;
129: VL vl;
130:
131: asir_assert(ARG0(arg),O_P,"prem");
132: asir_assert(ARG1(arg),O_P,"prem");
133: dnd = (P)ARG0(arg); dvr = (P)ARG1(arg);
134: if ( !dvr ) error("prem : division by 0");
135: if ( !dnd ) {
136: *rp = 0; return;
137: }
138: if ( argc(arg) == 3 ) {
139: v = VR((P)ARG2(arg));
140: change_mvar(CO,dnd,v,&dnd1); change_mvar(CO,dvr,v,&dvr1);
141: reordvar(CO,v,&vl);
142: premp(vl,dnd1,dvr1,&r);
143: restore_mvar(CO,r,v,rp);
144: } else
145: premp(CO,dnd,dvr,rp);
146: }
147:
148: void Psqr(arg,rp)
149: NODE arg;
150: LIST *rp;
151: {
152: P q,q1,r,r1,dnd,dnd1,dvr,dvr1;
153: NODE n,tn;
154: V v;
155: VL vl;
156:
157: asir_assert(ARG0(arg),O_P,"sqr");
158: asir_assert(ARG1(arg),O_P,"sqr");
159: dnd = (P)ARG0(arg); dvr = (P)ARG1(arg);
160: if ( argc(arg) == 3 ) {
161: v = VR((P)ARG2(arg));
162: change_mvar(CO,dnd,v,&dnd1); change_mvar(CO,dvr,v,&dvr1);
163: reordvar(CO,v,&vl);
164: divsrp(vl,dnd1,dvr1,&q1,&r1);
165: restore_mvar(CO,q1,v,&q); restore_mvar(CO,r1,v,&r);
166: } else
167: divsrp(CO,dnd,dvr,&q,&r);
168: MKNODE(tn,r,0); MKNODE(n,q,tn); MKLIST(*rp,n);
169: }
170:
171: void Psdiv_gf2n(arg,rp)
172: NODE arg;
173: GF2N *rp;
174: {
175: GF2N dnd,dvr;
176: UP2 q,r;
177:
178: dnd = (GF2N)ARG0(arg); dvr = (GF2N)ARG1(arg);
179: if ( !dvr )
180: error("sdiv_gf2n : division by 0");
181: else if ( !dnd )
182: *rp = 0;
183: else {
184: qrup2(dnd->body,dvr->body,&q,&r);
185: MKGF2N(q,*rp);
186: }
187: }
188:
189: void Psrem_gf2n(arg,rp)
190: NODE arg;
191: GF2N *rp;
192: {
193: GF2N dnd,dvr;
194: UP2 q,r;
195:
196: dnd = (GF2N)ARG0(arg); dvr = (GF2N)ARG1(arg);
197: if ( !dvr )
198: error("srem_gf2n : division by 0");
199: else if ( !dnd )
200: *rp = 0;
201: else {
202: qrup2(dnd->body,dvr->body,&q,&r);
203: MKGF2N(r,*rp);
204: }
205: }
206:
207: void Pgcd_gf2n(arg,rp)
208: NODE arg;
209: GF2N *rp;
210: {
211: GF2N p1,p2;
212: UP2 gcd;
213:
214: p1 = (GF2N)ARG0(arg); p2 = (GF2N)ARG1(arg);
215: if ( !p1 )
216: *rp = p2;
217: else if ( !p2 )
218: *rp = p1;
219: else {
220: gcdup2(p1->body,p2->body,&gcd);
221: MKGF2N(gcd,*rp);
222: }
223: }
224:
225: void Ptdiv(arg,rp)
226: NODE arg;
227: P *rp;
228: {
229: P p1,p2,q1,q2,q,c1,c2,c;
230: int m;
231:
232: p1 = (P)ARG0(arg); p2 = (P)ARG1(arg);
233: asir_assert(p1,O_P,"tdiv");
234: asir_assert(p2,O_P,"tdiv");
235: if ( !p1 || !p2 )
236: *rp = 0;
237: else if ( (OID(p1) > O_P) || (OID(p2) > O_P ) )
238: *rp = 0;
239: else if ( argc(arg) == 3 ) {
1.2 ! noro 240: m = ZTOS((Q)ARG2(arg));
1.1 noro 241: ptomp(m,p1,&q1); ptomp(m,p2,&q2);
242: if ( divtmp(CO,m,q1,q2,&q) )
243: mptop(q,rp);
244: else
245: *rp = 0;
246: } else if ( qpcheck((Obj)p1) && qpcheck((Obj)p2) ) {
247: ptozp(p1,1,(Q *)&c1,&q1); ptozp(p2,1,(Q *)&c2,&q2);
248: if ( divtpz(CO,q1,q2,&q) ) {
249: divq((Q)c1,(Q)c2,(Q *)&c); mulp(CO,q,c,rp);
250: } else
251: *rp = 0;
252: } else {
253: if ( !divtp(CO,p1,p2,rp) )
254: *rp = 0;
255: }
256: }
257:
258: void Pudiv(arg,rp)
259: NODE arg;
260: LIST *rp;
261: {
262: P q,r,dnd,dvr;
263: NODE n,tn;
264:
265: asir_assert(ARG0(arg),O_P,"udiv");
266: asir_assert(ARG1(arg),O_P,"udiv");
267: dnd = (P)ARG0(arg); dvr = (P)ARG1(arg);
268: udivpz(dnd,dvr,&q,&r);
269: MKNODE(tn,r,0); MKNODE(n,q,tn); MKLIST(*rp,n);
270: }
271:
272: void Psdivm(arg,rp)
273: NODE arg;
274: Obj *rp;
275: {
276: P q,r,dnd,dnd1,dndm,dvr,dvr1,dvrm,t;
277: V v;
278: VL vl;
279: int m;
280:
281: asir_assert(ARG0(arg),O_P,"sdivm");
282: asir_assert(ARG1(arg),O_P,"sdivm");
283: asir_assert(ARG2(arg),O_N,"sdivm");
1.2 ! noro 284: dnd = (P)ARG0(arg); dvr = (P)ARG1(arg); m = ZTOS((Q)ARG2(arg));
1.1 noro 285: if ( argc(arg) == 4 ) {
286: v = VR((P)ARG3(arg));
287: change_mvar(CO,dnd,v,&dnd1); change_mvar(CO,dvr,v,&dvr1);
288: reordvar(CO,v,&vl);
289: ptomp(m,dnd1,&dndm); ptomp(m,dvr1,&dvrm);
290: divsrmp(vl,m,dndm,dvrm,&t,&r); mptop(t,&q);
291: restore_mvar(CO,q,v,(P *)rp);
292: } else {
293: ptomp(m,dnd,&dndm); ptomp(m,dvr,&dvrm);
294: divsrmp(CO,m,dndm,dvrm,&t,&r); mptop(t,(P *)rp);
295: }
296: }
297:
298: void Psremm(arg,rp)
299: NODE arg;
300: Obj *rp;
301: {
302: P q,r,dnd,dnd1,dndm,dvr,dvr1,dvrm,t;
303: V v;
304: VL vl;
305: int m;
306:
307: asir_assert(ARG0(arg),O_P,"sremm");
308: asir_assert(ARG1(arg),O_P,"sremm");
309: asir_assert(ARG2(arg),O_N,"sremm");
1.2 ! noro 310: dnd = (P)ARG0(arg); dvr = (P)ARG1(arg); m = ZTOS((Q)ARG2(arg));
1.1 noro 311: if ( argc(arg) == 4 ) {
312: v = VR((P)ARG3(arg));
313: change_mvar(CO,dnd,v,&dnd1); change_mvar(CO,dvr,v,&dvr1);
314: reordvar(CO,v,&vl);
315: ptomp(m,dnd1,&dndm); ptomp(m,dvr1,&dvrm);
316: divsrmp(vl,m,dndm,dvrm,&q,&t); mptop(t,&r);
317: restore_mvar(CO,r,v,(P *)rp);
318: } else {
319: ptomp(m,dnd,&dndm); ptomp(m,dvr,&dvrm);
320: divsrmp(CO,m,dndm,dvrm,&q,&t); mptop(t,(P *)rp);
321: }
322: }
323:
324: void Psqrm(arg,rp)
325: NODE arg;
326: LIST *rp;
327: {
328: P q,q1,r,r1,dnd,dnd1,dndm,dvr,dvr1,dvrm;
329: NODE n,tn;
330: V v;
331: VL vl;
332: int m;
333:
334: asir_assert(ARG0(arg),O_P,"sqrm");
335: asir_assert(ARG1(arg),O_P,"sqrm");
336: asir_assert(ARG2(arg),O_N,"sqrm");
1.2 ! noro 337: dnd = (P)ARG0(arg); dvr = (P)ARG1(arg); m = ZTOS((Q)ARG2(arg));
1.1 noro 338: if ( argc(arg) == 4 ) {
339: v = VR((P)ARG3(arg));
340: change_mvar(CO,dnd,v,&dnd1); change_mvar(CO,dvr,v,&dvr1);
341: reordvar(CO,v,&vl);
342: ptomp(m,dnd1,&dndm); ptomp(m,dvr1,&dvrm);
343: divsrmp(vl,m,dndm,dvrm,&q,&r); mptop(q,&q1); mptop(r,&r1);
344: restore_mvar(CO,q1,v,&q); restore_mvar(CO,r1,v,&r);
345: } else {
346: ptomp(m,dnd,&dndm); ptomp(m,dvr,&dvrm);
347: divsrmp(CO,m,dndm,dvrm,&q1,&r1); mptop(q1,&q); mptop(r1,&r);
348: }
349: MKNODE(tn,r,0); MKNODE(n,q,tn); MKLIST(*rp,n);
350: }
351:
352: void Pinva_mod(arg,rp)
353: NODE arg;
354: P *rp;
355: {
356: P dp,f;
357: Z q;
358: int n,i;
359: int mod;
360: V v;
361: UM wf,wdp,winv;
362:
363: asir_assert(ARG0(arg),O_P,"gcda_mod");
364: asir_assert(ARG1(arg),O_N,"gcda_mod");
365: asir_assert(ARG2(arg),O_P,"gcda_mod");
366: dp = (P)ARG0(arg);
1.2 ! noro 367: mod = ZTOS((Q)ARG1(arg));
1.1 noro 368: f = (P)ARG2(arg);
369: if ( NUM(f) ) {
370: i = invm(remqi((Q)f,mod),mod);
1.2 ! noro 371: STOZ(i,q); *rp = (P)q;
1.1 noro 372: } else {
373: v = VR(dp);
374: n = MAX(UDEG(dp),UDEG(f));
375: wf = W_UMALLOC(n); wdp = W_UMALLOC(n);
376: winv = W_UMALLOC(n);
377: ptoum(mod,f,wf); ptoum(mod,dp,wdp);
378: invum(mod,wdp,wf,winv);
379: if ( DEG(winv) < 0 )
380: *rp = 0;
381: else {
382: umtop(v,winv,rp);
383: }
384: }
385: }
386:
387: void Psrem_mod(arg,rp)
388: NODE arg;
389: P *rp;
390: {
391: P p1,p2;
392: int n,dr;
393: int mod;
394: V v;
395: UM wp1,wp2,q;
396:
397: asir_assert(ARG0(arg),O_P,"srem_mod");
398: asir_assert(ARG1(arg),O_P,"srem_mod");
399: asir_assert(ARG2(arg),O_N,"srem_mod");
1.2 ! noro 400: p1 = (P)ARG0(arg); p2 = (P)ARG1(arg); mod = ZTOS((Q)ARG2(arg));
1.1 noro 401: if ( !p1 || NUM(p1) )
402: *rp = p1;
403: else {
404: v = VR(p1);
405: n = MAX(UDEG(p1),UDEG(p2));
406: wp1 = W_UMALLOC(n); wp2 = W_UMALLOC(n); q = W_UMALLOC(n);
407: ptoum(mod,p1,wp1); ptoum(mod,p2,wp2);
408: dr = divum(mod,wp1,wp2,q);
409: if ( ( DEG(wp1) = dr ) == -1 )
410: *rp = 0;
411: else
412: umtop(v,wp1,rp);
413: }
414: }
415:
416: void Purem(arg,rp)
417: NODE arg;
418: P *rp;
419: {
420: asir_assert(ARG0(arg),O_P,"urem");
421: asir_assert(ARG1(arg),O_P,"urem");
422: uremp((P)ARG0(arg),(P)ARG1(arg),rp);
423: }
424:
425: void Pugcd(arg,rp)
426: NODE arg;
427: P *rp;
428: {
429: asir_assert(ARG0(arg),O_P,"ugcd");
430: asir_assert(ARG1(arg),O_P,"ugcd");
431: ugcdp((P)ARG0(arg),(P)ARG1(arg),rp);
432: }
433:
434: void invum(mod,dp,f,inv)
435: int mod;
436: UM dp,f,inv;
437: {
438: UM g1,g2,a1,a2,a3,wm,q,tum;
439: int d,dr;
440:
441: d = DEG(dp)+DEG(f)+10;
442: g1 = W_UMALLOC(d); g2 = W_UMALLOC(d); a1 = W_UMALLOC(d);
443: a2 = W_UMALLOC(d); a3 = W_UMALLOC(d); wm = W_UMALLOC(d);
444: q = W_UMALLOC(d);
445: DEG(a1) = 0; COEF(a1)[0] = 1; DEG(a2) = -1;
446: cpyum(f,g1); cpyum(dp,g2);
447: while ( 1 ) {
448: dr = divum(mod,g1,g2,q); tum = g1; g1 = g2; g2 = tum;
449: if ( ( DEG(g2) = dr ) == -1 )
450: break;
451: mulum(mod,a2,q,wm); subum(mod,a1,wm,a3); dr = divum(mod,a3,dp,q);
452: tum = a1; a1 = a2; a2 = a3; a3 = tum; DEG(a3) = dr;
453: }
454: if ( DEG(g1) != 0 )
455: DEG(inv) = -1;
456: else if ( COEF(g1)[0] != 1 )
457: mulsum(mod,a2,invm(COEF(g1)[0],mod),inv);
458: else
459: cpyum(a2,inv);
460: }
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>