-
Notifications
You must be signed in to change notification settings - Fork 8
/
check.mpl
657 lines (607 loc) · 18 KB
/
check.mpl
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
pm1_stage1 := proc(n, a0, B1)
local p, a, q;
p := 2;
a := a0;
while p <= B1 do
q := 1;
while q*p <= B1 do
q := q*p;
a := Power(a, p) mod n;
od;
p := nextprime(p);
od;
a
end:
Powering["P-1"] := proc(a,i,p) Power(a,i) mod p end:
Select["P-1"] := proc(i) evalb(i mod 6=1) end:
Select["P+1"] := proc(i) member(i mod 6,{1,5}) end:
# stage2, method="P-1" or "P+1"
# p is the modulus
stage2 := proc(p, d, a0, k, method)
local x, F, inva, i, v, t, u, a, dF, G, j, H, ij;
a := a0;
F := 1;
dF := numtheory[phi](d) / 2;
if method = "P-1" then inva := 1/a mod p fi;
for i from 1 to d by 2 do
if Select[method](i) and igcd(i,d)=1 then
v := Powering[method](a,i,p);
if method="P-1" then v := v + Powering[method](inva,i,p) fi;
F := F * (x + v)
fi
od;
F := Expand(F) mod p;
a := Powering[method](a, d, p);
t := 1;
if method="P-1" then
inva := Powering[method](inva, d, p);
u := 1;
fi;
lprint("B2=", k*(dF-1)*d);
ij := 0;
for i to k do
G := 1;
for j from 1 to dF - 1 do
ij := ij + 1;
t := Powering[method](a, ij, p);
v := t;
if method="P-1" then
u := Powering[method](inva, ij, p);
v := v + u;
fi;
G := G * (x + v);
od;
G := Expand(G) mod p;
if i = 1 then H := G
else
G := Expand(G * H) mod p;
H := Rem(G, F, x) mod p;
fi;
od;
G := Gcd(F, H) mod p;
if degree(G)<>0 then lprint("****** Found factor in stage 2: ", p) fi;
end:
list_mul_opt := proc(n) option remember;
if n<=1 then n
elif member(n, {2,5,6,7,8,17,18,23,24,29,30}) then # Karatsuba
2*procname(ceil(n/2))+procname(floor(n/2))
elif member(n, {3,9,10,11,12,20,21,25,26,27}) then # toomcook3
4*procname(ceil(n/3))+procname(n-2*ceil(n/3))
else # toomcook4
6*procname(ceil(n/4))+procname(n-3*ceil(n/4))
fi
end:
# number of scalar multiplies from karatsuba
karatsuba := proc(n) option remember;
if n<=1 then n
else 2*procname(ceil(n/2))+procname(floor(n/2))
fi
end:
# number of scalar multiplies from karatsuba, short product
karatsuba_short := proc(n) option remember;
if n<=1 then n
else procname(ceil(n/2))+2*procname(floor(n/2))
fi
end:
# number of scalar multiplies from Karatsuba
# Mulders' short product (optimal cutoff)
karatsuba_short_mulders := proc(n) option remember; local p, m;
if n<=1 then n
else
m := infinity;
for p from ceil(n/2) to n-1 do
m := min(m, karatsuba(p)+2*procname(n-p))
od;
m
fi
end:
# get m terms, with entries of n terms
karatsuba_short2 := proc(n0, m0) option remember; local n, m;
n := n0;
m := m0;
if m>2*n-1 then m:=2*n-1
elif m<n then n:=m
fi;
if n<=1 then n
else
procname(ceil(n/2), ceil(m/2)) # evaluation at t=0
+ procname(ceil(n/2), ceil((m-1)/2)) # eval. at t=1
+ procname(ceil((n-1)/2), ceil((m-1)/2)) # t = inf
fi
end:
# number of scalar multiplies from toomcook3, as implemented in ecm-5.0.1
toomcook3 := proc(n) option remember; local l, k;
if member (n, {0,1,2,4}) then karatsuba(n)
else
l := iquo(n + 2, 3);
k := n - 2*l;
4*procname(l) + procname(k)
fi
end:
# number of scalar multiplies for Toom3
# Mulders' short product (optimal cutoff)
toomcook3_short_mulders := proc(n) option remember; local p, m, c, s;
if n<=1 then n
else
m := infinity;
for p from ceil(n/2) to n do
c := toomcook3(p)+2*procname(n-p);
if c<m then m:=c; s:={p}
elif c=m then s:=s union {p}
fi
od;
# lprint(n, s);
m
fi
end:
# conjectured optimal cutoff: largest of 3^k or 2*3^k that is between n/2 and n
# seems to be true: works for all n<=1000
toomcook3_short_mulders2 := proc(n) option remember; local p;
if n<=1 then n
else
p := 1;
while 3*p<=n do p:=3*p od;
p := floor(n/p)*p;
toomcook3(p)+2*procname(n-p);
fi
end:
# number of scalar multiplies from toomcook3, short product
# odd-even variant: short (a0(x^3) + x*a1(x^3)+ x^2*a2(x^3), n)
toomcook3_short := proc(n) option remember;
if n <= 2 or n = 5 then karatsuba_short(n)
else
if n mod 3 = 2 then # consider (x*a(x))^2
procname(iquo(n,3)) + 4*procname(iquo(n+2,3))
else procname(ceil(n/3)) + 4*procname(ceil((n-1)/3))
fi
fi
end:
# get m terms, with entries of n terms
toomcook3_short2 := proc(n0, m0) option remember; local n, m, c1, c2;
n := n0;
m := m0;
if m>2*n-1 then m:=2*n-1
elif m<n then n:=m
fi;
if n<=1 then n
elif member([n,m],{[2,2],[2,3],[4,5],[4,6],[4,7],[5,5]}) then karatsuba_short2(n,m)
else
c1 := toomcook3_short2(ceil(n/3), ceil(m/3)) # evaluation at t=0
+ 3*toomcook3_short2(ceil(n/3), ceil((m-1)/3)) # eval. at t=1, -1, 2
+ toomcook3_short2(ceil((n-2)/3), ceil((m-1)/3)); # t = inf
# shift entries by x: n->n+1, m->m+2
c2 := toomcook3_short2(ceil((n+1)/3)-1, ceil((m-4)/3)) # eval. at 0
+ 3*toomcook3_short2(ceil((n+1)/3), ceil((m+1)/3)) # eval. at t=1, -1, 2
+ toomcook3_short2(ceil((n-1)/3), ceil((m+1)/3)); # t = inf
c1 := min(c1, c2);
if karatsuba_short2(n,m) < c1 then
lprint("karatsuba_short2 faster for ", n, m)
fi;
c1
fi
end:
toomcook3_2 := proc(n) option remember; local l0, l1, l2;
if n<=2 or n=4 then karatsuba(n)
else
l2 := iquo(n + 2, 3);
l1 := iquo(n + 1, 3);
l0 := n - l2 - l1;
3*procname(l2) + procname(l0) + procname(l1)
fi
end:
# memory used by toomcook3
M := proc(len) local l;
option remember;
l := iquo(len + 2, 3);
4 * l + max (M(l), 1)
end:
M(0):=0:
M(1):=0:
M(2):=1:
M(4):=5:
# number of scalar multiplies from toomcook4
toomcook4 := proc(n) option remember; local l, k;
if member(n,{0,1,2,3,5,6,9,17,18,25,26,27,77,78,79,80,81}) then toomcook3(n)
else
l := iquo(n + 3, 4);
k := n - 3 * l;
6*procname(l) + procname(k)
fi
end:
# find optimal method between kara, toom3 and toom4
find_opt := proc(nmax) local n, T, kara, toom3, toom4;
T[0]:=0;
T[1]:=1;
for n from 2 to nmax do
kara := 2*T[ceil(n/2)]+T[floor(n/2)];
toom3 := 4*T[ceil(n/3)]+T[n-2*ceil(n/3)];
if n>=4 and n<>5 then
toom4 := 6*T[ceil(n/4)]+T[n-3*ceil(n/4)]
else
toom4 := kara;
fi;
if kara<=min(toom3,toom4) then lprint(n, "karatsuba", kara); T[n]:=kara
elif toom3<=toom4 then lprint(n, "toomcook3", toom3); T[n]:=toom3
else T[n]:=toom4
fi
od;
end:
# number of scalar multiplies for Toom3
# Mulders' short product (optimal cutoff)
toomcook4_short_mulders := proc(n) option remember; local p, m, c, s;
if n<=1 then n
else
m := infinity;
for p from ceil(n/2) to n do
c := toomcook4(p)+2*procname(n-p);
if c<m then m:=c; s:={p}
elif c=m then s:=s union {p}
fi
od;
lprint(n, s);
m
fi
end:
# conjectured optimal cutoff: 4^k or 2*4^k or 3*4^k that is between n/2 and n
# works for almost all n: exceptions are n=32 (p=27),
# n=125-130 (108)
toomcook4_short_mulders2 := proc(n) option remember; local p;
if n<=1 then n
else
p := 1;
while 4*p<=n do p:=4*p od;
p:=floor(n/p)*p;
toomcook4(p)+2*toomcook4_short_mulders(n-p);
fi
end:
# number of scalar multiplies from toomcook4, short product
toomcook4_short := proc(n) option remember;
if n <= 3 or member (n, {6,7,10,22,30,31,90,91,94}) then toomcook3_short(n)
else
min(procname(ceil(n/4)) + 6*procname(ceil((n-1)/4)),
procname(ceil((n+1)/4)-1) + 6*procname(ceil(n/4)))
fi
end:
toomcook4_2 := proc(n) option remember; local l2, l1, l0;
if n<=3 or member(n,{5,6,9,17,18,25,26,27,77,78,79,80,81}) then toomcook3_2(n)
else
l2 := iquo(n + 3, 4);
l1 := iquo(n - 2*l2 + 1, 2);
if member(n,{22,30,41,42,45,54,57,70,73,82,85,86,94}) then l1:=l2 fi;
l0 := n - 2*l2 - l1;
5*procname(l2) + procname(l1) + procname(l0)
fi
end:
short_mul := proc(n) option remember; local k;
if member(n,{1,4,14,15,16,$56..64,$221..256}) then toomcook4(n)
else
k := 1; while 2*k < n do k:=2*k od;
toomcook4(k) + 2*procname(n-k)
fi
end:
# assumes k>=l
list_mul := proc(k, l)
if k=l then toomcook4(l)
elif k=l+1 then toomcook4(l) + l # special important case
else toomcook4(l) + list_mul(max(k-l, l), min(k-l, l))
fi
end:
# number of multiplies of PolyFromRoots
PolyFromRoots := proc(k) option remember; local l, m;
if k<=1 then 0
elif k=2 then 1
else
m := iquo(k, 2);
l := k - m; # l = k or l = k + 1
procname(l) + procname (m) + list_mul(l, m)
fi
end:
# PolyFromRoots with optimal cutoff point
# (depends on the multiplication used, and the way list_mul deals with
# operands of different degree)
PolyFromRoots_opt := proc(k) option remember; local l, m, s, c, cmin;
if k<=1 then 0
elif k=2 then 1
else
cmin := infinity;
for m from 1 to iquo(k,2) do
l := k - m;
c := procname(l) + procname (m) + list_mul(l, m);
if c<cmin then
cmin := c;
s := {l};
elif c=cmin then s:=s union {l}
fi
od;
lprint(k, s);
cmin
fi
end:
# number of multiplies in RecursiveDivision
RecursiveDivision := proc(K) option remember; local k, l;
if K=1 then 1
else
k := iquo(K, 2);
l := K - k;
procname(l) + 2*toomcook4(k) + 2*k * (l-k) + procname(k)
fi
end:
Reduce := proc(n) 2*toomcook4(n-1) + 1 end:
rootsF := proc(d, S)
2.5 * 6 * (d/6) * S
end:
rootsG := proc(dF, S)
2.5 * 6 * dF * S
end:
step2_cost := proc(B2, d, S) option remember; local dF, k, a, b;
dF := numtheory[phi](d)/2;
k := ceil (B2 / d / dF);
a := PolyFromRoots(dF) + PolyInvert(dF-1) - toomcook4(dF) -
Reduce(dF)
+ PolyEval(dF) + rootsF(d, S);
b := PolyFromRoots(dF) # Building G from its roots
+ toomcook4(dF) # Computing G * H
+ Reduce(dF) # Reducing G * H mod F
+ rootsG(dF, S);
k, a, b, a + k * b;
end:
PolyInvert := proc(n) option remember; local k, l, v;
if n=1 then 0
else
k := iquo(n, 2);
l := n - k;
v := procname(l) + toomcook4(l) + toomcook4(k);
if k > 1 then v := v + list_mul (l-1, k-1) fi;
v
fi
end:
PolyEval := proc(k) option remember; local m, l, v;
if k=1 then 0
else
m := iquo(k, 2);
l := k - m;
v := RecursiveDivision(m);
if k > 2*m then v := v + m fi;
v + RecursiveDivision(l) + procname(l) + procname(m)
fi
end:
# output list with increasing phi(d) and decreasing step2_cost(d)
gen_bestD := proc(d0, d1) local l, d, c, p, i, j;
l := [[d0,numtheory[phi](d0),step2_cost(d0)]];
for d from d0+6 by 6 to d1 do
p := numtheory[phi](d);
c := step2_cost(d);
for i to nops(l) while p > l[i][2] and c < l[i][3] do od;
# now i > nops(l) or phi(d) <= phi(l[i]) or c >= step2_cost(l[i])
if i > nops(l) then l:=[op(l),[d,p,c]]
elif p <= l[i][2] then # p <= l[j][2] for j >= i
for j from i to nops(l) while c <= l[j][3] do od;
l:=[op(1..i-1, l), [d,p,c], op(j..nops(l), l)]
else # p > l[i][2] and c >= step2_cost(l[i])
fi
od;
l
end:
# estimate number of multiplies of PolyGcd(F, G)
# with deg(F)=n and deg(G)=n-1
PolyGcd := proc(n)
if n<=1 then 0 # gcd is G
else HalfGcd(n,0) + PolyGcd(ceil(n/2))
fi
end:
# deg(F)=n and deg(G)=n-1, assumes return cofactors
# and matrix if flag=1
HalfGcd := proc(n,flag) option remember; local k, l, c;
if n<=1 then 0
else
k := ceil(n/2);
l := ceil(n/4);
c := procname(k, 1) # return 2x2 matrix with degrees n/4
+ 8*toomcook4(l) # 4 multiplies of n/2 * n/4
+ procname(k, flag) # 2nd call
+ 4*toomcook4(l); # 4 multiplies of n/4 * n/4
if flag=1 then c:=c+7*toomcook4(l) fi; # multiply matrices
c
fi
end:
# auxiliary memory needed for karatsuba
M := proc(K) option remember; local l;
if K=1 then 0
else
l := iquo(K+1, 2);
max(2*l-1+l,2*l-2+M(l))
fi;
end:
# cf Williams, Math. of Comp. 39, 1982
# pp1_stage1(328006342451, 5, 7043); # 2^235+1
# pp1_stage1(6215074747201, 5, 199729); # 2^297+1
# pp1_stage1(8857714771093, 3, 49207); # 2^418+1
# pp1_stage1(236344687097, 3, 55001); # 2^602+1
# pp1_stage1(87251820842149, 5, 170249); # 2^642+1
# pp1_stage1(719571227339189, 4, 57679); # 3^134+1
# pp1_stage1(5468575720021, 6, 175759); # 3^161+1
# pp1_stage1(49804972211, 5, 268757); # 3^185-1
# pp1_stage1(329573417220613, 3, 101573); # 5^94+1
# pp1_stage1(4866979762781, 4, 97609); # 6^59-1
# pp1_stage1(187333846633, 3, 9851); # 6^111-1
# pp1_stage1(332526664667473, 3, 111919); # 6^132+1
# pp1_stage1(265043186297, 3, 152791); # 7^133+1
# pp1_stage1(207734163253, 3, 4211); # 7^231+1
# pp1_stage1(225974065503889, 5, 8243); # 10^102+1
# pp1_stage1(660198074631409, 5, 115679); # 12^81-1
# pp1_stage1(563215815517, 3, 109849); # 12^183+1
# pp1_stage1(409100738617, 3, 70957); # fib(247)
# pp1_stage1(7901346123803597, 3, 18307); # fib(313)
# pp1_stage1(5648966761, 15, 100519); # fib(367)
# pp1_stage1(14279673833, 3, 823); # fib(387)
# pp1_stage1(1795220677069, 6, 159931); # fib(483)
# pp1_stage1(1250839826281, 5, 4673); # fib(495)
# pp1_stage1(2192843129417, 3, 38803); # fib(531)
# pp1_stage1(10424083697, 3, 131); # fib(549)
pp1_stage1 := proc(n, A0, B1) local a0, p, a, q;
# suggested default choice from Montgomery
if A0=0 then a0:=2/7 mod n else a0:=A0 fi;
if igcd(a0^2-4, n)<>1 then ERROR("igcd(a0^2-4, n)<>1") fi;
if isprime(n) and numtheory[jacobi](a0^2-4, n)=1 then
lprint("Warning: jacobi(a0^2-4, n) = 1")
fi;
p := 2;
a := a0;
while p <= B1 do
q := p;
while q*p <= B1 do q:=q*p od;
a := Powering["P+1"](a, q, n);
p := nextprime(p);
od;
p:=igcd(a-2, n);
if p<>1 then lprint("****** Found factor in stage 1: ", p) fi;
a
end:
# compute P0 "power" p mod n for P+1
Powering["P+1"] := proc(P0, p, n) local l, i, P, Q, R;
l := convert(p-1, base, 2);
P := P0;
Q := 2;
for i from nops(l) by -1 to 1 do
if l[i] = 1 then # (i,i-1) to (2i,2i-1)
Q := P*Q-P0 mod n;
P := P^2-2 mod n;
else # (i,i-1) to (2i-1,2i-2)
P := P*Q-P0 mod n;
Q := Q^2-2 mod n;
fi
od;
P
end:
############################ ecm ###################################
# converts (x:1:1) to Weierstrass form (mod p)
# returns [X,Y,A]
montgomery_to_weierstrass := proc(x, a, p) local g;
g := (x^3 + a*x^2 + x) mod p;
[(3*x+a)/(3*g) mod p, 1/g mod p, (3-a^2)/(3*g^2) mod p]
end:
# return 2*(x1:z1)
dup := proc(x1, z1, b, n) local u, v, w, x2, z2;
u := x1+z1 mod n;
u := u^2 mod n;
v := x1-z1 mod n;
v := v^2 mod n;
x2 := u*v mod n;
w := u-v mod n;
u := w*b mod n;
u := u+v mod n;
z2 := w*u mod n;
return x2, z2;
end:
# return (x1:z1) + (x2:z2), where (x:z) = (x1:z1) - (x2:z2)
add3 := proc(x1, z1, x2, z2, x, z, n) local u, v, w, x3, z3;
u := x2-z2 mod n;
v := x1+z1 mod n;
u := u*v mod n;
w := x2+z2 mod n;
v := x1-z1 mod n;
v := w*v mod n;
w := u+v mod n;
v := u-v mod n;
w := w*w mod n;
v := v*v mod n;
x3 := w*z mod n;
z3 := x*v mod n;
return x3, z3;
end:
addW := proc(x1, y1, x2, y2, n) local u, v, p;
u := x2-x1;
v := 1/u mod n; # 1/(x2-x1)
p := (y2-y1)*v mod n; # lambda=(y2-y1)/(x2-x1)
u := p*p-x1; # lambda^2-x1
v := u-x2 mod n; # lambda^2-x1-x2
u := (x1-v)*p; # (2x1+x2-lambda^2)*lambda
[v, u-y1 mod n]
end:
# (x::y) -> 2*(x::y)
dupW := proc(x, y, n, a) local u, v, p;
v := 1/(2*y) mod n;
u := 3*x^2+a mod n;
p := u*v mod n;
u := p^2;
v := 2*x;
u := u-v mod n;
[u, (x-u)*p-y mod n]
end:
# (x::y) -> k*(x::y)
mulW := proc(x, y, k, n, a) local l, P, i;
if k=1 then [x,y]
else # k >= 3
l := convert(k, base, 2);
P := [x, y];
for i from nops(l)-1 by -1 to 1 do
P := dupW(op(P), n, a);
if l[i]=1 then P := addW(op(P), x, y, n) fi
od;
P
fi
end:
##############################################################################
# odd-even variant
kara_short_mul := proc(a, b, n)
local a0, a1, b0, b1, c0, c1, c2, p, q, r, i, res;
if n = 0 then []
elif n = 1 then [a[1]*b[1]]
else
p := ceil(n/2);
q := ceil((n-1)/2);
r := q;
a0 := [seq(a[2*i-1],i=1..p)];
b0 := [seq(b[2*i-1],i=1..p)];
a1 := [seq(a[2*i], i=1..q)];
b1 := [seq(b[2*i], i=1..q)];
c0 := procname(a0, b0, p);
c1 := procname(a0[1..q]+a1, b0[1..q]+b1, q);
c2 := procname(a1, b1, r);
c1 := c1 - c0[1..q] - [op(c2),0$(q-r)];
res := [0$n];
for i to p do res[2*i-1]:=c0[i] od;
for i to min(r,iquo(n-1,2)) do res[2*i+1]:=res[2*i+1]+c2[i] od;
for i to q do res[2*i]:=c1[i] od;
res
fi
end:
# odd-even variant
toom3_short_mul := proc(a, b, n)
local a0, a1, a2, b0, b1, b2, c0, c1, c2, c3, c4, p, q, r, i, res;
if n = 0 then []
elif n = 1 then [a[1]*b[1]]
elif n = 2 then [a[1]*b[1], a[1]*b[2]+a[2]*b[1]]
else
p := ceil(n/3);
q := ceil((n-1)/3);
r := ceil((n-2)/3);
a0 := [seq(a[3*i-2],i=1..p)];
b0 := [seq(b[3*i-2],i=1..p)];
a1 := [seq(a[3*i-1], i=1..q)];
b1 := [seq(b[3*i-1], i=1..q)];
a2 := [seq(a[3*i], i=1..r), 0$(q-r)];
b2 := [seq(b[3*i], i=1..r), 0$(q-r)];
c0 := procname(a0, b0, p); # 0
c1 := procname(a0[1..q]+a1+a2, b0[1..q]+b1+b2, q); # 1
c2 := procname(a0[1..q]-a1+a2, b0[1..q]-b1+b2, q); # -1
c3 := procname(a0[1..q]+2*a1+4*a2, b0[1..q]+2*b1+4*b2, q); # 2
c4 := procname(a2, b2, q);
c1 := c1 - c0[1..q] - c4; # d1+d2+d3
c2 := c2 - c0[1..q] - c4; # -d1+d2-d3
c3 := c3 - c0[1..q] - 16*c4; # 2*d1+4*d2+8*d3
c1 := (c1 + c2)/2; # d2
c2 := c2 - c1; # -d1-d3
c3 := c3 - 4*c1; # 2*d1+8*d3
c3 := c3 + 2*c2; # 6*d3
c3 := c3/6; # d3
c2 := -c2-c3; # d1
res := [0$n];
for i to p do res[3*i-2]:=c0[i] od;
for i to q do res[3*i-1]:=res[3*i-1]+c2[i] od;
for i to ceil((n-2)/3) do res[3*i]:=res[3*i]+c1[i] od;
for i to ceil((n-3)/3) do res[3*i+1]:=res[3*i+1]+c3[i] od;
for i to ceil((n-4)/3) do res[3*i+2]:=res[3*i+2]+c4[i] od;
res
fi
end: