This file is indexed.

/usr/include/linbox/algorithms/dyadic-to-rational.h is in liblinbox-dev 1.4.2-3.

This file is owned by root:root, with mode 0o644.

The actual contents of the file can be viewed below.

  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
#ifndef __LINBOX_algorithms_dyadic_to_rational_H
#define __LINBOX_algorithms_dyadic_to_rational_H

/* dyadic-to-rational.h
 *
 * dyadicToRational reconstructs a rational a/b from dyadic approximation n/2^k.
 *
 * It is used in rational-solver-sn
 *
 * "Rational reconstruction" starts from a p-adic approximation.
 * This is different though very similar
 *
 *  Evolved by bds from rational-reconstruction2.h by Z. Wan.
 *
 * Copyright (c) LinBox
 *
 * ========LICENCE========
 * This file is part of the library LinBox.
 *
 * LinBox is free software: you can redistribute it and/or modify
 * it under the terms of the  GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 * ========LICENCE========
 */

#include <stack>
#include <assert.h>

//#include "linbox/integer.h"

namespace LinBox{
template<class Ring>
bool partial_hegcd(Ring& Z, typename Ring::Element& e, typename Ring::Element& b, const typename Ring::Element& n, const typename Ring::Element& d, const typename Ring::Element& denBound);
/** Rational reconstruction of a/b from n/d with denominator bound B.
 * We give a/b, the continued fraction approximant of n/d that
 * satisfies |a/b - n/d| < 1/2d (well approximated) and 0 < b <= B.
 * Return value is 0, if no such approximant exists.
 * Return value is 1, if either
 *   (i) a second well approximated rational with denominator bounded by B may exist, or
 *   (ii) the well approximated condition is not met for a/b.
 *   In these cases, a/b may be used speculatively.
 * Return value is 2, if the approximant is guaranteed (because bB <= d).

 * If no fraction is well approximated the last b <= B in the remainder sequence of n,d is given.
 *
 * If d = 2^k and n = sum_i=l to k n_i 2^i, then * n/d = sum_{i=l down to 0} n_i/2^{k-i}
 * is a {\em dyadic rational}.  Numbers of this form are produced for example by
 * numeric-symbolic iterations.
 *
 * If it is known that n/d is the most accurate approximation with denominator d
 * to a/b, and that the denominator b is bounded by B, i.e. b <= B, then such a/b is
 * uniquely determined, provided d >= bB.
 * ...in that case, such a/b is returned by dyadicToRational().
 * This follows from two facts:
 * First, by definition, n/d is an accurate approximation to a/b
 * with b <= d when |n/d - a/b| < 1/2d.
 * Otherwise (n-1)/d or (n+1)/d would be a better approximation.
 * Second, if a/b and a'/b' are distinct rationals, then |a/b - a'/b'| >= 1/bb'.
 * Thus if a'/b' is another rational accurately approximated by n/d,
 * we have 1/bb' <= |a/b - a'/b'| <= |a/b - n/d| + |n/d - a'/b'| <= 1/2d + 1/2d = 1/d.
 * So bb' > d >= bB, thus b' > B.
 *
 * In summary: If it exists, the unique a/b is given such that n/d approximates a/b
 * to within 1/2d and b <= B.  Otherwise a plausible a/b is given or failure is signaled.
 *
 * "Symbolic-Numeric Exact Rational Linear System Solver" by Saunders, Wood, Youse.
 * describes the construction.
 */
template<class Ring>
int dyadicToRational (
	const Ring& Z,
	typename Ring::Element& a, typename Ring::Element& b,
	const typename Ring::Element& n, const typename Ring::Element& d,
	const typename Ring::Element& B)
{
	typedef typename Ring::Element Int;
	Int e; Z.init(e);// error term

	Int an; Z.init(an); Z.abs(an, n/*q*/);

	// Partial_hegcd is defined below.
	bool found = partial_hegcd(Z, e, b, an, d/*q*/, B); // e = b*an - ?*d and |b| <= B
	Z.axmyin(e, b, an);
	Z.div(a, e, d); //a = (e - b*an)/d, div is exact.
//std::cout << "Z.axmyin(e, b, an); " << e << " " << b << " " << an << ", a = e/d exact " << a << " " << d << std::endl;
	// now a/b is solution but signs may be wrong
	Z.abs(a,a);
	Z.abs(b,b);
	if (Z.compare(n, Z.zero) < 0)  Z.negin(a); // a = -a;

//std::cout << "DtR in n, d " << n << " "<< d << ", bound " << B << ", out a, b " << a << " " << b << std::endl;
	bool guarantee = b*B < d;
	if (found && guarantee) return 2;
	if (b == 0) return 0;
	return 1; //if ((!found && b > 0) || (found && ! guarantee)) return 1;
}

/** partial_hegcd() sets e, b from the remainder sequence of n,d.
 * It requires positive n and d.
 * It sets e to the first r_i (remainder) and
 * b to the corresponding q_i (coefficient of n)
 * such that 2r_i < |q_i| and |q_i| <= B (the given denominator bound).
 * True is returned iff such e, b exist.
 *
 * If not, b is the largest q_i such that |q_i| <= B,
 * and e is the corresponding remainder.  In this case b is the denominator
 * of a plausibly approximated but not well approximated rational.
 * It can be used speculatively.
 */
// later reintroduce swapping trick
template<class Ring>
bool partial_hegcd(Ring& Z, typename Ring::Element& e, typename Ring::Element& b, const typename Ring::Element& n, const typename Ring::Element& d, const typename Ring::Element& denBound){
	typedef typename Ring::Element Int;
	Int quo, r, tmp;  Z.init(quo); Z.init(r); Z.init(tmp);
	bool withinbound, wellapproximated;

	Int b0; Z.init(b0); Z.assign(b0, Z.one); // and a0 = -0
	Int r0; Z.init(r0); Z.assign(r0, n); // so that r0 = b0*n - a0*d
	Int b1; Z.init(b1); Z.assign(b1, Z.zero); // and a1 = 1
	Int r1; Z.init(r1); Z.assign(r1, d); // so that r1 = b1*n - a1*d
//std::cout << "init 1 -0: " << b0 << " " << b1 << std::endl;

	do {
//std::cout << "quorem from " << r0 << " " << b0 << " and " << r1 << " " << b1 << std::endl;
		Z.quoRem(quo, e, r0, r1);
		//integer::divmod (quo, e, r0, r1);
		b = b0;
		Z.axmyin(b, quo, b1);
		Z.negin(b); // b = b0 - quo*b1;
//std::cout << "Z.axmyin(b, quo, b1);// b = b0 - quo*b1;" << b << " " << b0 << " " << quo << " " << b1 << std::endl;
		r0 = r1; b0 = b1;
		r1 = e; b1 = b;
		//assert(r1 >= 0);
		Z.abs(tmp, b);
		withinbound = (Z.compare(tmp, denBound) <= 0);
	    wellapproximated = (Z.compare(2*e , tmp) <= 0);
	}
	while ( ! wellapproximated && withinbound );
	if (! withinbound) {e = r0; b = b0;} // make available for speculation
//std::cout << "withinbound " << withinbound << " e b " << e << " " << b << ", n/d " << n << "/" << d << ", denBound " << denBound << std::endl;
	return withinbound;
	// returning with first well approximated (small remainder e) or last within bound denom b.

} // partial_hegcd

// vector rational reconstruction building num, den from numx, denx
template<class Ring>
int dyadicToRational(
	const Ring& Z,
	BlasVector<Ring>& num, typename Ring::Element& den,
	BlasVector<Ring>& numx, typename Ring::Element& denx,
	typename Ring::Element& denBound)
{
	typedef typename Ring::Element Int;
	Int q, rem, tmp_den, nx;
	Z.init(q); Z.init(rem); Z.init(tmp_den); Z.init(nx);
	Int two; Z.init(two, int64_t(2));
	Int denx2; Z.init(denx2);
	Z.assign(denx2, denx); Z.divin(denx2, two);// denx2 = denx/2, for balancing remainders.
	std::stack<std::pair<size_t, Int> > S;
	Int tmp; Z.init(tmp);

	Int den_lcm; Z.init(den_lcm); Z.assign(den_lcm, Z.one);
	den = den_lcm; // = 1.

	S.push(std::pair<int, Int>(0, Z.one));
	Int e; Z.init(e);// e for error
	int ret = 2; // 2 means guaranteed, 1 possible, 0 fail.
	for (size_t i = 0; i < num.size(); ++i) {
		Z.abs(nx, numx[i]);
		Z.mul(tmp, nx, den);
		Z.quoRem(num[i], e, tmp, denx); //nx*den - num[i]*denx = e, with num[i] and e nonneg.
		// we need |nx/denx - num[i]/den| == e/den*denx <= 1/2denx, so 2e <= den.
		// adjust to balanced remainder e.
		if (Z.compare(e, denx2) >= 0) {Z.subin(e, denx), Z.addin(num[i], Z.one); }
		//nx*den = num[i]*denx + e , thus |nx/denx - num[i]/den| = e/denx*den

	// can try e < den && 2*e < den for speed
		Z.mul(tmp, two, Z.abs(e));
		if ( Z.compare(tmp, den) > 0)// 2|e| > den
		{   // we failed, so need another reconstruction
			int oneret = dyadicToRational (Z, tmp, tmp_den, nx, denx, denBound);
			if (oneret == 1) ret = 1;
			if ( oneret == 0 ) return oneret;
			//std::cerr << i << " tmp " << tmp << " num[i] " << num[i] << std::endl;
			num[i] = tmp;

			Z.lcm (den_lcm, tmp_den, den);
			Z.div( tmp, den_lcm, tmp_den ); // exact
			Z.mulin( num[i], tmp ); // num[i]/den_lcm = prev num[i]/tmp_den.

			Z.div(tmp, den_lcm, den); // exact
			// must multiply all prior num[i] by this tmp.
			S.push(std::pair<size_t, Int>(i, tmp));
			den = den_lcm;
			//assert(Z.compare(den, denBound)<=0);
			if (Z.compare(den, denBound)>0) return false; // den > denBound
		}

		if (Z.compare(numx[i], Z.zero) < 0) Z.negin(num[i]); // numx[i] < 0

	}
	// now fix shorties
	Int t; Z.assign(t, Z.one);
	while ( S.size() > 1 ) {
		Z.mulin(t, S.top().second);
		int k = (int)S.top().first;
		S.pop();
		int j = (int)S.top().first;
		for (int i = k-1; i >= j; --i) {
			Z.mulin(num[(size_t)i], t);
		}
	}
	S.pop();
	return ret;

} // vector dyadicToRational

#if 0
// vector rational reconstruction building num, den from numx, denx
// This one -- very inefficient -- just reconstructs each one, then goes thru to fix for lcm.
void rational_reconstruction(BlasVector<Givaro::ZRing<Integer> >& num, integer& den, BlasVector<Givaro::ZRing<Integer> >& numx, integer& denx, integer& denBound) {
	integer den_tmp, missing_factor;
	den = 1;
	for (size_t i = 0; i < numx.size(); ++i) {
		rational_reconstruction(num[i], den_tmp, numx[i], denx, denBound);
		lcm(missing_factor, den_tmp, den);
		den = missing_factor;
	}
	for (size_t i = 0; i < numx.size(); ++i) {
		rational_reconstruction(num[i], den_tmp, numx[i], denx, denBound);
		integer::divexact (missing_factor, den, den_tmp);
		num[i] *= missing_factor;
	}
} // vector rational_reconstruction
****
#endif

}// LinBox
#endif // __LINBOX_algorithms_dyadic_to_rational_H



// Local Variables:
// mode: C++
// tab-width: 8
// indent-tabs-mode: nil
// c-basic-offset: 8
// End:
// vim:sts=8:sw=8:ts=8:noet:sr:cino=>s,f0,{0,g0,(0,:0,t0,+0,=s