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
|
(**
This file is part of the Flocq formalization of floating-point
arithmetic in Coq: http://flocq.gforge.inria.fr/
Copyright (C) 2010-2011 Sylvie Boldo
#<br />#
Copyright (C) 2010-2011 Guillaume Melquiond
This library 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 3 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
COPYING file for more details.
*)
(** * Helper function and theorem for computing the rounded quotient of two floating-point numbers. *)
Require Import Fcore_Raux.
Require Import Fcore_defs.
Require Import Fcore_float_prop.
Require Import Fcore_digits.
Require Import Fcalc_bracket.
Require Import Fcalc_digits.
Section Fcalc_div.
Variable beta : radix.
Notation bpow e := (bpow beta e).
(** Computes a mantissa of precision p, the corresponding exponent,
and the position with respect to the real quotient of the
input floating-point numbers.
The algorithm performs the following steps:
- Shift dividend mantissa so that it has at least p2 + p digits.
- Perform the Euclidean division.
- Compute the position according to the division remainder.
Complexity is fine as long as p1 <= 2p and p2 <= p.
*)
Definition Fdiv_core prec m1 e1 m2 e2 :=
let d1 := Zdigits beta m1 in
let d2 := Zdigits beta m2 in
let e := (e1 - e2)%Z in
let (m, e') :=
match (d2 + prec - d1)%Z with
| Zpos p => (m1 * Zpower_pos beta p, e + Zneg p)%Z
| _ => (m1, e)
end in
let '(q, r) := Zdiv_eucl m m2 in
(q, e', new_location m2 r loc_Exact).
Theorem Fdiv_core_correct :
forall prec m1 e1 m2 e2,
(0 < prec)%Z ->
(0 < m1)%Z -> (0 < m2)%Z ->
let '(m, e, l) := Fdiv_core prec m1 e1 m2 e2 in
(prec <= Zdigits beta m)%Z /\
inbetween_float beta m e (F2R (Float beta m1 e1) / F2R (Float beta m2 e2)) l.
Proof.
intros prec m1 e1 m2 e2 Hprec Hm1 Hm2.
unfold Fdiv_core.
set (d1 := Zdigits beta m1).
set (d2 := Zdigits beta m2).
case_eq
(match (d2 + prec - d1)%Z with
| Zpos p => ((m1 * Zpower_pos beta p)%Z, (e1 - e2 + Zneg p)%Z)
| _ => (m1, (e1 - e2)%Z)
end).
intros m' e' Hme.
(* . the shifted mantissa m' has enough digits *)
assert (Hs: F2R (Float beta m' (e' + e2)) = F2R (Float beta m1 e1) /\ (0 < m')%Z /\ (d2 + prec <= Zdigits beta m')%Z).
replace (d2 + prec)%Z with (d2 + prec - d1 + d1)%Z by ring.
destruct (d2 + prec - d1)%Z as [|p|p] ;
unfold d1 ;
inversion Hme.
ring_simplify (e1 - e2 + e2)%Z.
repeat split.
now rewrite <- H0.
apply Zle_refl.
replace (e1 - e2 + Zneg p + e2)%Z with (e1 - Zpos p)%Z by (unfold Zminus ; simpl ; ring).
fold (Zpower beta (Zpos p)).
split.
pattern (Zpos p) at 1 ; replace (Zpos p) with (e1 - (e1 - Zpos p))%Z by ring.
apply sym_eq.
apply F2R_change_exp.
assert (0 < Zpos p)%Z by easy.
omega.
split.
apply Zmult_lt_0_compat.
exact Hm1.
now apply Zpower_gt_0.
rewrite Zdigits_mult_Zpower.
rewrite Zplus_comm.
apply Zle_refl.
apply sym_not_eq.
now apply Zlt_not_eq.
easy.
split.
now ring_simplify (e1 - e2 + e2)%Z.
assert (Zneg p < 0)%Z by easy.
omega.
(* . *)
destruct Hs as (Hs1, (Hs2, Hs3)).
rewrite <- Hs1.
generalize (Z_div_mod m' m2 (Zlt_gt _ _ Hm2)).
destruct (Zdiv_eucl m' m2) as (q, r).
intros (Hq, Hr).
split.
(* . the result mantissa q has enough digits *)
cut (Zdigits beta m' <= d2 + Zdigits beta q)%Z. omega.
unfold d2.
rewrite Hq.
assert (Hq': (0 < q)%Z).
apply Zmult_lt_reg_r with (1 := Hm2).
assert (m2 < m')%Z.
apply lt_Zdigits with beta.
now apply Zlt_le_weak.
unfold d2 in Hs3.
clear -Hprec Hs3 ; omega.
cut (q * m2 = m' - r)%Z. clear -Hr H ; omega.
rewrite Hq.
ring.
apply Zle_trans with (Zdigits beta (m2 + q + m2 * q)).
apply Zdigits_le.
rewrite <- Hq.
now apply Zlt_le_weak.
clear -Hr Hq'. omega.
apply Zdigits_mult_strong ; apply Zlt_le_weak.
now apply Zle_lt_trans with r.
exact Hq'.
(* . the location is correctly computed *)
unfold inbetween_float, F2R. simpl.
rewrite bpow_plus, Z2R_plus.
rewrite Hq, Z2R_plus, Z2R_mult.
replace ((Z2R m2 * Z2R q + Z2R r) * (bpow e' * bpow e2) / (Z2R m2 * bpow e2))%R
with ((Z2R q + Z2R r / Z2R m2) * bpow e')%R.
apply inbetween_mult_compat.
apply bpow_gt_0.
destruct (Z_lt_le_dec 1 m2) as [Hm2''|Hm2''].
replace (Z2R 1) with (Z2R m2 * /Z2R m2)%R.
apply new_location_correct ; try easy.
apply Rinv_0_lt_compat.
now apply (Z2R_lt 0).
now constructor.
apply Rinv_r.
apply Rgt_not_eq.
now apply (Z2R_lt 0).
assert (r = 0 /\ m2 = 1)%Z by (clear -Hr Hm2'' ; omega).
rewrite (proj1 H), (proj2 H).
unfold Rdiv.
rewrite Rmult_0_l, Rplus_0_r.
now constructor.
field.
split ; apply Rgt_not_eq.
apply bpow_gt_0.
now apply (Z2R_lt 0).
Qed.
End Fcalc_div.
|
