Theory Approximation_Bounds
theory Approximation_Bounds
imports
Complex_Main
"HOL-Library.Interval_Float"
Dense_Linear_Order
begin
declare powr_neg_one [simp]
declare powr_neg_numeral [simp]
context includes interval.lifting begin
section "Horner Scheme"
subsection ‹Define auxiliary helper ‹horner› function›
primrec horner :: "(nat ⇒ nat) ⇒ (nat ⇒ nat ⇒ nat) ⇒ nat ⇒ nat ⇒ nat ⇒ real ⇒ real" where
"horner F G 0 i k x = 0" |
"horner F G (Suc n) i k x = 1 / k - x * horner F G n (F i) (G i k) x"
lemma horner_schema':
fixes x :: real and a :: "nat ⇒ real"
shows "a 0 - x * (∑ i=0..<n. (-1)^i * a (Suc i) * x^i) = (∑ i=0..<Suc n. (-1)^i * a i * x^i)"
proof -
have shift_pow: "⋀i. - (x * ((-1)^i * a (Suc i) * x ^ i)) = (-1)^(Suc i) * a (Suc i) * x ^ (Suc i)"
by auto
show ?thesis
unfolding sum_distrib_left shift_pow uminus_add_conv_diff [symmetric] sum_negf[symmetric]
sum.atLeast_Suc_lessThan[OF zero_less_Suc]
sum.reindex[OF inj_Suc, unfolded comp_def, symmetric, of "λ n. (-1)^n *a n * x^n"] by auto
qed
lemma horner_schema:
fixes f :: "nat ⇒ nat" and G :: "nat ⇒ nat ⇒ nat" and F :: "nat ⇒ nat"
assumes f_Suc: "⋀n. f (Suc n) = G ((F ^^ n) s) (f n)"
shows "horner F G n ((F ^^ j') s) (f j') x = (∑ j = 0..< n. (- 1) ^ j * (1 / (f (j' + j))) * x ^ j)"
proof (induct n arbitrary: j')
case 0
then show ?case by auto
next
case (Suc n)
show ?case unfolding horner.simps Suc[where j'="Suc j'", unfolded funpow.simps comp_def f_Suc]
using horner_schema'[of "λ j. 1 / (f (j' + j))"] by auto
qed
lemma horner_bounds':
fixes lb :: "nat ⇒ nat ⇒ nat ⇒ float ⇒ float" and ub :: "nat ⇒ nat ⇒ nat ⇒ float ⇒ float"
assumes "0 ≤ real_of_float x" and f_Suc: "⋀n. f (Suc n) = G ((F ^^ n) s) (f n)"
and lb_0: "⋀ i k x. lb 0 i k x = 0"
and lb_Suc: "⋀ n i k x. lb (Suc n) i k x = float_plus_down prec
(lapprox_rat prec 1 k)
(- float_round_up prec (x * (ub n (F i) (G i k) x)))"
and ub_0: "⋀ i k x. ub 0 i k x = 0"
and ub_Suc: "⋀ n i k x. ub (Suc n) i k x = float_plus_up prec
(rapprox_rat prec 1 k)
(- float_round_down prec (x * (lb n (F i) (G i k) x)))"
shows "(lb n ((F ^^ j') s) (f j') x) ≤ horner F G n ((F ^^ j') s) (f j') x ∧
horner F G n ((F ^^ j') s) (f j') x ≤ (ub n ((F ^^ j') s) (f j') x)"
(is "?lb n j' ≤ ?horner n j' ∧ ?horner n j' ≤ ?ub n j'")
proof (induct n arbitrary: j')
case 0
thus ?case unfolding lb_0 ub_0 horner.simps by auto
next
case (Suc n)
thus ?case using lapprox_rat[of prec 1 "f j'"] using rapprox_rat[of 1 "f j'" prec]
Suc[where j'="Suc j'"] ‹0 ≤ real_of_float x›
by (auto intro!: add_mono mult_left_mono float_round_down_le float_round_up_le
order_trans[OF add_mono[OF _ float_plus_down_le]]
order_trans[OF _ add_mono[OF _ float_plus_up_le]]
simp add: lb_Suc ub_Suc field_simps f_Suc)
qed
subsection "Theorems for floating point functions implementing the horner scheme"
text ‹
Here \<^term_type>‹f :: nat ⇒ nat› is the sequence defining the Taylor series, the coefficients are
all alternating and reciprocs. We use \<^term>‹G› and \<^term>‹F› to describe the computation of \<^term>‹f›.
›
lemma horner_bounds:
fixes F :: "nat ⇒ nat" and G :: "nat ⇒ nat ⇒ nat"
assumes "0 ≤ real_of_float x" and f_Suc: "⋀n. f (Suc n) = G ((F ^^ n) s) (f n)"
and lb_0: "⋀ i k x. lb 0 i k x = 0"
and lb_Suc: "⋀ n i k x. lb (Suc n) i k x = float_plus_down prec
(lapprox_rat prec 1 k)
(- float_round_up prec (x * (ub n (F i) (G i k) x)))"
and ub_0: "⋀ i k x. ub 0 i k x = 0"
and ub_Suc: "⋀ n i k x. ub (Suc n) i k x = float_plus_up prec
(rapprox_rat prec 1 k)
(- float_round_down prec (x * (lb n (F i) (G i k) x)))"
shows "(lb n ((F ^^ j') s) (f j') x) ≤ (∑j=0..<n. (- 1) ^ j * (1 / (f (j' + j))) * (x ^ j))"
(is "?lb")
and "(∑j=0..<n. (- 1) ^ j * (1 / (f (j' + j))) * (x ^ j)) ≤ (ub n ((F ^^ j') s) (f j') x)"
(is "?ub")
proof -
have "?lb ∧ ?ub"
using horner_bounds'[where lb=lb, OF ‹0 ≤ real_of_float x› f_Suc lb_0 lb_Suc ub_0 ub_Suc]
unfolding horner_schema[where f=f, OF f_Suc] by simp
thus "?lb" and "?ub" by auto
qed
lemma horner_bounds_nonpos:
fixes F :: "nat ⇒ nat" and G :: "nat ⇒ nat ⇒ nat"
assumes "real_of_float x ≤ 0" and f_Suc: "⋀n. f (Suc n) = G ((F ^^ n) s) (f n)"
and lb_0: "⋀ i k x. lb 0 i k x = 0"
and lb_Suc: "⋀ n i k x. lb (Suc n) i k x = float_plus_down prec
(lapprox_rat prec 1 k)
(float_round_down prec (x * (ub n (F i) (G i k) x)))"
and ub_0: "⋀ i k x. ub 0 i k x = 0"
and ub_Suc: "⋀ n i k x. ub (Suc n) i k x = float_plus_up prec
(rapprox_rat prec 1 k)
(float_round_up prec (x * (lb n (F i) (G i k) x)))"
shows "(lb n ((F ^^ j') s) (f j') x) ≤ (∑j=0..<n. (1 / (f (j' + j))) * real_of_float x ^ j)" (is "?lb")
and "(∑j=0..<n. (1 / (f (j' + j))) * real_of_float x ^ j) ≤ (ub n ((F ^^ j') s) (f j') x)" (is "?ub")
proof -
have diff_mult_minus: "x - y * z = x + - y * z" for x y z :: float by simp
have sum_eq: "(∑j=0..<n. (1 / (f (j' + j))) * real_of_float x ^ j) =
(∑j = 0..<n. (- 1) ^ j * (1 / (f (j' + j))) * real_of_float (- x) ^ j)"
by (auto simp add: field_simps power_mult_distrib[symmetric])
have "0 ≤ real_of_float (-x)" using assms by auto
from horner_bounds[where G=G and F=F and f=f and s=s and prec=prec
and lb="λ n i k x. lb n i k (-x)" and ub="λ n i k x. ub n i k (-x)",
unfolded lb_Suc ub_Suc diff_mult_minus,
OF this f_Suc lb_0 _ ub_0 _]
show "?lb" and "?ub" unfolding minus_minus sum_eq
by (auto simp: minus_float_round_up_eq minus_float_round_down_eq)
qed
subsection ‹Selectors for next even or odd number›
text ‹
The horner scheme computes alternating series. To get the upper and lower bounds we need to
guarantee to access a even or odd member. To do this we use \<^term>‹get_odd› and \<^term>‹get_even›.
›
definition get_odd :: "nat ⇒ nat" where
"get_odd n = (if odd n then n else (Suc n))"
definition get_even :: "nat ⇒ nat" where
"get_even n = (if even n then n else (Suc n))"
lemma get_odd[simp]: "odd (get_odd n)"
unfolding get_odd_def by (cases "odd n") auto
lemma get_even[simp]: "even (get_even n)"
unfolding get_even_def by (cases "even n") auto
lemma get_odd_ex: "∃ k. Suc k = get_odd n ∧ odd (Suc k)"
by (auto simp: get_odd_def odd_pos intro!: exI[of _ "n - 1"])
lemma get_even_double: "∃i. get_even n = 2 * i"
using get_even by (blast elim: evenE)
lemma get_odd_double: "∃i. get_odd n = 2 * i + 1"
using get_odd by (blast elim: oddE)
section "Power function"
definition float_power_bnds :: "nat ⇒ nat ⇒ float ⇒ float ⇒ float * float" where
"float_power_bnds prec n l u =
(if 0 < l then (power_down_fl prec l n, power_up_fl prec u n)
else if odd n then
(- power_up_fl prec ¦l¦ n,
if u < 0 then - power_down_fl prec ¦u¦ n else power_up_fl prec u n)
else if u < 0 then (power_down_fl prec ¦u¦ n, power_up_fl prec ¦l¦ n)
else (0, power_up_fl prec (max ¦l¦ ¦u¦) n))"
lemma le_minus_power_downI: "0 ≤ x ⟹ x ^ n ≤ - a ⟹ a ≤ - power_down prec x n"
by (subst le_minus_iff) (auto intro: power_down_le power_mono_odd)
lemma float_power_bnds:
"(l1, u1) = float_power_bnds prec n l u ⟹ x ∈ {l .. u} ⟹ (x::real) ^ n ∈ {l1..u1}"
by (auto
simp: float_power_bnds_def max_def real_power_up_fl real_power_down_fl minus_le_iff
split: if_split_asm
intro!: power_up_le power_down_le le_minus_power_downI
intro: power_mono_odd power_mono power_mono_even zero_le_even_power)
lemma bnds_power:
"∀(x::real) l u. (l1, u1) = float_power_bnds prec n l u ∧ x ∈ {l .. u} ⟶
l1 ≤ x ^ n ∧ x ^ n ≤ u1"
using float_power_bnds by auto
lift_definition power_float_interval :: "nat ⇒ nat ⇒ float interval ⇒ float interval"
is "λp n (l, u). float_power_bnds p n l u"
using float_power_bnds
by (auto simp: bnds_power dest!: float_power_bnds[OF sym])
lemma lower_power_float_interval:
"lower (power_float_interval p n x) = fst (float_power_bnds p n (lower x) (upper x))"
by transfer auto
lemma upper_power_float_interval:
"upper (power_float_interval p n x) = snd (float_power_bnds p n (lower x) (upper x))"
by transfer auto
lemma power_float_intervalI: "x ∈⇩r X ⟹ x ^ n ∈⇩r power_float_interval p n X"
using float_power_bnds[OF prod.collapse]
by (auto simp: set_of_eq lower_power_float_interval upper_power_float_interval)
lemma power_float_interval_mono:
"set_of (power_float_interval prec n A)
⊆ set_of (power_float_interval prec n B)"
if "set_of A ⊆ set_of B"
proof -
define la where "la = real_of_float (lower A)"
define ua where "ua = real_of_float (upper A)"
define lb where "lb = real_of_float (lower B)"
define ub where "ub = real_of_float (upper B)"
have ineqs: "lb ≤ la" "la ≤ ua" "ua ≤ ub" "lb ≤ ub"
using that lower_le_upper[of A] lower_le_upper[of B]
by (auto simp: la_def ua_def lb_def ub_def set_of_eq)
show ?thesis
using ineqs
by (simp add: set_of_subset_iff float_power_bnds_def max_def
power_down_fl.rep_eq power_up_fl.rep_eq
lower_power_float_interval upper_power_float_interval
la_def[symmetric] ua_def[symmetric] lb_def[symmetric] ub_def[symmetric])
(auto intro!: power_down_mono power_up_mono intro: order_trans[where y=0])
qed
section ‹Approximation utility functions›
lift_definition plus_float_interval::"nat ⇒ float interval ⇒ float interval ⇒ float interval"
is "λprec. λ(a1, a2). λ(b1, b2). (float_plus_down prec a1 b1, float_plus_up prec a2 b2)"
by (auto intro!: add_mono simp: float_plus_down_le float_plus_up_le)
lemma lower_plus_float_interval:
"lower (plus_float_interval prec ivl ivl') = float_plus_down prec (lower ivl) (lower ivl')"
by transfer auto
lemma upper_plus_float_interval:
"upper (plus_float_interval prec ivl ivl') = float_plus_up prec (upper ivl) (upper ivl')"
by transfer auto
lemma mult_float_interval_ge:
"real_interval A + real_interval B ≤ real_interval (plus_float_interval prec A B)"
unfolding less_eq_interval_def
by transfer
(auto simp: lower_plus_float_interval upper_plus_float_interval
intro!: order.trans[OF float_plus_down] order.trans[OF _ float_plus_up])
lemma plus_float_interval:
"set_of (real_interval A) + set_of (real_interval B) ⊆
set_of (real_interval (plus_float_interval prec A B))"
proof -
have "set_of (real_interval A) + set_of (real_interval B) ⊆
set_of (real_interval A + real_interval B)"
by (simp add: set_of_plus)
also have "… ⊆ set_of (real_interval (plus_float_interval prec A B))"
using mult_float_interval_ge[of A B prec] by (simp add: set_of_subset_iff')
finally show ?thesis .
qed
lemma plus_float_intervalI:
"x + y ∈⇩r plus_float_interval prec A B"
if "x ∈⇩i real_interval A" "y ∈⇩i real_interval B"
using plus_float_interval[of A B] that by auto
lemma plus_float_interval_mono:
"plus_float_interval prec A B ≤ plus_float_interval prec X Y"
if "A ≤ X" "B ≤ Y"
using that
by (auto simp: less_eq_interval_def lower_plus_float_interval upper_plus_float_interval
float_plus_down.rep_eq float_plus_up.rep_eq plus_down_mono plus_up_mono)
lemma plus_float_interval_monotonic:
"set_of (ivl + ivl') ⊆ set_of (plus_float_interval prec ivl ivl')"
using float_plus_down_le float_plus_up_le lower_plus_float_interval upper_plus_float_interval
by (simp add: set_of_subset_iff)
definition bnds_mult :: "nat ⇒ float ⇒ float ⇒ float ⇒ float ⇒ float × float" where
"bnds_mult prec a1 a2 b1 b2 =
(float_plus_down prec (nprt a1 * pprt b2)
(float_plus_down prec (nprt a2 * nprt b2)
(float_plus_down prec (pprt a1 * pprt b1) (pprt a2 * nprt b1))),
float_plus_up prec (pprt a2 * pprt b2)
(float_plus_up prec (pprt a1 * nprt b2)
(float_plus_up prec (nprt a2 * pprt b1) (nprt a1 * nprt b1))))"
lemma bnds_mult:
fixes prec :: nat and a1 aa2 b1 b2 :: float
assumes "(l, u) = bnds_mult prec a1 a2 b1 b2"
assumes "a ∈ {real_of_float a1..real_of_float a2}"
assumes "b ∈ {real_of_float b1..real_of_float b2}"
shows "a * b ∈ {real_of_float l..real_of_float u}"
proof -
from assms have "real_of_float l ≤ a * b"
by (intro order.trans[OF _ mult_ge_prts[of a1 a a2 b1 b b2]])
(auto simp: bnds_mult_def intro!: float_plus_down_le)
moreover from assms have "real_of_float u ≥ a * b"
by (intro order.trans[OF mult_le_prts[of a1 a a2 b1 b b2]])
(auto simp: bnds_mult_def intro!: float_plus_up_le)
ultimately show ?thesis by simp
qed
lift_definition mult_float_interval::"nat ⇒ float interval ⇒ float interval ⇒ float interval"
is "λprec. λ(a1, a2). λ(b1, b2). bnds_mult prec a1 a2 b1 b2"
by (auto dest!: bnds_mult[OF sym])
lemma lower_mult_float_interval:
"lower (mult_float_interval p x y) = fst (bnds_mult p (lower x) (upper x) (lower y) (upper y))"
by transfer auto
lemma upper_mult_float_interval:
"upper (mult_float_interval p x y) = snd (bnds_mult p (lower x) (upper x) (lower y) (upper y))"
by transfer auto
lemma mult_float_interval:
"set_of (real_interval A) * set_of (real_interval B) ⊆
set_of (real_interval (mult_float_interval prec A B))"
proof -
let ?bm = "bnds_mult prec (lower A) (upper A) (lower B) (upper B)"
show ?thesis
using bnds_mult[of "fst ?bm" "snd ?bm", simplified, OF refl]
by (auto simp: set_of_eq set_times_def upper_mult_float_interval lower_mult_float_interval)
qed
lemma mult_float_intervalI:
"x * y ∈⇩r mult_float_interval prec A B"
if "x ∈⇩i real_interval A" "y ∈⇩i real_interval B"
using mult_float_interval[of A B] that
by auto
lemma mult_float_interval_mono':
"set_of (mult_float_interval prec A B) ⊆ set_of (mult_float_interval prec X Y)"
if "set_of A ⊆ set_of X" "set_of B ⊆ set_of Y"
using that
apply transfer
unfolding bnds_mult_def atLeastatMost_subset_iff float_plus_down.rep_eq float_plus_up.rep_eq
by (auto simp: float_plus_down.rep_eq float_plus_up.rep_eq mult_float_mono1 mult_float_mono2)
lemma mult_float_interval_mono:
"mult_float_interval prec A B ≤ mult_float_interval prec X Y"
if "A ≤ X" "B ≤ Y"
using mult_float_interval_mono'[of A X B Y prec] that
by (simp add: set_of_subset_iff')
definition map_bnds :: "(nat ⇒ float ⇒ float) ⇒ (nat ⇒ float ⇒ float) ⇒
nat ⇒ (float × float) ⇒ (float × float)" where
"map_bnds lb ub prec = (λ(l,u). (lb prec l, ub prec u))"
lemma map_bnds:
assumes "(lf, uf) = map_bnds lb ub prec (l, u)"
assumes "mono f"
assumes "x ∈ {real_of_float l..real_of_float u}"
assumes "real_of_float (lb prec l) ≤ f (real_of_float l)"
assumes "real_of_float (ub prec u) ≥ f (real_of_float u)"
shows "f x ∈ {real_of_float lf..real_of_float uf}"
proof -
from assms have "real_of_float lf = real_of_float (lb prec l)"
by (simp add: map_bnds_def)
also have "real_of_float (lb prec l) ≤ f (real_of_float l)" by fact
also from assms have "… ≤ f x"
by (intro monoD[OF ‹mono f›]) auto
finally have lf: "real_of_float lf ≤ f x" .
from assms have "f x ≤ f (real_of_float u)"
by (intro monoD[OF ‹mono f›]) auto
also have "… ≤ real_of_float (ub prec u)" by fact
also from assms have "… = real_of_float uf"
by (simp add: map_bnds_def)
finally have uf: "f x ≤ real_of_float uf" .
from lf uf show ?thesis by simp
qed
section "Square root"
text ‹
The square root computation is implemented as newton iteration. As first first step we use the
nearest power of two greater than the square root.
›
fun sqrt_iteration :: "nat ⇒ nat ⇒ float ⇒ float" where
"sqrt_iteration prec 0 x = Float 1 ((bitlen ¦mantissa x¦ + exponent x) div 2 + 1)" |
"sqrt_iteration prec (Suc m) x = (let y = sqrt_iteration prec m x
in Float 1 (- 1) * float_plus_up prec y (float_divr prec x y))"
lemma compute_sqrt_iteration_base[code]:
shows "sqrt_iteration prec n (Float m e) =
(if n = 0 then Float 1 ((if m = 0 then 0 else bitlen ¦m¦ + e) div 2 + 1)
else (let y = sqrt_iteration prec (n - 1) (Float m e) in
Float 1 (- 1) * float_plus_up prec y (float_divr prec (Float m e) y)))"
using bitlen_Float by (cases n) simp_all
function ub_sqrt lb_sqrt :: "nat ⇒ float ⇒ float" where
"ub_sqrt prec x = (if 0 < x then (sqrt_iteration prec prec x)
else if x < 0 then - lb_sqrt prec (- x)
else 0)" |
"lb_sqrt prec x = (if 0 < x then (float_divl prec x (sqrt_iteration prec prec x))
else if x < 0 then - ub_sqrt prec (- x)
else 0)"
by pat_completeness auto
termination by (relation "measure (λ v. let (prec, x) = case_sum id id v in (if x < 0 then 1 else 0))", auto)
declare lb_sqrt.simps[simp del]
declare ub_sqrt.simps[simp del]
lemma sqrt_ub_pos_pos_1:
assumes "sqrt x < b" and "0 < b" and "0 < x"
shows "sqrt x < (b + x / b)/2"
proof -
from assms have "0 < (b - sqrt x)⇧2 " by simp
also have "… = b⇧2 - 2 * b * sqrt x + (sqrt x)⇧2" by algebra
also have "… = b⇧2 - 2 * b * sqrt x + x" using assms by simp
finally have "0 < b⇧2 - 2 * b * sqrt x + x" .
hence "0 < b / 2 - sqrt x + x / (2 * b)" using assms
by (simp add: field_simps power2_eq_square)
thus ?thesis by (simp add: field_simps)
qed
lemma sqrt_iteration_bound:
assumes "0 < real_of_float x"
shows "sqrt x < sqrt_iteration prec n x"
proof (induct n)
case 0
show ?case
proof (cases x)
case (Float m e)
hence "0 < m"
using assms
by (auto simp: algebra_split_simps)
hence "0 < sqrt m" by auto
have int_nat_bl: "(nat (bitlen m)) = bitlen m"
using bitlen_nonneg by auto
have "x = (m / 2^nat (bitlen m)) * 2 powr (e + (nat (bitlen m)))"
unfolding Float by (auto simp: powr_realpow[symmetric] field_simps powr_add)
also have "… < 1 * 2 powr (e + nat (bitlen m))"
proof (rule mult_strict_right_mono, auto)
show "m < 2^nat (bitlen m)"
using bitlen_bounds[OF ‹0 < m›, THEN conjunct2]
unfolding of_int_less_iff[of m, symmetric] by auto
qed
finally have "sqrt x < sqrt (2 powr (e + bitlen m))"
unfolding int_nat_bl by auto
also have "… ≤ 2 powr ((e + bitlen m) div 2 + 1)"
proof -
let ?E = "e + bitlen m"
have E_mod_pow: "2 powr (?E mod 2) < 4"
proof (cases "?E mod 2 = 1")
case True
thus ?thesis by auto
next
case False
have "0 ≤ ?E mod 2" by auto
have "?E mod 2 < 2" by auto
from this[THEN zless_imp_add1_zle]
have "?E mod 2 ≤ 0" using False by auto
from xt1(5)[OF ‹0 ≤ ?E mod 2› this]
show ?thesis by auto
qed
hence "sqrt (2 powr (?E mod 2)) < sqrt (2 * 2)"
by (intro real_sqrt_less_mono) auto
hence E_mod_pow: "sqrt (2 powr (?E mod 2)) < 2" by auto
have E_eq: "2 powr ?E = 2 powr (?E div 2 + ?E div 2 + ?E mod 2)"
by auto
have "sqrt (2 powr ?E) = sqrt (2 powr (?E div 2) * 2 powr (?E div 2) * 2 powr (?E mod 2))"
unfolding E_eq unfolding powr_add[symmetric] by (metis of_int_add)
also have "… = 2 powr (?E div 2) * sqrt (2 powr (?E mod 2))"
unfolding real_sqrt_mult[of _ "2 powr (?E mod 2)"] real_sqrt_abs2 by auto
also have "… < 2 powr (?E div 2) * 2 powr 1"
by (rule mult_strict_left_mono) (auto intro: E_mod_pow)
also have "… = 2 powr (?E div 2 + 1)"
unfolding add.commute[of _ 1] powr_add[symmetric] by simp
finally show ?thesis by auto
qed
finally show ?thesis using ‹0 < m›
unfolding Float
by (subst compute_sqrt_iteration_base) (simp add: ac_simps)
qed
next
case (Suc n)
let ?b = "sqrt_iteration prec n x"
have "0 < sqrt x"
using ‹0 < real_of_float x› by auto
also have "… < real_of_float ?b"
using Suc .
finally have "sqrt x < (?b + x / ?b)/2"
using sqrt_ub_pos_pos_1[OF Suc _ ‹0 < real_of_float x›] by auto
also have "… ≤ (?b + (float_divr prec x ?b))/2"
by (rule divide_right_mono, auto simp add: float_divr)
also have "… = (Float 1 (- 1)) * (?b + (float_divr prec x ?b))"
by simp
also have "… ≤ (Float 1 (- 1)) * (float_plus_up prec ?b (float_divr prec x ?b))"
by (auto simp add: algebra_simps float_plus_up_le)
finally show ?case
unfolding sqrt_iteration.simps Let_def distrib_left .
qed
lemma sqrt_iteration_lower_bound:
assumes "0 < real_of_float x"
shows "0 < real_of_float (sqrt_iteration prec n x)" (is "0 < ?sqrt")
proof -
have "0 < sqrt x" using assms by auto
also have "… < ?sqrt" using sqrt_iteration_bound[OF assms] .
finally show ?thesis .
qed
lemma lb_sqrt_lower_bound:
assumes "0 ≤ real_of_float x"
shows "0 ≤ real_of_float (lb_sqrt prec x)"
proof (cases "0 < x")
case True
hence "0 < real_of_float x" and "0 ≤ x"
using ‹0 ≤ real_of_float x› by auto
hence "0 < sqrt_iteration prec prec x"
using sqrt_iteration_lower_bound by auto
hence "0 ≤ real_of_float (float_divl prec x (sqrt_iteration prec prec x))"
using float_divl_lower_bound[OF ‹0 ≤ x›] unfolding less_eq_float_def by auto
thus ?thesis
unfolding lb_sqrt.simps using True by auto
next
case False
with ‹0 ≤ real_of_float x› have "real_of_float x = 0" by auto
thus ?thesis
unfolding lb_sqrt.simps by auto
qed
lemma bnds_sqrt': "sqrt x ∈ {(lb_sqrt prec x) .. (ub_sqrt prec x)}"
proof -
have lb: "lb_sqrt prec x ≤ sqrt x" if "0 < x" for x :: float
proof -
from that have "0 < real_of_float x" and "0 ≤ real_of_float x" by auto
hence sqrt_gt0: "0 < sqrt x" by auto
hence sqrt_ub: "sqrt x < sqrt_iteration prec prec x"
using sqrt_iteration_bound by auto
have "(float_divl prec x (sqrt_iteration prec prec x)) ≤
x / (sqrt_iteration prec prec x)" by (rule float_divl)
also have "… < x / sqrt x"
by (rule divide_strict_left_mono[OF sqrt_ub ‹0 < real_of_float x›
mult_pos_pos[OF order_less_trans[OF sqrt_gt0 sqrt_ub] sqrt_gt0]])
also have "… = sqrt x"
unfolding inverse_eq_iff_eq[of _ "sqrt x", symmetric]
sqrt_divide_self_eq[OF ‹0 ≤ real_of_float x›, symmetric] by auto
finally show ?thesis
unfolding lb_sqrt.simps if_P[OF ‹0 < x›] by auto
qed
have ub: "sqrt x ≤ ub_sqrt prec x" if "0 < x" for x :: float
proof -
from that have "0 < real_of_float x" by auto
hence "0 < sqrt x" by auto
hence "sqrt x < sqrt_iteration prec prec x"
using sqrt_iteration_bound by auto
then show ?thesis
unfolding ub_sqrt.simps if_P[OF ‹0 < x›] by auto
qed
show ?thesis
using lb[of "-x"] ub[of "-x"] lb[of x] ub[of x]
by (auto simp add: lb_sqrt.simps ub_sqrt.simps real_sqrt_minus)
qed
lemma bnds_sqrt: "∀(x::real) lx ux.
(l, u) = (lb_sqrt prec lx, ub_sqrt prec ux) ∧ x ∈ {lx .. ux} ⟶ l ≤ sqrt x ∧ sqrt x ≤ u"
proof ((rule allI) +, rule impI, erule conjE, rule conjI)
fix x :: real
fix lx ux
assume "(l, u) = (lb_sqrt prec lx, ub_sqrt prec ux)"
and x: "x ∈ {lx .. ux}"
hence l: "l = lb_sqrt prec lx " and u: "u = ub_sqrt prec ux" by auto
have "sqrt lx ≤ sqrt x" using x by auto
from order_trans[OF _ this]
show "l ≤ sqrt x" unfolding l using bnds_sqrt'[of lx prec] by auto
have "sqrt x ≤ sqrt ux" using x by auto
from order_trans[OF this]
show "sqrt x ≤ u" unfolding u using bnds_sqrt'[of ux prec] by auto
qed
lift_definition sqrt_float_interval::"nat ⇒ float interval ⇒ float interval"
is "λprec. λ(lx, ux). (lb_sqrt prec lx, ub_sqrt prec ux)"
using bnds_sqrt'
by auto (meson order_trans real_sqrt_le_iff)
lemma lower_float_interval: "lower (sqrt_float_interval prec X) = lb_sqrt prec (lower X)"
by transfer auto
lemma upper_float_interval: "upper (sqrt_float_interval prec X) = ub_sqrt prec (upper X)"
by transfer auto
lemma sqrt_float_interval:
"sqrt ` set_of (real_interval X) ⊆ set_of (real_interval (sqrt_float_interval prec X))"
using bnds_sqrt
by (auto simp: set_of_eq lower_float_interval upper_float_interval)
lemma sqrt_float_intervalI: "sqrt x ∈⇩r sqrt_float_interval p X" if "x ∈⇩r X"
using sqrt_float_interval[of X p] that
by auto
section "Arcus tangens and π"
subsection "Compute arcus tangens series"
text ‹
As first step we implement the computation of the arcus tangens series. This is only valid in the range
\<^term>‹{-1 :: real .. 1}›. This is used to compute π and then the entire arcus tangens.
›
fun ub_arctan_horner :: "nat ⇒ nat ⇒ nat ⇒ float ⇒ float"
and lb_arctan_horner :: "nat ⇒ nat ⇒ nat ⇒ float ⇒ float" where
"ub_arctan_horner prec 0 k x = 0"
| "ub_arctan_horner prec (Suc n) k x = float_plus_up prec
(rapprox_rat prec 1 k) (- float_round_down prec (x * (lb_arctan_horner prec n (k + 2) x)))"
| "lb_arctan_horner prec 0 k x = 0"
| "lb_arctan_horner prec (Suc n) k x = float_plus_down prec
(lapprox_rat prec 1 k) (- float_round_up prec (x * (ub_arctan_horner prec n (k + 2) x)))"
lemma arctan_0_1_bounds':
assumes "0 ≤ real_of_float y" "real_of_float y ≤ 1"
and "even n"
shows "arctan (sqrt y) ∈
{(sqrt y * lb_arctan_horner prec n 1 y) .. (sqrt y * ub_arctan_horner prec (Suc n) 1 y)}"
proof -
let ?c = "λi. (- 1) ^ i * (1 / (i * 2 + (1::nat)) * sqrt y ^ (i * 2 + 1))"
let ?S = "λn. ∑ i=0..<n. ?c i"
have "0 ≤ sqrt y" using assms by auto
have "sqrt y ≤ 1" using assms by auto
from ‹even n› obtain m where "2 * m = n" by (blast elim: evenE)
have "arctan (sqrt y) ∈ { ?S n .. ?S (Suc n) }"
proof (cases "sqrt y = 0")
case True
then show ?thesis by simp
next
case False
hence "0 < sqrt y" using ‹0 ≤ sqrt y› by auto
hence prem: "0 < 1 / (0 * 2 + (1::nat)) * sqrt y ^ (0 * 2 + 1)" by auto
have "¦ sqrt y ¦ ≤ 1" using ‹0 ≤ sqrt y› ‹sqrt y ≤ 1› by auto
from mp[OF summable_Leibniz(2)[OF zeroseq_arctan_series[OF this]
monoseq_arctan_series[OF this]] prem, THEN spec, of m, unfolded ‹2 * m = n›]
show ?thesis unfolding arctan_series[OF ‹¦ sqrt y ¦ ≤ 1›] Suc_eq_plus1 atLeast0LessThan .
qed
note arctan_bounds = this[unfolded atLeastAtMost_iff]
have F: "⋀n. 2 * Suc n + 1 = 2 * n + 1 + 2" by auto
note bounds = horner_bounds[where s=1 and f="λi. 2 * i + 1" and j'=0
and lb="λn i k x. lb_arctan_horner prec n k x"
and ub="λn i k x. ub_arctan_horner prec n k x",
OF ‹0 ≤ real_of_float y› F lb_arctan_horner.simps ub_arctan_horner.simps]
have "(sqrt y * lb_arctan_horner prec n 1 y) ≤ arctan (sqrt y)"
proof -
have "(sqrt y * lb_arctan_horner prec n 1 y) ≤ ?S n"
using bounds(1) ‹0 ≤ sqrt y›
apply (simp only: power_add power_one_right mult.assoc[symmetric] sum_distrib_right[symmetric])
apply (simp only: mult.commute[where 'a=real] mult.commute[of _ "2::nat"] power_mult)
apply (auto intro!: mult_left_mono)
done
also have "… ≤ arctan (sqrt y)" using arctan_bounds ..
finally show ?thesis .
qed
moreover
have "arctan (sqrt y) ≤ (sqrt y * ub_arctan_horner prec (Suc n) 1 y)"
proof -
have "arctan (sqrt y) ≤ ?S (Suc n)" using arctan_bounds ..
also have "… ≤ (sqrt y * ub_arctan_horner prec (Suc n) 1 y)"
using bounds(2)[of "Suc n"] ‹0 ≤ sqrt y›
apply (simp only: power_add power_one_right mult.assoc[symmetric] sum_distrib_right[symmetric])
apply (simp only: mult.commute[where 'a=real] mult.commute[of _ "2::nat"] power_mult)
apply (auto intro!: mult_left_mono)
done
finally show ?thesis .
qed
ultimately show ?thesis by auto
qed
lemma arctan_0_1_bounds:
assumes "0 ≤ real_of_float y" "real_of_float y ≤ 1"
shows "arctan (sqrt y) ∈
{(sqrt y * lb_arctan_horner prec (get_even n) 1 y) ..
(sqrt y * ub_arctan_horner prec (get_odd n) 1 y)}"
using
arctan_0_1_bounds'[OF assms, of n prec]
arctan_0_1_bounds'[OF assms, of "n + 1" prec]
arctan_0_1_bounds'[OF assms, of "n - 1" prec]
by (auto simp: get_even_def get_odd_def odd_pos
simp del: ub_arctan_horner.simps lb_arctan_horner.simps)
lemma arctan_lower_bound:
assumes "0 ≤ x"
shows "x / (1 + x⇧2) ≤ arctan x" (is "?l x ≤ _")
proof -
have "?l x - arctan x ≤ ?l 0 - arctan 0"
using assms
by (intro DERIV_nonpos_imp_nonincreasing[where f="λx. ?l x - arctan x"])
(auto intro!: derivative_eq_intros simp: add_nonneg_eq_0_iff field_simps)
thus ?thesis by simp
qed
lemma arctan_divide_mono: "0 < x ⟹ x ≤ y ⟹ arctan y / y ≤ arctan x / x"
by (rule DERIV_nonpos_imp_nonincreasing[where f="λx. arctan x / x"])
(auto intro!: derivative_eq_intros divide_nonpos_nonneg
simp: inverse_eq_divide arctan_lower_bound)
lemma arctan_mult_mono: "0 ≤ x ⟹ x ≤ y ⟹ x * arctan y ≤ y * arctan x"
using arctan_divide_mono[of x y] by (cases "x = 0") (simp_all add: field_simps)
lemma arctan_mult_le:
assumes "0 ≤ x" "x ≤ y" "y * z ≤ arctan y"
shows "x * z ≤ arctan x"
proof (cases "x = 0")
case True
then show ?thesis by simp
next
case False
with assms have "z ≤ arctan y / y" by (simp add: field_simps)
also have "… ≤ arctan x / x" using assms ‹x ≠ 0› by (auto intro!: arctan_divide_mono)
finally show ?thesis using assms ‹x ≠ 0› by (simp add: field_simps)
qed
lemma arctan_le_mult:
assumes "0 < x" "x ≤ y" "arctan x ≤ x * z"
shows "arctan y ≤ y * z"
proof -
from assms have "arctan y / y ≤ arctan x / x" by (auto intro!: arctan_divide_mono)
also have "… ≤ z" using assms by (auto simp: field_simps)
finally show ?thesis using assms by (simp add: field_simps)
qed
lemma arctan_0_1_bounds_le:
assumes "0 ≤ x" "x ≤ 1" "0 < real_of_float xl" "real_of_float xl ≤ x * x" "x * x ≤ real_of_float xu" "real_of_float xu ≤ 1"
shows "arctan x ∈
{x * lb_arctan_horner p1 (get_even n) 1 xu .. x * ub_arctan_horner p2 (get_odd n) 1 xl}"
proof -
from assms have "real_of_float xl ≤ 1" "sqrt (real_of_float xl) ≤ x" "x ≤ sqrt (real_of_float xu)" "0 ≤ real_of_float xu"
"0 ≤ real_of_float xl" "0 < sqrt (real_of_float xl)"
by (auto intro!: real_le_rsqrt real_le_lsqrt simp: power2_eq_square)
from arctan_0_1_bounds[OF ‹0 ≤ real_of_float xu› ‹real_of_float xu ≤ 1›]
have "sqrt (real_of_float xu) * real_of_float (lb_arctan_horner p1 (get_even n) 1 xu) ≤ arctan (sqrt (real_of_float xu))"
by simp
from arctan_mult_le[OF ‹0 ≤ x› ‹x ≤ sqrt _› this]
have "x * real_of_float (lb_arctan_horner p1 (get_even n) 1 xu) ≤ arctan x" .
moreover
from arctan_0_1_bounds[OF ‹0 ≤ real_of_float xl› ‹real_of_float xl ≤ 1›]
have "arctan (sqrt (real_of_float xl)) ≤ sqrt (real_of_float xl) * real_of_float (ub_arctan_horner p2 (get_odd n) 1 xl)"
by simp
from arctan_le_mult[OF ‹0 < sqrt xl› ‹sqrt xl ≤ x› this]
have "arctan x ≤ x * real_of_float (ub_arctan_horner p2 (get_odd n) 1 xl)" .
ultimately show ?thesis by simp
qed
lemma arctan_0_1_bounds_round:
assumes "0 ≤ real_of_float x" "real_of_float x ≤ 1"
shows "arctan x ∈
{real_of_float x * lb_arctan_horner p1 (get_even n) 1 (float_round_up (Suc p2) (x * x)) ..
real_of_float x * ub_arctan_horner p3 (get_odd n) 1 (float_round_down (Suc p4) (x * x))}"
using assms
apply (cases "x > 0")
apply (intro arctan_0_1_bounds_le)
apply (auto simp: float_round_down.rep_eq float_round_up.rep_eq
intro!: truncate_up_le1 mult_le_one truncate_down_le truncate_up_le truncate_down_pos
mult_pos_pos)
done
subsection "Compute π"
definition ub_pi :: "nat ⇒ float" where
"ub_pi prec =
(let
A = rapprox_rat prec 1 5 ;
B = lapprox_rat prec 1 239
in ((Float 1 2) * float_plus_up prec
((Float 1 2) * float_round_up prec (A * (ub_arctan_horner prec (get_odd (prec div 4 + 1)) 1
(float_round_down (Suc prec) (A * A)))))
(- float_round_down prec (B * (lb_arctan_horner prec (get_even (prec div 14 + 1)) 1
(float_round_up (Suc prec) (B * B)))))))"
definition lb_pi :: "nat ⇒ float" where
"lb_pi prec =
(let
A = lapprox_rat prec 1 5 ;
B = rapprox_rat prec 1 239
in ((Float 1 2) * float_plus_down prec
((Float 1 2) * float_round_down prec (A * (lb_arctan_horner prec (get_even (prec div 4 + 1)) 1
(float_round_up (Suc prec) (A * A)))))
(- float_round_up prec (B * (ub_arctan_horner prec (get_odd (prec div 14 + 1)) 1
(float_round_down (Suc prec) (B * B)))))))"
lemma pi_boundaries: "pi ∈ {(lb_pi n) .. (ub_pi n)}"
proof -
have machin_pi: "pi = 4 * (4 * arctan (1 / 5) - arctan (1 / 239))"
unfolding machin[symmetric] by auto
{
fix prec n :: nat
fix k :: int
assume "1 < k" hence "0 ≤ k" and "0 < k" and "1 ≤ k" by auto
let ?k = "rapprox_rat prec 1 k"
let ?kl = "float_round_down (Suc prec) (?k * ?k)"
have "1 div k = 0" using div_pos_pos_trivial[OF _ ‹1 < k›] by auto
have "0 ≤ real_of_float ?k" by (rule order_trans[OF _ rapprox_rat]) (auto simp add: ‹0 ≤ k›)
have "real_of_float ?k ≤ 1"
by (auto simp add: ‹0 < k› ‹1 ≤ k› less_imp_le
intro!: mult_le_one order_trans[OF _ rapprox_rat] rapprox_rat_le1)
have "1 / k ≤ ?k" using rapprox_rat[where x=1 and y=k] by auto
hence "arctan (1 / k) ≤ arctan ?k" by (rule arctan_monotone')
also have "… ≤ (?k * ub_arctan_horner prec (get_odd n) 1 ?kl)"
using arctan_0_1_bounds_round[OF ‹0 ≤ real_of_float ?k› ‹real_of_float ?k ≤ 1›]
by auto
finally have "arctan (1 / k) ≤ ?k * ub_arctan_horner prec (get_odd n) 1 ?kl" .
} note ub_arctan = this
{
fix prec n :: nat
fix k :: int
assume "1 < k" hence "0 ≤ k" and "0 < k" by auto
let ?k = "lapprox_rat prec 1 k"
let ?ku = "float_round_up (Suc prec) (?k * ?k)"
have "1 div k = 0" using div_pos_pos_trivial[OF _ ‹1 < k›] by auto
have "1 / k ≤ 1" using ‹1 < k› by auto
have "0 ≤ real_of_float ?k" using lapprox_rat_nonneg[where x=1 and y=k, OF zero_le_one ‹0 ≤ k›]
by (auto simp add: ‹1 div k = 0›)
have "0 ≤ real_of_float (?k * ?k)" by simp
have "real_of_float ?k ≤ 1" using lapprox_rat by (rule order_trans, auto simp add: ‹1 / k ≤ 1›)
hence "real_of_float (?k * ?k) ≤ 1" using ‹0 ≤ real_of_float ?k› by (auto intro!: mult_le_one)
have "?k ≤ 1 / k" using lapprox_rat[where x=1 and y=k] by auto
have "?k * lb_arctan_horner prec (get_even n) 1 ?ku ≤ arctan ?k"
using arctan_0_1_bounds_round[OF ‹0 ≤ real_of_float ?k› ‹real_of_float ?k ≤ 1›]
by auto
also have "… ≤ arctan (1 / k)" using ‹?k ≤ 1 / k› by (rule arctan_monotone')
finally have "?k * lb_arctan_horner prec (get_even n) 1 ?ku ≤ arctan (1 / k)" .
} note lb_arctan = this
have "pi ≤ ub_pi n "
unfolding ub_pi_def machin_pi Let_def times_float.rep_eq Float_num
using lb_arctan[of 239] ub_arctan[of 5] powr_realpow[of 2 2]
by (intro mult_left_mono float_plus_up_le float_plus_down_le)
(auto intro!: mult_left_mono float_round_down_le float_round_up_le diff_mono)
moreover have "lb_pi n ≤ pi"
unfolding lb_pi_def machin_pi Let_def times_float.rep_eq Float_num
using lb_arctan[of 5] ub_arctan[of 239]
by (intro mult_left_mono float_plus_up_le float_plus_down_le)
(auto intro!: mult_left_mono float_round_down_le float_round_up_le diff_mono)
ultimately show ?thesis by auto
qed
lift_definition pi_float_interval::"nat ⇒ float interval" is "λprec. (lb_pi prec, ub_pi prec)"
using pi_boundaries
by (auto intro: order_trans)
lemma lower_pi_float_interval: "lower (pi_float_interval prec) = lb_pi prec"
by transfer auto
lemma upper_pi_float_interval: "upper (pi_float_interval prec) = ub_pi prec"
by transfer auto
lemma pi_float_interval: "pi ∈ set_of (real_interval (pi_float_interval prec))"
using pi_boundaries
by (auto simp: set_of_eq lower_pi_float_interval upper_pi_float_interval)
subsection "Compute arcus tangens in the entire domain"
function lb_arctan :: "nat ⇒ float ⇒ float" and ub_arctan :: "nat ⇒ float ⇒ float" where
"lb_arctan prec x =
(let
ub_horner = λ x. float_round_up prec
(x *
ub_arctan_horner prec (get_odd (prec div 4 + 1)) 1 (float_round_down (Suc prec) (x * x)));
lb_horner = λ x. float_round_down prec
(x *
lb_arctan_horner prec (get_even (prec div 4 + 1)) 1 (float_round_up (Suc prec) (x * x)))
in
if x < 0 then - ub_arctan prec (-x)
else if x ≤ Float 1 (- 1) then lb_horner x
else if x ≤ Float 1 1 then
Float 1 1 *
lb_horner
(float_divl prec x
(float_plus_up prec 1
(ub_sqrt prec (float_plus_up prec 1 (float_round_up prec (x * x))))))
else let inv = float_divr prec 1 x in
if inv > 1 then 0
else float_plus_down prec (lb_pi prec * Float 1 (- 1)) ( - ub_horner inv))"
| "ub_arctan prec x =
(let
lb_horner = λ x. float_round_down prec
(x *
lb_arctan_horner prec (get_even (prec div 4 + 1)) 1 (float_round_up (Suc prec) (x * x))) ;
ub_horner = λ x. float_round_up prec
(x *
ub_arctan_horner prec (get_odd (prec div 4 + 1)) 1 (float_round_down (Suc prec) (x * x)))
in if x < 0 then - lb_arctan prec (-x)
else if x ≤ Float 1 (- 1) then ub_horner x
else if x ≤ Float 1 1 then
let y = float_divr prec x
(float_plus_down
(Suc prec) 1 (lb_sqrt prec (float_plus_down prec 1 (float_round_down prec (x * x)))))
in if y > 1 then ub_pi prec * Float 1 (- 1) else Float 1 1 * ub_horner y
else float_plus_up prec (ub_pi prec * Float 1 (- 1)) ( - lb_horner (float_divl prec 1 x)))"
by pat_completeness auto
termination
by (relation "measure (λ v. let (prec, x) = case_sum id id v in (if x < 0 then 1 else 0))", auto)
declare ub_arctan_horner.simps[simp del]
declare lb_arctan_horner.simps[simp del]
lemma lb_arctan_bound':
assumes "0 ≤ real_of_float x"
shows "lb_arctan prec x ≤ arctan x"
proof -
have "¬ x < 0" and "0 ≤ x"
using ‹0 ≤ real_of_float x› by (auto intro!: truncate_up_le )
let "?ub_horner x" =
"x * ub_arctan_horner prec (get_odd (prec div 4 + 1)) 1 (float_round_down (Suc prec) (x * x))"
and "?lb_horner x" =
"x * lb_arctan_horner prec (get_even (prec div 4 + 1)) 1 (float_round_up (Suc prec) (x * x))"
show ?thesis
proof (cases "x ≤ Float 1 (- 1)")
case True
hence "real_of_float x ≤ 1" by simp
from arctan_0_1_bounds_round[OF ‹0 ≤ real_of_float x› ‹real_of_float x ≤ 1›]
show ?thesis
unfolding lb_arctan.simps Let_def if_not_P[OF ‹¬ x < 0›] if_P[OF True] using ‹0 ≤ x›
by (auto intro!: float_round_down_le)
next
case False
hence "0 < real_of_float x" by auto
let ?R = "1 + sqrt (1 + real_of_float x * real_of_float x)"
let ?sxx = "float_plus_up prec 1 (float_round_up prec (x * x))"
let ?fR = "float_plus_up prec 1 (ub_sqrt prec ?sxx)"
let ?DIV = "float_divl prec x ?fR"
have divisor_gt0: "0 < ?R" by (auto intro: add_pos_nonneg)
have "sqrt (1 + x*x) ≤ sqrt ?sxx"
by (auto simp: float_plus_up.rep_eq plus_up_def float_round_up.rep_eq intro!: truncate_up_le)
also have "… ≤ ub_sqrt prec ?sxx"
using bnds_sqrt'[of ?sxx prec] by auto
finally
have "sqrt (1 + x*x) ≤ ub_sqrt prec ?sxx" .
hence "?R ≤ ?fR" by (auto simp: float_plus_up.rep_eq plus_up_def intro!: truncate_up_le)
hence "0 < ?fR" and "0 < real_of_float ?fR" using ‹0 < ?R› by auto
have monotone: "?DIV ≤ x / ?R"
proof -
have "?DIV ≤ real_of_float x / ?fR" by (rule float_divl)
also have "… ≤ x / ?R" by (rule divide_left_mono[OF ‹?R ≤ ?fR› ‹0 ≤ real_of_float x› mult_pos_pos[OF order_less_le_trans[OF divisor_gt0 ‹?R ≤ real_of_float ?fR›] divisor_gt0]])
finally show ?thesis .
qed
show ?thesis
proof (cases "x ≤ Float 1 1")
case True
have "x ≤ sqrt (1 + x * x)"
using real_sqrt_sum_squares_ge2[where x=1, unfolded numeral_2_eq_2] by auto
also note ‹… ≤ (ub_sqrt prec ?sxx)›
finally have "real_of_float x ≤ ?fR"
by (auto simp: float_plus_up.rep_eq plus_up_def intro!: truncate_up_le)
moreover have "?DIV ≤ real_of_float x / ?fR"
by (rule float_divl)
ultimately have "real_of_float ?DIV ≤ 1"
unfolding divide_le_eq_1_pos[OF ‹0 < real_of_float ?fR›, symmetric] by auto
have "0 ≤ real_of_float ?DIV"
using float_divl_lower_bound[OF ‹0 ≤ x›] ‹0 < ?fR›
unfolding less_eq_float_def by auto
from arctan_0_1_bounds_round[OF ‹0 ≤ real_of_float (?DIV)› ‹real_of_float (?DIV) ≤ 1›]
have "Float 1 1 * ?lb_horner ?DIV ≤ 2 * arctan ?DIV"
by simp
also have "… ≤ 2 * arctan (x / ?R)"
using arctan_monotone'[OF monotone] by (auto intro!: mult_left_mono arctan_monotone')
also have "2 * arctan (x / ?R) = arctan x"
using arctan_half[symmetric] unfolding numeral_2_eq_2 power_Suc2 power_0 mult_1_left .
finally show ?thesis
unfolding lb_arctan.simps Let_def if_not_P[OF ‹¬ x < 0›]
if_not_P[OF ‹¬ x ≤ Float 1 (- 1)›] if_P[OF True]
by (auto simp: float_round_down.rep_eq
intro!: order_trans[OF mult_left_mono[OF truncate_down]])
next
case False
hence "2 < real_of_float x" by auto
hence "1 ≤ real_of_float x" by auto
let "?invx" = "float_divr prec 1 x"
have "0 ≤ arctan x" using arctan_monotone'[OF ‹0 ≤ real_of_float x›]
using arctan_tan[of 0, unfolded tan_zero] by auto
show ?thesis
proof (cases "1 < ?invx")
case True
show ?thesis
unfolding lb_arctan.simps Let_def if_not_P[OF ‹¬ x < 0›]
if_not_P[OF ‹¬ x ≤ Float 1 (- 1)›] if_not_P[OF False] if_P[OF True]
using ‹0 ≤ arctan x› by auto
next
case False
hence "real_of_float ?invx ≤ 1" by auto
have "0 ≤ real_of_float ?invx"
by (rule order_trans[OF _ float_divr]) (auto simp add: ‹0 ≤ real_of_float x›)
have "1 / x ≠ 0" and "0 < 1 / x"
using ‹0 < real_of_float x› by auto
have "arctan (1 / x) ≤ arctan ?invx"
unfolding one_float.rep_eq[symmetric] by (rule arctan_monotone', rule float_divr)
also have "… ≤ ?ub_horner ?invx"
using arctan_0_1_bounds_round[OF ‹0 ≤ real_of_float ?invx› ‹real_of_float ?invx ≤ 1›]
by (auto intro!: float_round_up_le)
also note float_round_up
finally have "pi / 2 - float_round_up prec (?ub_horner ?invx) ≤ arctan x"
using ‹0 ≤ arctan x› arctan_inverse[OF ‹1 / x ≠ 0›]
unfolding sgn_pos[OF ‹0 < 1 / real_of_float x›] le_diff_eq by auto
moreover
have "lb_pi prec * Float 1 (- 1) ≤ pi / 2"
unfolding Float_num times_divide_eq_right mult_1_left using pi_boundaries by simp
ultimately
show ?thesis
unfolding lb_arctan.simps Let_def if_not_P[OF ‹¬ x < 0›]
if_not_P[OF ‹¬ x ≤ Float 1 (- 1)›] if_not_P[OF ‹¬ x ≤ Float 1 1›] if_not_P[OF False]
by (auto intro!: float_plus_down_le)
qed
qed
qed
qed
lemma ub_arctan_bound':
assumes "0 ≤ real_of_float x"
shows "arctan x ≤ ub_arctan prec x"
proof -
have "¬ x < 0" and "0 ≤ x"
using ‹0 ≤ real_of_float x› by auto
let "?ub_horner x" =
"float_round_up prec (x * ub_arctan_horner prec (get_odd (prec div 4 + 1)) 1 (float_round_down (Suc prec) (x * x)))"
let "?lb_horner x" =
"float_round_down prec (x * lb_arctan_horner prec (get_even (prec div 4 + 1)) 1 (float_round_up (Suc prec) (x * x)))"
show ?thesis
proof (cases "x ≤ Float 1 (- 1)")
case True
hence "real_of_float x ≤ 1" by auto
show ?thesis
unfolding ub_arctan.simps Let_def if_not_P[OF ‹¬ x < 0›] if_P[OF True]
using arctan_0_1_bounds_round[OF ‹0 ≤ real_of_float x› ‹real_of_float x ≤ 1›]
by (auto intro!: float_round_up_le)
next
case False
hence "0 < real_of_float x" by auto
let ?R = "1 + sqrt (1 + real_of_float x * real_of_float x)"
let ?sxx = "float_plus_down prec 1 (float_round_down prec (x * x))"
let ?fR = "float_plus_down (Suc prec) 1 (lb_sqrt prec ?sxx)"
let ?DIV = "float_divr prec x ?fR"
have sqr_ge0: "0 ≤ 1 + real_of_float x * real_of_float x"
using sum_power2_ge_zero[of 1 "real_of_float x", unfolded numeral_2_eq_2] by auto
hence "0 ≤ real_of_float (1 + x*x)" by auto
hence divisor_gt0: "0 < ?R" by (auto intro: add_pos_nonneg)
have "lb_sqrt prec ?sxx ≤ sqrt ?sxx"
using bnds_sqrt'[of ?sxx] by auto
also have "… ≤ sqrt (1 + x*x)"
by (auto simp: float_plus_down.rep_eq plus_down_def float_round_down.rep_eq truncate_down_le)
finally have "lb_sqrt prec ?sxx ≤ sqrt (1 + x*x)" .
hence "?fR ≤ ?R"
by (auto simp: float_plus_down.rep_eq plus_down_def truncate_down_le)
have "0 < real_of_float ?fR"
by (auto simp: float_plus_down.rep_eq plus_down_def float_round_down.rep_eq
intro!: truncate_down_ge1 lb_sqrt_lower_bound order_less_le_trans[OF zero_less_one]
truncate_down_nonneg add_nonneg_nonneg)
have monotone: "x / ?R ≤ (float_divr prec x ?fR)"
proof -
from divide_left_mono[OF ‹?fR ≤ ?R› ‹0 ≤ real_of_float x› mult_pos_pos[OF divisor_gt0 ‹0 < real_of_float ?fR›]]
have "x / ?R ≤ x / ?fR" .
also have "… ≤ ?DIV" by (rule float_divr)
finally show ?thesis .
qed
show ?thesis
proof (cases "x ≤ Float 1 1")
case True
show ?thesis
proof (cases "?DIV > 1")
case True
have "pi / 2 ≤ ub_pi prec * Float 1 (- 1)"
unfolding Float_num times_divide_eq_right mult_1_left using pi_boundaries by auto
from order_less_le_trans[OF arctan_ubound this, THEN less_imp_le]
show ?thesis
unfolding ub_arctan.simps Let_def if_not_P[OF ‹¬ x < 0›]
if_not_P[OF ‹¬ x ≤ Float 1 (- 1)›] if_P[OF ‹x ≤ Float 1 1›] if_P[OF True] .
next
case False
hence "real_of_float ?DIV ≤ 1" by auto
have "0 ≤ x / ?R"
using ‹0 ≤ real_of_float x› ‹0 < ?R› unfolding zero_le_divide_iff by auto
hence "0 ≤ real_of_float ?DIV"
using monotone by (rule order_trans)
have "arctan x = 2 * arctan (x / ?R)"
using arctan_half unfolding numeral_2_eq_2 power_Suc2 power_0 mult_1_left .
also have "… ≤ 2 * arctan (?DIV)"
using arctan_monotone'[OF monotone] by (auto intro!: mult_left_mono)
also have "… ≤ (Float 1 1 * ?ub_horner ?DIV)" unfolding Float_num
using arctan_0_1_bounds_round[OF ‹0 ≤ real_of_float ?DIV› ‹real_of_float ?DIV ≤ 1›]
by (auto intro!: float_round_up_le)
finally show ?thesis
unfolding ub_arctan.simps Let_def if_not_P[OF ‹¬ x < 0›]
if_not_P[OF ‹¬ x ≤ Float 1 (- 1)›] if_P[OF ‹x ≤ Float 1 1›] if_not_P[OF False] .
qed
next
case False
hence "2 < real_of_float x" by auto
hence "1 ≤ real_of_float x" by auto
hence "0 < real_of_float x" by auto
hence "0 < x" by auto
let "?invx" = "float_divl prec 1 x"
have "0 ≤ arctan x"
using arctan_monotone'[OF ‹0 ≤ real_of_float x›] and arctan_tan[of 0, unfolded tan_zero] by auto
have "real_of_float ?invx ≤ 1"
unfolding less_float_def
by (rule order_trans[OF float_divl])
(auto simp add: ‹1 ≤ real_of_float x› divide_le_eq_1_pos[OF ‹0 < real_of_float x›])
have "0 ≤ real_of_float ?invx"
using ‹0 < x› by (intro float_divl_lower_bound) auto
have "1 / x ≠ 0" and "0 < 1 / x"
using ‹0 < real_of_float x› by auto
have "(?lb_horner ?invx) ≤ arctan (?invx)"
using arctan_0_1_bounds_round[OF ‹0 ≤ real_of_float ?invx› ‹real_of_float ?invx ≤ 1›]
by (auto intro!: float_round_down_le)
also have "… ≤ arctan (1 / x)"
unfolding one_float.rep_eq[symmetric] by (rule arctan_monotone') (rule float_divl)
finally have "arctan x ≤ pi / 2 - (?lb_horner ?invx)"
using ‹0 ≤ arctan x› arctan_inverse[OF ‹1 / x ≠ 0›]
unfolding sgn_pos[OF ‹0 < 1 / x›] le_diff_eq by auto
moreover
have "pi / 2 ≤ ub_pi prec * Float 1 (- 1)"
unfolding Float_num times_divide_eq_right mult_1_right
using pi_boundaries by auto
ultimately
show ?thesis
unfolding ub_arctan.simps Let_def if_not_P[OF ‹¬ x < 0›]
if_not_P[OF ‹¬ x ≤ Float 1 (- 1)›] if_not_P[OF False]
by (auto intro!: float_round_up_le float_plus_up_le)
qed
qed
qed
lemma arctan_boundaries: "arctan x ∈ {(lb_arctan prec x) .. (ub_arctan prec x)}"
proof (cases "0 ≤ x")
case True
hence "0 ≤ real_of_float x" by auto
show ?thesis
using ub_arctan_bound'[OF ‹0 ≤ real_of_float x›] lb_arctan_bound'[OF ‹0 ≤ real_of_float x›]
unfolding atLeastAtMost_iff by auto
next
case False
let ?mx = "-x"
from False have "x < 0" and "0 ≤ real_of_float ?mx"
by auto
hence bounds: "lb_arctan prec ?mx ≤ arctan ?mx ∧ arctan ?mx ≤ ub_arctan prec ?mx"
using ub_arctan_bound'[OF ‹0 ≤ real_of_float ?mx›] lb_arctan_bound'[OF ‹0 ≤ real_of_float ?mx›] by auto
show ?thesis
unfolding minus_float.rep_eq arctan_minus lb_arctan.simps[where x=x]
ub_arctan.simps[where x=x] Let_def if_P[OF ‹x < 0›]
unfolding atLeastAtMost_iff using bounds[unfolded minus_float.rep_eq arctan_minus]
by (simp add: arctan_minus)
qed
lemma bnds_arctan: "∀ (x::real) lx ux. (l, u) = (lb_arctan prec lx, ub_arctan prec ux) ∧ x ∈ {lx .. ux} ⟶ l ≤ arctan x ∧ arctan x ≤ u"
proof (rule allI, rule allI, rule allI, rule impI)
fix x :: real
fix lx ux
assume "(l, u) = (lb_arctan prec lx, ub_arctan prec ux) ∧ x ∈ {lx .. ux}"
hence l: "lb_arctan prec lx = l "
and u: "ub_arctan prec ux = u"
and x: "x ∈ {lx .. ux}"
by auto
show "l ≤ arctan x ∧ arctan x ≤ u"
proof
show "l ≤ arctan x"
proof -
from arctan_boundaries[of lx prec, unfolded l]
have "l ≤ arctan lx" by (auto simp del: lb_arctan.simps)
also have "… ≤ arctan x" using x by (auto intro: arctan_monotone')
finally show ?thesis .
qed
show "arctan x ≤ u"
proof -
have "arctan x ≤ arctan ux" using x by (auto intro: arctan_monotone')
also have "… ≤ u" using arctan_boundaries[of ux prec, unfolded u] by (auto simp del: ub_arctan.simps)
finally show ?thesis .
qed
qed
qed
lemmas [simp del] = lb_arctan.simps ub_arctan.simps
lemma lb_arctan: "arctan (real_of_float x) ≤ y ⟹ real_of_float (lb_arctan prec x) ≤ y"
and ub_arctan: "y ≤ arctan x ⟹ y ≤ ub_arctan prec x"
for x::float and y::real
using arctan_boundaries[of x prec] by auto
lift_definition arctan_float_interval :: "nat ⇒ float interval ⇒ float interval"
is "λprec. λ(lx, ux). (lb_arctan prec lx, ub_arctan prec ux)"
by (auto intro!: lb_arctan ub_arctan arctan_monotone')
lemma lower_arctan_float_interval: "lower (arctan_float_interval p x) = lb_arctan p (lower x)"
by transfer auto
lemma upper_arctan_float_interval: "upper (arctan_float_interval p x) = ub_arctan p (upper x)"
by transfer auto
lemma arctan_float_interval:
"arctan ` set_of (real_interval x) ⊆ set_of (real_interval (arctan_float_interval p x))"
by (auto simp: set_of_eq lower_arctan_float_interval upper_arctan_float_interval
intro!: lb_arctan ub_arctan arctan_monotone')
lemma arctan_float_intervalI:
"arctan x ∈⇩r arctan_float_interval p X" if "x ∈⇩r X"
using arctan_float_interval[of X p] that
by auto
section "Sinus and Cosinus"
subsection "Compute the cosinus and sinus series"
fun ub_sin_cos_aux :: "nat ⇒ nat ⇒ nat ⇒ nat ⇒ float ⇒ float"
and lb_sin_cos_aux :: "nat ⇒ nat ⇒ nat ⇒ nat ⇒ float ⇒ float" where
"ub_sin_cos_aux prec 0 i k x = 0"
| "ub_sin_cos_aux prec (Suc n) i k x = float_plus_up prec
(rapprox_rat prec 1 k) (-
float_round_down prec (x * (lb_sin_cos_aux prec n (i + 2) (k * i * (i + 1)) x)))"
| "lb_sin_cos_aux prec 0 i k x = 0"
| "lb_sin_cos_aux prec (Suc n) i k x = float_plus_down prec
(lapprox_rat prec 1 k) (-
float_round_up prec (x * (ub_sin_cos_aux prec n (i + 2) (k * i * (i + 1)) x)))"
lemma cos_aux:
shows "(lb_sin_cos_aux prec n 1 1 (x * x)) ≤ (∑ i=0..<n. (- 1) ^ i * (1/(fact (2 * i))) * x ^(2 * i))" (is "?lb")
and "(∑ i=0..<n. (- 1) ^ i * (1/(fact (2 * i))) * x^(2 * i)) ≤ (ub_sin_cos_aux prec n 1 1 (x * x))" (is "?ub")
proof -
have "0 ≤ real_of_float (x * x)" by auto
let "?f n" = "fact (2 * n) :: nat"
have f_eq: "?f (Suc n) = ?f n * ((λi. i + 2) ^^ n) 1 * (((λi. i + 2) ^^ n) 1 + 1)" for n
proof -
have "⋀m. ((λi. i + 2) ^^ n) m = m + 2 * n" by (induct n) auto
then show ?thesis by auto
qed
from horner_bounds[where lb="lb_sin_cos_aux prec" and ub="ub_sin_cos_aux prec" and j'=0,
OF ‹0 ≤ real_of_float (x * x)› f_eq lb_sin_cos_aux.simps ub_sin_cos_aux.simps]
show ?lb and ?ub
by (auto simp add: power_mult power2_eq_square[of "real_of_float x"])
qed
lemma lb_sin_cos_aux_zero_le_one: "lb_sin_cos_aux prec n i j 0 ≤ 1"
by (cases j n rule: nat.exhaust[case_product nat.exhaust])
(auto intro!: float_plus_down_le order_trans[OF lapprox_rat])
lemma one_le_ub_sin_cos_aux: "odd n ⟹ 1 ≤ ub_sin_cos_aux prec n i (Suc 0) 0"
by (cases n) (auto intro!: float_plus_up_le order_trans[OF _ rapprox_rat])
lemma cos_boundaries:
assumes "0 ≤ real_of_float x" and "x ≤ pi / 2"
shows "cos x ∈ {(lb_sin_cos_aux prec (get_even n) 1 1 (x * x)) .. (ub_sin_cos_aux prec (get_odd n) 1 1 (x * x))}"
proof (cases "real_of_float x = 0")
case False
hence "real_of_float x ≠ 0" by auto
hence "0 < x" and "0 < real_of_float x"
using ‹0 ≤ real_of_float x› by auto
have "0 < x * x"
using ‹0 < x› by simp
have morph_to_if_power: "(∑ i=0..<n. (-1::real) ^ i * (1/(fact (2 * i))) * x ^ (2 * i)) =
(∑ i = 0 ..< 2 * n. (if even(i) then ((- 1) ^ (i div 2))/((fact i)) else 0) * x ^ i)"
(is "?sum = ?ifsum") for x n
proof -
have "?sum = ?sum + (∑ j = 0 ..< n. 0)" by auto
also have "… =
(∑ j = 0 ..< n. (- 1) ^ ((2 * j) div 2) / ((fact (2 * j))) * x ^(2 * j)) + (∑ j = 0 ..< n. 0)" by auto
also have "… = (∑ i = 0 ..< 2 * n. if even i then (- 1) ^ (i div 2) / ((fact i)) * x ^ i else 0)"
unfolding sum_split_even_odd atLeast0LessThan ..
also have "… = (∑ i = 0 ..< 2 * n. (if even i then (- 1) ^ (i div 2) / ((fact i)) else 0) * x ^ i)"
by (rule sum.cong) auto
finally show ?thesis .
qed
{ fix n :: nat assume "0 < n"
hence "0 < 2 * n" by auto
obtain t where "0 < t" and "t < real_of_float x" and
cos_eq: "cos x = (∑ i = 0 ..< 2 * n. (if even(i) then ((- 1) ^ (i div 2))/((fact i)) else 0) * (real_of_float x) ^ i)
+ (cos (t + 1/2 * (2 * n) * pi) / (fact (2*n))) * (real_of_float x)^(2*n)"
(is "_ = ?SUM + ?rest / ?fact * ?pow")
using Maclaurin_cos_expansion2[OF ‹0 < real_of_float x› ‹0 < 2 * n›]
unfolding cos_coeff_def atLeast0LessThan by auto
have "cos t * (- 1) ^ n = cos t * cos (n * pi) + sin t * sin (n * pi)" by auto
also have "… = cos (t + n * pi)" by (simp add: cos_add)
also have "… = ?rest" by auto
finally have "cos t * (- 1) ^ n = ?rest" .
moreover
have "t ≤ pi / 2" using ‹t < real_of_float x› and ‹x ≤ pi / 2› by auto
hence "0 ≤ cos t" using ‹0 < t› and cos_ge_zero by auto
ultimately have even: "even n ⟹ 0 ≤ ?rest" and odd: "odd n ⟹ 0 ≤ - ?rest " by auto
have "0 < ?fact" by auto
have "0 < ?pow" using ‹0 < real_of_float x› by auto
{
assume "even n"
have "(lb_sin_cos_aux prec n 1 1 (x * x)) ≤ ?SUM"
unfolding morph_to_if_power[symmetric] using cos_aux by auto
also have "… ≤ cos x"
proof -
from even[OF ‹even n›] ‹0 < ?fact› ‹0 < ?pow›
have "0 ≤ (?rest / ?fact) * ?pow" by simp
thus ?thesis unfolding cos_eq by auto
qed
finally have "(lb_sin_cos_aux prec n 1 1 (x * x)) ≤ cos x" .
} note lb = this
{
assume "odd n"
have "cos x ≤ ?SUM"
proof -
from ‹0 < ?fact› and ‹0 < ?pow› and odd[OF ‹odd n›]
have "0 ≤ (- ?rest) / ?fact * ?pow"
by (metis mult_nonneg_nonneg divide_nonneg_pos less_imp_le)
thus ?thesis unfolding cos_eq by auto
qed
also have "… ≤ (ub_sin_cos_aux prec n 1 1 (x * x))"
unfolding morph_to_if_power[symmetric] using cos_aux by auto
finally have "cos x ≤ (ub_sin_cos_aux prec n 1 1 (x * x))" .
} note ub = this and lb
} note ub = this(1) and lb = this(2)
have "cos x ≤ (ub_sin_cos_aux prec (get_odd n) 1 1 (x * x))"
using ub[OF odd_pos[OF get_odd] get_odd] .
moreover have "(lb_sin_cos_aux prec (get_even n) 1 1 (x * x)) ≤ cos x"
proof (cases "0 < get_even n")
case True
show ?thesis using lb[OF True get_even] .
next
case False
hence "get_even n = 0" by auto
have "- (pi / 2) ≤ x"
by (rule order_trans[OF _ ‹0 < real_of_float x›[THEN less_imp_le]]) auto
with ‹x ≤ pi / 2› show ?thesis
unfolding ‹get_even n = 0› lb_sin_cos_aux.simps minus_float.rep_eq zero_float.rep_eq
using cos_ge_zero by auto
qed
ultimately show ?thesis by auto
next
case True
hence "x = 0"
by (simp add: real_of_float_eq)
thus ?thesis
using lb_sin_cos_aux_zero_le_one one_le_ub_sin_cos_aux
by simp
qed
lemma sin_aux:
assumes "0 ≤ real_of_float x"
shows "(x * lb_sin_cos_aux prec n 2 1 (x * x)) ≤
(∑ i=0..<n. (- 1) ^ i * (1/(fact (2 * i + 1))) * x^(2 * i + 1))" (is "?lb")
and "(∑ i=0..<n. (- 1) ^ i * (1/(fact (2 * i + 1))) * x^(2 * i + 1)) ≤
(x * ub_sin_cos_aux prec n 2 1 (x * x))" (is "?ub")
proof -
have "0 ≤ real_of_float (x * x)" by auto
let "?f n" = "fact (2 * n + 1) :: nat"
have f_eq: "?f (Suc n) = ?f n * ((λi. i + 2) ^^ n) 2 * (((λi. i + 2) ^^ n) 2 + 1)" for n
proof -
have F: "⋀m. ((λi. i + 2) ^^ n) m = m + 2 * n" by (induct n) auto
show ?thesis
unfolding F by auto
qed
from horner_bounds[where lb="lb_sin_cos_aux prec" and ub="ub_sin_cos_aux prec" and j'=0,
OF ‹0 ≤ real_of_float (x * x)› f_eq lb_sin_cos_aux.simps ub_sin_cos_aux.simps]
show "?lb" and "?ub" using ‹0 ≤ real_of_float x›
apply (simp_all only: power_add power_one_right mult.assoc[symmetric] sum_distrib_right[symmetric])
apply (simp_all only: mult.commute[where 'a=real] of_nat_fact)
apply (auto intro!: mult_left_mono simp add: power_mult power2_eq_square[of "real_of_float x"])
done
qed
lemma sin_boundaries:
assumes "0 ≤ real_of_float x"
and "x ≤ pi / 2"
shows "sin x ∈ {(x * lb_sin_cos_aux prec (get_even n) 2 1 (x * x)) .. (x * ub_sin_cos_aux prec (get_odd n) 2 1 (x * x))}"
proof (cases "real_of_float x = 0")
case False
hence "real_of_float x ≠ 0" by auto
hence "0 < x" and "0 < real_of_float x"
using ‹0 ≤ real_of_float x› by auto
have "0 < x * x"
using ‹0 < x› by simp
have sum_morph: "(∑j = 0 ..< n. (- 1) ^ (((2 * j + 1) - Suc 0) div 2) / ((fact (2 * j + 1))) * x ^(2 * j + 1)) =
(∑ i = 0 ..< 2 * n. (if even(i) then 0 else ((- 1) ^ ((i - Suc 0) div 2))/((fact i))) * x ^ i)"
(is "?SUM = _") for x :: real and n
proof -
have pow: "!!i. x ^ (2 * i + 1) = x * x ^ (2 * i)"
by auto
have "?SUM = (∑ j = 0 ..< n. 0) + ?SUM"
by auto
also have "… = (∑ i = 0 ..< 2 * n. if even i then 0 else (- 1) ^ ((i - Suc 0) div 2) / ((fact i)) * x ^ i)"
unfolding sum_split_even_odd atLeast0LessThan ..
also have "… = (∑ i = 0 ..< 2 * n. (if even i then 0 else (- 1) ^ ((i - Suc 0) div 2) / ((fact i))) * x ^ i)"
by (rule sum.cong) auto
finally show ?thesis .
qed
{ fix n :: nat assume "0 < n"
hence "0 < 2 * n + 1" by auto
obtain t where "0 < t" and "t < real_of_float x" and
sin_eq: "sin x = (∑ i = 0 ..< 2 * n + 1. (if even(i) then 0 else ((- 1) ^ ((i - Suc 0) div 2))/((fact i))) * (real_of_float x) ^ i)
+ (sin (t + 1/2 * (2 * n + 1) * pi) / (fact (2*n + 1))) * (real_of_float x)^(2*n + 1)"
(is "_ = ?SUM + ?rest / ?fact * ?pow")
using Maclaurin_sin_expansion3[OF ‹0 < 2 * n + 1› ‹0 < real_of_float x›]
unfolding sin_coeff_def atLeast0LessThan by auto
have "?rest = cos t * (- 1) ^ n"
unfolding sin_add cos_add of_nat_add distrib_right distrib_left by auto
moreover
have "t ≤ pi / 2"
using ‹t < real_of_float x› and ‹x ≤ pi / 2› by auto
hence "0 ≤ cos t"
using ‹0 < t› and cos_ge_zero by auto
ultimately have even: "even n ⟹ 0 ≤ ?rest" and odd: "odd n ⟹ 0 ≤ - ?rest"
by auto
have "0 < ?fact"
by (simp del: fact_Suc)
have "0 < ?pow"
using ‹0 < real_of_float x› by (rule zero_less_power)
{
assume "even n"
have "(x * lb_sin_cos_aux prec n 2 1 (x * x)) ≤
(∑ i = 0 ..< 2 * n. (if even(i) then 0 else ((- 1) ^ ((i - Suc 0) div 2))/((fact i))) * (real_of_float x) ^ i)"
using sin_aux[OF ‹0 ≤ real_of_float x›] unfolding sum_morph[symmetric] by auto
also have "… ≤ ?SUM" by auto
also have "… ≤ sin x"
proof -
from even[OF ‹even n›] ‹0 < ?fact› ‹0 < ?pow›
have "0 ≤ (?rest / ?fact) * ?pow" by simp
thus ?thesis unfolding sin_eq by auto
qed
finally have "(x * lb_sin_cos_aux prec n 2 1 (x * x)) ≤ sin x" .
} note lb = this
{
assume "odd n"
have "sin x ≤ ?SUM"
proof -
from ‹0 < ?fact› and ‹0 < ?pow› and odd[OF ‹odd n›]
have "0 ≤ (- ?rest) / ?fact * ?pow"
by (metis mult_nonneg_nonneg divide_nonneg_pos less_imp_le)
thus ?thesis unfolding sin_eq by auto
qed
also have "… ≤ (∑ i = 0 ..< 2 * n. (if even(i) then 0 else ((- 1) ^ ((i - Suc 0) div 2))/((fact i))) * (real_of_float x) ^ i)"
by auto
also have "… ≤ (x * ub_sin_cos_aux prec n 2 1 (x * x))"
using sin_aux[OF ‹0 ≤ real_of_float x›] unfolding sum_morph[symmetric] by auto
finally have "sin x ≤ (x * ub_sin_cos_aux prec n 2 1 (x * x))" .
} note ub = this and lb
} note ub = this(1) and lb = this(2)
have "sin x ≤ (x * ub_sin_cos_aux prec (get_odd n) 2 1 (x * x))"
using ub[OF odd_pos[OF get_odd] get_odd] .
moreover have "(x * lb_sin_cos_aux prec (get_even n) 2 1 (x * x)) ≤ sin x"
proof (cases "0 < get_even n")
case True
show ?thesis
using lb[OF True get_even] .
next
case False
hence "get_even n = 0" by auto
with ‹x ≤ pi / 2› ‹0 ≤ real_of_float x›
show ?thesis
unfolding ‹get_even n = 0› ub_sin_cos_aux.simps minus_float.rep_eq
using sin_ge_zero by auto
qed
ultimately show ?thesis by auto
next
case True
show ?thesis
proof (cases "n = 0")
case True
thus ?thesis
unfolding ‹n = 0› get_even_def get_odd_def
using ‹real_of_float x = 0› lapprox_rat[where x="-1" and y=1] by auto
next
case False
with not0_implies_Suc obtain m where "n = Suc m" by blast
thus ?thesis
unfolding ‹n = Suc m› get_even_def get_odd_def
using ‹real_of_float x = 0› rapprox_rat[where x=1 and y=1] lapprox_rat[where x=1 and y=1]
by (cases "even (Suc m)") auto
qed
qed
subsection "Compute the cosinus in the entire domain"
definition lb_cos :: "nat ⇒ float ⇒ float" where
"lb_cos prec x = (let
horner = λ x. lb_sin_cos_aux prec (get_even (prec div 4 + 1)) 1 1 (x * x) ;
half = λ x. if x < 0 then - 1 else float_plus_down prec (Float 1 1 * x * x) (- 1)
in if x < Float 1 (- 1) then horner x
else if x < 1 then half (horner (x * Float 1 (- 1)))
else half (half (horner (x * Float 1 (- 2)))))"
definition ub_cos :: "nat ⇒ float ⇒ float" where
"ub_cos prec x = (let
horner = λ x. ub_sin_cos_aux prec (get_odd (prec div 4 + 1)) 1 1 (x * x) ;
half = λ x. float_plus_up prec (Float 1 1 * x * x) (- 1)
in if x < Float 1 (- 1) then horner x
else if x < 1 then half (horner (x * Float 1 (- 1)))
else half (half (horner (x * Float 1 (- 2)))))"
lemma lb_cos:
assumes "0 ≤ real_of_float x" and "x ≤ pi"
shows "cos x ∈ {(lb_cos prec x) .. (ub_cos prec x)}" (is "?cos x ∈ {(?lb x) .. (?ub x) }")
proof -
have x_half[symmetric]: "cos x = 2 * cos (x / 2) * cos (x / 2) - 1" for x :: real
proof -
have "cos x = cos (x / 2 + x / 2)"
by auto
also have "… = cos (x / 2) * cos (x / 2) + sin (x / 2) * sin (x / 2) - sin (x / 2) * sin (x / 2) + cos (x / 2) * cos (x / 2) - 1"
unfolding cos_add by auto
also have "… = 2 * cos (x / 2) * cos (x / 2) - 1"
by algebra
finally show ?thesis .
qed
have "¬ x < 0" using ‹0 ≤ real_of_float x› by auto
let "?ub_horner x" = "ub_sin_cos_aux prec (get_odd (prec div 4 + 1)) 1 1 (x * x)"
let "?lb_horner x" = "lb_sin_cos_aux prec (get_even (prec div 4 + 1)) 1 1 (x * x)"
let "?ub_half x" = "float_plus_up prec (Float 1 1 * x * x) (- 1)"
let "?lb_half x" = "if x < 0 then - 1 else float_plus_down prec (Float 1 1 * x * x) (- 1)"
show ?thesis
proof (cases "x < Float 1 (- 1)")
case True
hence "x ≤ pi / 2"
using pi_ge_two by auto
show ?thesis
unfolding lb_cos_def[where x=x] ub_cos_def[where x=x]
if_not_P[OF ‹¬ x < 0›] if_P[OF ‹x < Float 1 (- 1)›] Let_def
using cos_boundaries[OF ‹0 ≤ real_of_float x› ‹x ≤ pi / 2›] .
next
case False
{ fix y x :: float let ?x2 = "(x * Float 1 (- 1))"
assume "y ≤ cos ?x2" and "-pi ≤ x" and "x ≤ pi"
hence "- (pi / 2) ≤ ?x2" and "?x2 ≤ pi / 2"
using pi_ge_two unfolding Float_num by auto
hence "0 ≤ cos ?x2"
by (rule cos_ge_zero)
have "(?lb_half y) ≤ cos x"
proof (cases "y < 0")
case True
show ?thesis
using cos_ge_minus_one unfolding if_P[OF True] by auto
next
case False
hence "0 ≤ real_of_float y" by auto
from mult_mono[OF ‹y ≤ cos ?x2› ‹y ≤ cos ?x2› ‹0 ≤ cos ?x2› this]
have "real_of_float y * real_of_float y ≤ cos ?x2 * cos ?x2" .
hence "2 * real_of_float y * real_of_float y ≤ 2 * cos ?x2 * cos ?x2"
by auto
hence "2 * real_of_float y * real_of_float y - 1 ≤ 2 * cos (x / 2) * cos (x / 2) - 1"
unfolding Float_num by auto
thus ?thesis
unfolding if_not_P[OF False] x_half Float_num
by (auto intro!: float_plus_down_le)
qed
} note lb_half = this
{ fix y x :: float let ?x2 = "(x * Float 1 (- 1))"
assume ub: "cos ?x2 ≤ y" and "- pi ≤ x" and "x ≤ pi"
hence "- (pi / 2) ≤ ?x2" and "?x2 ≤ pi / 2"
using pi_ge_two unfolding Float_num by auto
hence "0 ≤ cos ?x2" by (rule cos_ge_zero)
have "cos x ≤ (?ub_half y)"
proof -
have "0 ≤ real_of_float y"
using ‹0 ≤ cos ?x2› ub by (rule order_trans)
from mult_mono[OF ub ub this ‹0 ≤ cos ?x2›]
have "cos ?x2 * cos ?x2 ≤ real_of_float y * real_of_float y" .
hence "2 * cos ?x2 * cos ?x2 ≤ 2 * real_of_float y * real_of_float y"
by auto
hence "2 * cos (x / 2) * cos (x / 2) - 1 ≤ 2 * real_of_float y * real_of_float y - 1"
unfolding Float_num by auto
thus ?thesis
unfolding x_half Float_num
by (auto intro!: float_plus_up_le)
qed
} note ub_half = this
let ?x2 = "x * Float 1 (- 1)"
let ?x4 = "x * Float 1 (- 1) * Float 1 (- 1)"
have "-pi ≤ x"
using pi_ge_zero[THEN le_imp_neg_le, unfolded minus_zero] ‹0 ≤ real_of_float x›
by (rule order_trans)
show ?thesis
proof (cases "x < 1")
case True
hence "real_of_float x ≤ 1" by auto
have "0 ≤ real_of_float ?x2" and "?x2 ≤ pi / 2"
using pi_ge_two ‹0 ≤ real_of_float x› using assms by auto
from cos_boundaries[OF this]
have lb: "(?lb_horner ?x2) ≤ ?cos ?x2" and ub: "?cos ?x2 ≤ (?ub_horner ?x2)"
by auto
have "(?lb x) ≤ ?cos x"
proof -
from lb_half[OF lb ‹-pi ≤ x› ‹x ≤ pi›]
show ?thesis
unfolding lb_cos_def[where x=x] Let_def
using ‹¬ x < 0› ‹¬ x < Float 1 (- 1)› ‹x < 1› by auto
qed
moreover have "?cos x ≤ (?ub x)"
proof -
from ub_half[OF ub ‹-pi ≤ x› ‹x ≤ pi›]
show ?thesis
unfolding ub_cos_def[where x=x] Let_def
using ‹¬ x < 0› ‹¬ x < Float 1 (- 1)› ‹x < 1› by auto
qed
ultimately show ?thesis by auto
next
case False
have "0 ≤ real_of_float ?x4" and "?x4 ≤ pi / 2"
using pi_ge_two ‹0 ≤ real_of_float x› ‹x ≤ pi› unfolding Float_num by auto
from cos_boundaries[OF this]
have lb: "(?lb_horner ?x4) ≤ ?cos ?x4" and ub: "?cos ?x4 ≤ (?ub_horner ?x4)"
by auto
have eq_4: "?x2 * Float 1 (- 1) = x * Float 1 (- 2)"
by (auto simp: real_of_float_eq)
have "(?lb x) ≤ ?cos x"
proof -
have "-pi ≤ ?x2" and "?x2 ≤ pi"
using pi_ge_two ‹0 ≤ real_of_float x› ‹x ≤ pi› by auto
from lb_half[OF lb_half[OF lb this] ‹-pi ≤ x› ‹x ≤ pi›, unfolded eq_4]
show ?thesis
unfolding lb_cos_def[where x=x] if_not_P[OF ‹¬ x < 0›]
if_not_P[OF ‹¬ x < Float 1 (- 1)›] if_not_P[OF ‹¬ x < 1›] Let_def .
qed
moreover have "?cos x ≤ (?ub x)"
proof -
have "-pi ≤ ?x2" and "?x2 ≤ pi"
using pi_ge_two ‹0 ≤ real_of_float x› ‹ x ≤ pi› by auto
from ub_half[OF ub_half[OF ub this] ‹-pi ≤ x› ‹x ≤ pi›, unfolded eq_4]
show ?thesis
unfolding ub_cos_def[where x=x] if_not_P[OF ‹¬ x < 0›]
if_not_P[OF ‹¬ x < Float 1 (- 1)›] if_not_P[OF ‹¬ x < 1›] Let_def .
qed
ultimately show ?thesis by auto
qed
qed
qed
lemma lb_cos_minus:
assumes "-pi ≤ x"
and "real_of_float x ≤ 0"
shows "cos (real_of_float(-x)) ∈ {(lb_cos prec (-x)) .. (ub_cos prec (-x))}"
proof -
have "0 ≤ real_of_float (-x)" and "(-x) ≤ pi"
using ‹-pi ≤ x› ‹real_of_float x ≤ 0› by auto
from lb_cos[OF this] show ?thesis .
qed
definition bnds_cos :: "nat ⇒ float ⇒ float ⇒ float * float" where
"bnds_cos prec lx ux = (let
lpi = float_round_down prec (lb_pi prec) ;
upi = float_round_up prec (ub_pi prec) ;
k = floor_fl (float_divr prec (lx + lpi) (2 * lpi)) ;
lx = float_plus_down prec lx (- k * 2 * (if k < 0 then lpi else upi)) ;
ux = float_plus_up prec ux (- k * 2 * (if k < 0 then upi else lpi))
in if - lpi ≤ lx ∧ ux ≤ 0 then (lb_cos prec (-lx), ub_cos prec (-ux))
else if 0 ≤ lx ∧ ux ≤ lpi then (lb_cos prec ux, ub_cos prec lx)
else if - lpi ≤ lx ∧ ux ≤ lpi then (min (lb_cos prec (-lx)) (lb_cos prec ux), Float 1 0)
else if 0 ≤ lx ∧ ux ≤ 2 * lpi then (Float (- 1) 0, max (ub_cos prec lx) (ub_cos prec (- (ux - 2 * lpi))))
else if -2 * lpi ≤ lx ∧ ux ≤ 0 then (Float (- 1) 0, max (ub_cos prec (lx + 2 * lpi)) (ub_cos prec (-ux)))
else (Float (- 1) 0, Float 1 0))"
lemma floor_int: obtains k :: int where "real_of_int k = (floor_fl f)"
by (simp add: floor_fl_def)
lemma cos_periodic_nat[simp]:
fixes n :: nat
shows "cos (x + n * (2 * pi)) = cos x"
proof (induct n arbitrary: x)
case 0
then show ?case by simp
next
case (Suc n)
have split_pi_off: "x + (Suc n) * (2 * pi) = (x + n * (2 * pi)) + 2 * pi"
unfolding Suc_eq_plus1 of_nat_add of_int_1 distrib_right by auto
show ?case
unfolding split_pi_off using Suc by auto
qed
lemma cos_periodic_int[simp]:
fixes i :: int
shows "cos (x + i * (2 * pi)) = cos x"
proof (cases "0 ≤ i")
case True
hence i_nat: "real_of_int i = nat i" by auto
show ?thesis
unfolding i_nat by auto
next
case False
hence i_nat: "i = - real (nat (-i))" by auto
have "cos x = cos (x + i * (2 * pi) - i * (2 * pi))"
by auto
also have "… = cos (x + i * (2 * pi))"
unfolding i_nat mult_minus_left diff_minus_eq_add by (rule cos_periodic_nat)
finally show ?thesis by auto
qed
lemma bnds_cos: "∀(x::real) lx ux. (l, u) =
bnds_cos prec lx ux ∧ x ∈ {lx .. ux} ⟶ l ≤ cos x ∧ cos x ≤ u"
proof (rule allI | rule impI | erule conjE)+
fix x :: real
fix lx ux
assume bnds: "(l, u) = bnds_cos prec lx ux" and x: "x ∈ {lx .. ux}"
let ?lpi = "float_round_down prec (lb_pi prec)"
let ?upi = "float_round_up prec (ub_pi prec)"
let ?k = "floor_fl (float_divr prec (lx + ?lpi) (2 * ?lpi))"
let ?lx2 = "(- ?k * 2 * (if ?k < 0 then ?lpi else ?upi))"
let ?ux2 = "(- ?k * 2 * (if ?k < 0 then ?upi else ?lpi))"
let ?lx = "float_plus_down prec lx ?lx2"
let ?ux = "float_plus_up prec ux ?ux2"
obtain k :: int where k: "k = real_of_float ?k"
by (rule floor_int)
have upi: "pi ≤ ?upi" and lpi: "?lpi ≤ pi"
using float_round_up[of "ub_pi prec" prec] pi_boundaries[of prec]
float_round_down[of prec "lb_pi prec"]
by auto
hence "lx + ?lx2 ≤ x - k * (2 * pi) ∧ x - k * (2 * pi) ≤ ux + ?ux2"
using x
by (cases "k = 0")
(auto intro!: add_mono
simp add: k [symmetric] uminus_add_conv_diff [symmetric]
simp del: uminus_add_conv_diff)
hence "?lx ≤ x - k * (2 * pi) ∧ x - k * (2 * pi) ≤ ?ux"
by (auto intro!: float_plus_down_le float_plus_up_le)
note lx = this[THEN conjunct1] and ux = this[THEN conjunct2]
hence lx_less_ux: "?lx ≤ real_of_float ?ux" by (rule order_trans)
{ assume "- ?lpi ≤ ?lx" and x_le_0: "x - k * (2 * pi) ≤ 0"
with lpi[THEN le_imp_neg_le] lx
have pi_lx: "- pi ≤ ?lx" and lx_0: "real_of_float ?lx ≤ 0"
by simp_all
have "(lb_cos prec (- ?lx)) ≤ cos (real_of_float (- ?lx))"
using lb_cos_minus[OF pi_lx lx_0] by simp
also have "… ≤ cos (x + (-k) * (2 * pi))"
using cos_monotone_minus_pi_0'[OF pi_lx lx x_le_0]
by (simp only: uminus_float.rep_eq of_int_minus
cos_minus mult_minus_left) simp
finally have "(lb_cos prec (- ?lx)) ≤ cos x"
unfolding cos_periodic_int . }
note negative_lx = this
{ assume "0 ≤ ?lx" and pi_x: "x - k * (2 * pi) ≤ pi"
with lx
have pi_lx: "?lx ≤ pi" and lx_0: "0 ≤ real_of_float ?lx"
by auto
have "cos (x + (-k) * (2 * pi)) ≤ cos ?lx"
using cos_monotone_0_pi_le[OF lx_0 lx pi_x]
by (simp only: of_int_minus
cos_minus mult_minus_left) simp
also have "… ≤ (ub_cos prec ?lx)"
using lb_cos[OF lx_0 pi_lx] by simp
finally have "cos x ≤ (ub_cos prec ?lx)"
unfolding cos_periodic_int . }
note positive_lx = this
{ assume pi_x: "- pi ≤ x - k * (2 * pi)" and "?ux ≤ 0"
with ux
have pi_ux: "- pi ≤ ?ux" and ux_0: "real_of_float ?ux ≤ 0"
by simp_all
have "cos (x + (-k) * (2 * pi)) ≤ cos (real_of_float (- ?ux))"
using cos_monotone_minus_pi_0'[OF pi_x ux ux_0]
by (simp only: uminus_float.rep_eq of_int_minus
cos_minus mult_minus_left) simp
also have "… ≤ (ub_cos prec (- ?ux))"
using lb_cos_minus[OF pi_ux ux_0, of prec] by simp
finally have "cos x ≤ (ub_cos prec (- ?ux))"
unfolding cos_periodic_int . }
note negative_ux = this
{ assume "?ux ≤ ?lpi" and x_ge_0: "0 ≤ x - k * (2 * pi)"
with lpi ux
have pi_ux: "?ux ≤ pi" and ux_0: "0 ≤ real_of_float ?ux"
by simp_all
have "(lb_cos prec ?ux) ≤ cos ?ux"
using lb_cos[OF ux_0 pi_ux] by simp
also have "… ≤ cos (x + (-k) * (2 * pi))"
using cos_monotone_0_pi_le[OF x_ge_0 ux pi_ux]
by (simp only: of_int_minus
cos_minus mult_minus_left) simp
finally have "(lb_cos prec ?ux) ≤ cos x"
unfolding cos_periodic_int . }
note positive_ux = this
show "l ≤ cos x ∧ cos x ≤ u"
proof (cases "- ?lpi ≤ ?lx ∧ ?ux ≤ 0")
case True
with bnds have l: "l = lb_cos prec (-?lx)" and u: "u = ub_cos prec (-?ux)"
by (auto simp add: bnds_cos_def Let_def)
from True lpi[THEN le_imp_neg_le] lx ux
have "- pi ≤ x - k * (2 * pi)" and "x - k * (2 * pi) ≤ 0"
by auto
with True negative_ux negative_lx show ?thesis
unfolding l u by simp
next
case 1: False
show ?thesis
proof (cases "0 ≤ ?lx ∧ ?ux ≤ ?lpi")
case True with bnds 1
have l: "l = lb_cos prec ?ux"
and u: "u = ub_cos prec ?lx"
by (auto simp add: bnds_cos_def Let_def)
from True lpi lx ux
have "0 ≤ x - k * (2 * pi)" and "x - k * (2 * pi) ≤ pi"
by auto
with True positive_ux positive_lx show ?thesis
unfolding l u by simp
next
case 2: False
show ?thesis
proof (cases "- ?lpi ≤ ?lx ∧ ?ux ≤ ?lpi")
case Cond: True
with bnds 1 2 have l: "l = min (lb_cos prec (-?lx)) (lb_cos prec ?ux)"
and u: "u = Float 1 0"
by (auto simp add: bnds_cos_def Let_def)
show ?thesis
unfolding u l using negative_lx positive_ux Cond
by (cases "x - k * (2 * pi) < 0") (auto simp add: real_of_float_min)
next
case 3: False
show ?thesis
proof (cases "0 ≤ ?lx ∧ ?ux ≤ 2 * ?lpi")
case Cond: True
with bnds 1 2 3
have l: "l = Float (- 1) 0"
and u: "u = max (ub_cos prec ?lx) (ub_cos prec (- (?ux - 2 * ?lpi)))"
by (auto simp add: bnds_cos_def Let_def)
have "cos x ≤ real_of_float u"
proof (cases "x - k * (2 * pi) < pi")
case True
hence "x - k * (2 * pi) ≤ pi" by simp
from positive_lx[OF Cond[THEN conjunct1] this] show ?thesis
unfolding u by (simp add: real_of_float_max)
next
case False
hence "pi ≤ x - k * (2 * pi)" by simp
hence pi_x: "- pi ≤ x - k * (2 * pi) - 2 * pi" by simp
have "?ux ≤ 2 * pi"
using Cond lpi by auto
hence "x - k * (2 * pi) - 2 * pi ≤ 0"
using ux by simp
have ux_0: "real_of_float (?ux - 2 * ?lpi) ≤ 0"
using Cond by auto
from 2 and Cond have "¬ ?ux ≤ ?lpi" by auto
hence "- ?lpi ≤ ?ux - 2 * ?lpi" by auto
hence pi_ux: "- pi ≤ (?ux - 2 * ?lpi)"
using lpi[THEN le_imp_neg_le] by auto
have x_le_ux: "x - k * (2 * pi) - 2 * pi ≤ (?ux - 2 * ?lpi)"
using ux lpi by auto
have "cos x = cos (x + (-k) * (2 * pi) + (-1::int) * (2 * pi))"
unfolding cos_periodic_int ..
also have "… ≤ cos ((?ux - 2 * ?lpi))"
using cos_monotone_minus_pi_0'[OF pi_x x_le_ux ux_0]
by (simp only: minus_float.rep_eq of_int_minus of_int_1
mult_minus_left mult_1_left) simp
also have "… = cos ((- (?ux - 2 * ?lpi)))"
unfolding uminus_float.rep_eq cos_minus ..
also have "… ≤ (ub_cos prec (- (?ux - 2 * ?lpi)))"
using lb_cos_minus[OF pi_ux ux_0] by simp
finally show ?thesis unfolding u by (simp add: real_of_float_max)
qed
thus ?thesis unfolding l by auto
next
case 4: False
show ?thesis
proof (cases "-2 * ?lpi ≤ ?lx ∧ ?ux ≤ 0")
case Cond: True
with bnds 1 2 3 4 have l: "l = Float (- 1) 0"
and u: "u = max (ub_cos prec (?lx + 2 * ?lpi)) (ub_cos prec (-?ux))"
by (auto simp add: bnds_cos_def Let_def)
have "cos x ≤ u"
proof (cases "-pi < x - k * (2 * pi)")
case True
hence "-pi ≤ x - k * (2 * pi)" by simp
from negative_ux[OF this Cond[THEN conjunct2]] show ?thesis
unfolding u by (simp add: real_of_float_max)
next
case False
hence "x - k * (2 * pi) ≤ -pi" by simp
hence pi_x: "x - k * (2 * pi) + 2 * pi ≤ pi" by simp
have "-2 * pi ≤ ?lx" using Cond lpi by auto
hence "0 ≤ x - k * (2 * pi) + 2 * pi" using lx by simp
have lx_0: "0 ≤ real_of_float (?lx + 2 * ?lpi)"
using Cond lpi by auto
from 1 and Cond have "¬ -?lpi ≤ ?lx" by auto
hence "?lx + 2 * ?lpi ≤ ?lpi" by auto
hence pi_lx: "(?lx + 2 * ?lpi) ≤ pi"
using lpi[THEN le_imp_neg_le] by auto
have lx_le_x: "(?lx + 2 * ?lpi) ≤ x - k * (2 * pi) + 2 * pi"
using lx lpi by auto
have "cos x = cos (x + (-k) * (2 * pi) + (1 :: int) * (2 * pi))"
unfolding cos_periodic_int ..
also have "… ≤ cos ((?lx + 2 * ?lpi))"
using cos_monotone_0_pi_le[OF lx_0 lx_le_x pi_x]
by (simp only: minus_float.rep_eq of_int_minus of_int_1
mult_minus_left mult_1_left) simp
also have "… ≤ (ub_cos prec (?lx + 2 * ?lpi))"
using lb_cos[OF lx_0 pi_lx] by simp
finally show ?thesis unfolding u by (simp add: real_of_float_max)
qed
thus ?thesis unfolding l by auto
next
case False
with bnds 1 2 3 4 show ?thesis
by (auto simp add: bnds_cos_def Let_def)
qed
qed
qed
qed
qed
qed
lemma bnds_cos_lower: "⋀x. real_of_float xl ≤ x ⟹ x ≤ real_of_float xu ⟹ cos x ≤ y ⟹ real_of_float (fst (bnds_cos prec xl xu)) ≤ y"
and bnds_cos_upper: "⋀x. real_of_float xl ≤ x ⟹ x ≤ real_of_float xu ⟹ y ≤ cos x ⟹ y ≤ real_of_float (snd (bnds_cos prec xl xu))"
for xl xu::float and y::real
using bnds_cos[of "fst (bnds_cos prec xl xu)" "snd (bnds_cos prec xl xu)" prec]
by force+
lift_definition cos_float_interval :: "nat ⇒ float interval ⇒ float interval"
is "λprec. λ(lx, ux). bnds_cos prec lx ux"
using bnds_cos
by auto (metis (full_types) order_refl order_trans)
lemma lower_cos_float_interval: "lower (cos_float_interval p x) = fst (bnds_cos p (lower x) (upper x))"
by transfer auto
lemma upper_cos_float_interval: "upper (cos_float_interval p x) = snd (bnds_cos p (lower x) (upper x))"
by transfer auto
lemma cos_float_interval:
"cos ` set_of (real_interval x) ⊆ set_of (real_interval (cos_float_interval p x))"
by (auto simp: set_of_eq bnds_cos_lower bnds_cos_upper lower_cos_float_interval
upper_cos_float_interval)
lemma cos_float_intervalI: "cos x ∈⇩r cos_float_interval p X" if "x ∈⇩r X"
using cos_float_interval[of X p] that
by auto
section "Exponential function"
subsection "Compute the series of the exponential function"
fun ub_exp_horner :: "nat ⇒ nat ⇒ nat ⇒ nat ⇒ float ⇒ float"
and lb_exp_horner :: "nat ⇒ nat ⇒ nat ⇒ nat ⇒ float ⇒ float"
where
"ub_exp_horner prec 0 i k x = 0" |
"ub_exp_horner prec (Suc n) i k x = float_plus_up prec
(rapprox_rat prec 1 (int k)) (float_round_up prec (x * lb_exp_horner prec n (i + 1) (k * i) x))" |
"lb_exp_horner prec 0 i k x = 0" |
"lb_exp_horner prec (Suc n) i k x = float_plus_down prec
(lapprox_rat prec 1 (int k)) (float_round_down prec (x * ub_exp_horner prec n (i + 1) (k * i) x))"
lemma bnds_exp_horner:
assumes "real_of_float x ≤ 0"
shows "exp x ∈ {lb_exp_horner prec (get_even n) 1 1 x .. ub_exp_horner prec (get_odd n) 1 1 x}"
proof -
have f_eq: "fact (Suc n) = fact n * ((λi::nat. i + 1) ^^ n) 1" for n
proof -
have F: "⋀ m. ((λi. i + 1) ^^ n) m = n + m"
by (induct n) auto
show ?thesis
unfolding F by auto
qed
note bounds = horner_bounds_nonpos[where f="fact" and lb="lb_exp_horner prec" and ub="ub_exp_horner prec" and j'=0 and s=1,
OF assms f_eq lb_exp_horner.simps ub_exp_horner.simps]
have "lb_exp_horner prec (get_even n) 1 1 x ≤ exp x"
proof -
have "lb_exp_horner prec (get_even n) 1 1 x ≤ (∑j = 0..<get_even n. 1 / (fact j) * real_of_float x ^ j)"
using bounds(1) by auto
also have "… ≤ exp x"
proof -
obtain t where "¦t¦ ≤ ¦real_of_float x¦" and "exp x = (∑m = 0..<get_even n. real_of_float x ^ m / (fact m)) + exp t / (fact (get_even n)) * (real_of_float x) ^ (get_even n)"
using Maclaurin_exp_le unfolding atLeast0LessThan by blast
moreover have "0 ≤ exp t / (fact (get_even n)) * (real_of_float x) ^ (get_even n)"
by (auto simp: zero_le_even_power)
ultimately show ?thesis using get_odd exp_gt_zero by auto
qed
finally show ?thesis .
qed
moreover
have "exp x ≤ ub_exp_horner prec (get_odd n) 1 1 x"
proof -
have x_less_zero: "real_of_float x ^ get_odd n ≤ 0"
proof (cases "real_of_float x = 0")
case True
have "(get_odd n) ≠ 0" using get_odd[THEN odd_pos] by auto
thus ?thesis unfolding True power_0_left by auto
next
case False hence "real_of_float x < 0" using ‹real_of_float x ≤ 0› by auto
show ?thesis by (rule less_imp_le, auto simp add: ‹real_of_float x < 0›)
qed
obtain t where "¦t¦ ≤ ¦real_of_float x¦"
and "exp x = (∑m = 0..<get_odd n. (real_of_float x) ^ m / (fact m)) + exp t / (fact (get_odd n)) * (real_of_float x) ^ (get_odd n)"
using Maclaurin_exp_le unfolding atLeast0LessThan by blast
moreover have "exp t / (fact (get_odd n)) * (real_of_float x) ^ (get_odd n) ≤ 0"
by (auto intro!: mult_nonneg_nonpos divide_nonpos_pos simp add: x_less_zero)
ultimately have "exp x ≤ (∑j = 0..<get_odd n. 1 / (fact j) * real_of_float x ^ j)"
using get_odd exp_gt_zero by auto
also have "… ≤ ub_exp_horner prec (get_odd n) 1 1 x"
using bounds(2) by auto
finally show ?thesis .
qed
ultimately show ?thesis by auto
qed
lemma ub_exp_horner_nonneg: "real_of_float x ≤ 0 ⟹
0 ≤ real_of_float (ub_exp_horner prec (get_odd n) (Suc 0) (Suc 0) x)"
using bnds_exp_horner[of x prec n]
by (intro order_trans[OF exp_ge_zero]) auto
subsection "Compute the exponential function on the entire domain"
function ub_exp :: "nat ⇒ float ⇒ float" and lb_exp :: "nat ⇒ float ⇒ float" where
"lb_exp prec x =
(if 0 < x then float_divl prec 1 (ub_exp prec (-x))
else
let
horner = (λ x. let y = lb_exp_horner prec (get_even (prec + 2)) 1 1 x in
if y ≤ 0 then Float 1 (- 2) else y)
in
if x < - 1 then
power_down_fl prec (horner (float_divl prec x (- floor_fl x))) (nat (- int_floor_fl x))
else horner x)" |
"ub_exp prec x =
(if 0 < x then float_divr prec 1 (lb_exp prec (-x))
else if x < - 1 then
power_up_fl prec
(ub_exp_horner prec (get_odd (prec + 2)) 1 1
(float_divr prec x (- floor_fl x))) (nat (- int_floor_fl x))
else ub_exp_horner prec (get_odd (prec + 2)) 1 1 x)"
by pat_completeness auto
termination
by (relation "measure (λ v. let (prec, x) = case_sum id id v in (if 0 < x then 1 else 0))") auto
lemma exp_m1_ge_quarter: "(1 / 4 :: real) ≤ exp (- 1)"
proof -
have eq4: "4 = Suc (Suc (Suc (Suc 0)))" by auto
have "1 / 4 = (Float 1 (- 2))"
unfolding Float_num by auto
also have "… ≤ lb_exp_horner 3 (get_even 3) 1 1 (- 1)"
by (subst less_eq_float.rep_eq [symmetric]) code_simp
also have "… ≤ exp (- 1 :: float)"
using bnds_exp_horner[where x="- 1"] by auto
finally show ?thesis
by simp
qed
lemma lb_exp_pos:
assumes "¬ 0 < x"
shows "0 < lb_exp prec x"
proof -
let "?lb_horner x" = "lb_exp_horner prec (get_even (prec + 2)) 1 1 x"
let "?horner x" = "let y = ?lb_horner x in if y ≤ 0 then Float 1 (- 2) else y"
have pos_horner: "0 < ?horner x" for x
unfolding Let_def by (cases "?lb_horner x ≤ 0") auto
moreover have "0 < real_of_float ((?horner x) ^ num)" for x :: float and num :: nat
proof -
have "0 < real_of_float (?horner x) ^ num" using ‹0 < ?horner x› by simp
also have "… = (?horner x) ^ num" by auto
finally show ?thesis .
qed
ultimately show ?thesis
unfolding lb_exp.simps if_not_P[OF ‹¬ 0 < x›] Let_def
by (cases "floor_fl x", cases "x < - 1")
(auto simp: real_power_up_fl real_power_down_fl intro!: power_up_less power_down_pos)
qed
lemma exp_boundaries':
assumes "x ≤ 0"
shows "exp x ∈ { (lb_exp prec x) .. (ub_exp prec x)}"
proof -
let "?lb_exp_horner x" = "lb_exp_horner prec (get_even (prec + 2)) 1 1 x"
let "?ub_exp_horner x" = "ub_exp_horner prec (get_odd (prec + 2)) 1 1 x"
have "real_of_float x ≤ 0" and "¬ x > 0"
using ‹x ≤ 0› by auto
show ?thesis
proof (cases "x < - 1")
case False
hence "- 1 ≤ real_of_float x" by auto
show ?thesis
proof (cases "?lb_exp_horner x ≤ 0")
case True
from ‹¬ x < - 1›
have "- 1 ≤ real_of_float x" by auto
hence "exp (- 1) ≤ exp x"
unfolding exp_le_cancel_iff .
from order_trans[OF exp_m1_ge_quarter this] have "Float 1 (- 2) ≤ exp x"
unfolding Float_num .
with True show ?thesis
using bnds_exp_horner ‹real_of_float x ≤ 0› ‹¬ x > 0› ‹¬ x < - 1› by auto
next
case False
thus ?thesis
using bnds_exp_horner ‹real_of_float x ≤ 0› ‹¬ x > 0› ‹¬ x < - 1› by (auto simp add: Let_def)
qed
next
case True
let ?num = "nat (- int_floor_fl x)"
have "real_of_int (int_floor_fl x) < - 1"
using int_floor_fl[of x] ‹x < - 1› by simp
hence "real_of_int (int_floor_fl x) < 0" by simp
hence "int_floor_fl x < 0" by auto
hence "1 ≤ - int_floor_fl x" by auto
hence "0 < nat (- int_floor_fl x)" by auto
hence "0 < ?num" by auto
hence "real ?num ≠ 0" by auto
have num_eq: "real ?num = - int_floor_fl x"
using ‹0 < nat (- int_floor_fl x)› by auto
have "0 < - int_floor_fl x"
using ‹0 < ?num›[unfolded of_nat_less_iff[symmetric]] by simp
hence "real_of_int (int_floor_fl x) < 0"
unfolding less_float_def by auto
have fl_eq: "real_of_int (- int_floor_fl x) = real_of_float (- floor_fl x)"
by (simp add: floor_fl_def int_floor_fl_def)
from ‹0 < - int_floor_fl x› have "0 ≤ real_of_float (- floor_fl x)"
by (simp add: floor_fl_def int_floor_fl_def)
from ‹real_of_int (int_floor_fl x) < 0› have "real_of_float (floor_fl x) < 0"
by (simp add: floor_fl_def int_floor_fl_def)
have "exp x ≤ ub_exp prec x"
proof -
have div_less_zero: "real_of_float (float_divr prec x (- floor_fl x)) ≤ 0"
using float_divr_nonpos_pos_upper_bound[OF ‹real_of_float x ≤ 0› ‹0 ≤ real_of_float (- floor_fl x)›]
unfolding less_eq_float_def zero_float.rep_eq .
have "exp x = exp (?num * (x / ?num))"
using ‹real ?num ≠ 0› by auto
also have "… = exp (x / ?num) ^ ?num"
unfolding exp_of_nat_mult ..
also have "… ≤ exp (float_divr prec x (- floor_fl x)) ^ ?num"
unfolding num_eq fl_eq
by (rule power_mono, rule exp_le_cancel_iff[THEN iffD2], rule float_divr) auto
also have "… ≤ (?ub_exp_horner (float_divr prec x (- floor_fl x))) ^ ?num"
unfolding real_of_float_power
by (rule power_mono, rule bnds_exp_horner[OF div_less_zero, unfolded atLeastAtMost_iff, THEN conjunct2], auto)
also have "… ≤ real_of_float (power_up_fl prec (?ub_exp_horner (float_divr prec x (- floor_fl x))) ?num)"
by (auto simp add: real_power_up_fl intro!: power_up ub_exp_horner_nonneg div_less_zero)
finally show ?thesis
unfolding ub_exp.simps if_not_P[OF ‹¬ 0 < x›] if_P[OF ‹x < - 1›] floor_fl_def Let_def .
qed
moreover
have "lb_exp prec x ≤ exp x"
proof -
let ?divl = "float_divl prec x (- floor_fl x)"
let ?horner = "?lb_exp_horner ?divl"
show ?thesis
proof (cases "?horner ≤ 0")
case False
hence "0 ≤ real_of_float ?horner" by auto
have div_less_zero: "real_of_float (float_divl prec x (- floor_fl x)) ≤ 0"
using ‹real_of_float (floor_fl x) < 0› ‹real_of_float x ≤ 0›
by (auto intro!: order_trans[OF float_divl] divide_nonpos_neg)
have "(?lb_exp_horner (float_divl prec x (- floor_fl x))) ^ ?num ≤
exp (float_divl prec x (- floor_fl x)) ^ ?num"
using ‹0 ≤ real_of_float ?horner›[unfolded floor_fl_def[symmetric]]
bnds_exp_horner[OF div_less_zero, unfolded atLeastAtMost_iff, THEN conjunct1]
by (auto intro!: power_mono)
also have "… ≤ exp (x / ?num) ^ ?num"
unfolding num_eq fl_eq
using float_divl by (auto intro!: power_mono simp del: uminus_float.rep_eq)
also have "… = exp (?num * (x / ?num))"
unfolding exp_of_nat_mult ..
also have "… = exp x"
using ‹real ?num ≠ 0› by auto
finally show ?thesis
using False
unfolding lb_exp.simps if_not_P[OF ‹¬ 0 < x›] if_P[OF ‹x < - 1›]
int_floor_fl_def Let_def if_not_P[OF False]
by (auto simp: real_power_down_fl intro!: power_down_le)
next
case True
have "power_down_fl prec (Float 1 (- 2)) ?num ≤ (Float 1 (- 2)) ^ ?num"
by (metis Float_le_zero_iff less_imp_le linorder_not_less
not_numeral_le_zero numeral_One power_down_fl)
then have "power_down_fl prec (Float 1 (- 2)) ?num ≤ real_of_float (Float 1 (- 2)) ^ ?num"
by simp
also
have "real_of_float (floor_fl x) ≠ 0" and "real_of_float (floor_fl x) ≤ 0"
using ‹real_of_float (floor_fl x) < 0› by auto
from divide_right_mono_neg[OF floor_fl[of x] ‹real_of_float (floor_fl x) ≤ 0›, unfolded divide_self[OF ‹real_of_float (floor_fl x) ≠ 0›]]
have "- 1 ≤ x / (- floor_fl x)"
unfolding minus_float.rep_eq by auto
from order_trans[OF exp_m1_ge_quarter this[unfolded exp_le_cancel_iff[where x="- 1", symmetric]]]
have "Float 1 (- 2) ≤ exp (x / (- floor_fl x))"
unfolding Float_num .
hence "real_of_float (Float 1 (- 2)) ^ ?num ≤ exp (x / (- floor_fl x)) ^ ?num"
by (metis Float_num(5) power_mono zero_le_divide_1_iff zero_le_numeral)
also have "… = exp x"
unfolding num_eq fl_eq exp_of_nat_mult[symmetric]
using ‹real_of_float (floor_fl x) ≠ 0› by auto
finally show ?thesis
unfolding lb_exp.simps if_not_P[OF ‹¬ 0 < x›] if_P[OF ‹x < - 1›]
int_floor_fl_def Let_def if_P[OF True] real_of_float_power .
qed
qed
ultimately show ?thesis by auto
qed
qed
lemma exp_boundaries: "exp x ∈ { lb_exp prec x .. ub_exp prec x }"
proof -
show ?thesis
proof (cases "0 < x")
case False
hence "x ≤ 0" by auto
from exp_boundaries'[OF this] show ?thesis .
next
case True
hence "-x ≤ 0" by auto
have "lb_exp prec x ≤ exp x"
proof -
from exp_boundaries'[OF ‹-x ≤ 0›]
have ub_exp: "exp (- real_of_float x) ≤ ub_exp prec (-x)"
unfolding atLeastAtMost_iff minus_float.rep_eq by auto
have "float_divl prec 1 (ub_exp prec (-x)) ≤ 1 / ub_exp prec (-x)"
using float_divl[where x=1] by auto
also have "… ≤ exp x"
using ub_exp[unfolded inverse_le_iff_le[OF order_less_le_trans[OF exp_gt_zero ub_exp]
exp_gt_zero, symmetric]]
unfolding exp_minus nonzero_inverse_inverse_eq[OF exp_not_eq_zero] inverse_eq_divide
by auto
finally show ?thesis
unfolding lb_exp.simps if_P[OF True] .
qed
moreover
have "exp x ≤ ub_exp prec x"
proof -
have "¬ 0 < -x" using ‹0 < x› by auto
from exp_boundaries'[OF ‹-x ≤ 0›]
have lb_exp: "lb_exp prec (-x) ≤ exp (- real_of_float x)"
unfolding atLeastAtMost_iff minus_float.rep_eq by auto
have "exp x ≤ (1 :: float) / lb_exp prec (-x)"
using lb_exp lb_exp_pos[OF ‹¬ 0 < -x›, of prec]
by (simp del: lb_exp.simps add: exp_minus field_simps)
also have "… ≤ float_divr prec 1 (lb_exp prec (-x))"
using float_divr .
finally show ?thesis
unfolding ub_exp.simps if_P[OF True] .
qed
ultimately show ?thesis
by auto
qed
qed
lemma bnds_exp: "∀(x::real) lx ux. (l, u) =
(lb_exp prec lx, ub_exp prec ux) ∧ x ∈ {lx .. ux} ⟶ l ≤ exp x ∧ exp x ≤ u"
proof (rule allI, rule allI, rule allI, rule impI)
fix x :: real and lx ux
assume "(l, u) = (lb_exp prec lx, ub_exp prec ux) ∧ x ∈ {lx .. ux}"
hence l: "lb_exp prec lx = l " and u: "ub_exp prec ux = u" and x: "x ∈ {lx .. ux}"
by auto
show "l ≤ exp x ∧ exp x ≤ u"
proof
show "l ≤ exp x"
proof -
from exp_boundaries[of lx prec, unfolded l]
have "l ≤ exp lx" by (auto simp del: lb_exp.simps)
also have "… ≤ exp x" using x by auto
finally show ?thesis .
qed
show "exp x ≤ u"
proof -
have "exp x ≤ exp ux" using x by auto
also have "… ≤ u" using exp_boundaries[of ux prec, unfolded u] by (auto simp del: ub_exp.simps)
finally show ?thesis .
qed
qed
qed
lemmas [simp del] = lb_exp.simps ub_exp.simps
lemma lb_exp: "exp x ≤ y ⟹ lb_exp prec x ≤ y"
and ub_exp: "y ≤ exp x ⟹ y ≤ ub_exp prec x"
for x::float and y::real using exp_boundaries[of x prec] by auto
lift_definition exp_float_interval :: "nat ⇒ float interval ⇒ float interval"
is "λprec. λ(lx, ux). (lb_exp prec lx, ub_exp prec ux)"
by (auto simp: lb_exp ub_exp)
lemma lower_exp_float_interval: "lower (exp_float_interval p x) = lb_exp p (lower x)"
by transfer auto
lemma upper_exp_float_interval: "upper (exp_float_interval p x) = ub_exp p (upper x)"
by transfer auto
lemma exp_float_interval:
"exp ` set_of (real_interval x) ⊆ set_of (real_interval (exp_float_interval p x))"
using exp_boundaries
by (auto simp: set_of_eq lower_exp_float_interval upper_exp_float_interval lb_exp ub_exp)
lemma exp_float_intervalI:
"exp x ∈⇩r exp_float_interval p X" if "x ∈⇩r X"
using exp_float_interval[of X p] that
by auto
section "Logarithm"
subsection "Compute the logarithm series"
fun ub_ln_horner :: "nat ⇒ nat ⇒ nat ⇒ float ⇒ float"
and lb_ln_horner :: "nat ⇒ nat ⇒ nat ⇒ float ⇒ float" where
"ub_ln_horner prec 0 i x = 0" |
"ub_ln_horner prec (Suc n) i x = float_plus_up prec
(rapprox_rat prec 1 (int i)) (- float_round_down prec (x * lb_ln_horner prec n (Suc i) x))" |
"lb_ln_horner prec 0 i x = 0" |
"lb_ln_horner prec (Suc n) i x = float_plus_down prec
(lapprox_rat prec 1 (int i)) (- float_round_up prec (x * ub_ln_horner prec n (Suc i) x))"
lemma ln_bounds:
assumes "0 ≤ x"
and "x < 1"
shows "(∑i=0..<2*n. (- 1) ^ i * (1 / real (i + 1)) * x ^ (Suc i)) ≤ ln (x + 1)" (is "?lb")
and "ln (x + 1) ≤ (∑i=0..<2*n + 1. (- 1) ^ i * (1 / real (i + 1)) * x ^ (Suc i))" (is "?ub")
proof -
let "?a n" = "(1/real (n +1)) * x ^ (Suc n)"
have ln_eq: "(∑ i. (- 1) ^ i * ?a i) = ln (x + 1)"
using ln_series[of "x + 1"] ‹0 ≤ x› ‹x < 1› by auto
have "norm x < 1" using assms by auto
have "?a ⇢ 0" unfolding Suc_eq_plus1[symmetric] inverse_eq_divide[symmetric]
using tendsto_mult[OF LIMSEQ_inverse_real_of_nat LIMSEQ_Suc[OF LIMSEQ_power_zero[OF ‹norm x < 1›]]] by auto
have "0 ≤ ?a n" for n
by (rule mult_nonneg_nonneg) (auto simp: ‹0 ≤ x›)
have "?a (Suc n) ≤ ?a n" for n
unfolding inverse_eq_divide[symmetric]
proof (rule mult_mono)
show "0 ≤ x ^ Suc (Suc n)"
by (auto simp add: ‹0 ≤ x›)
have "x ^ Suc (Suc n) ≤ x ^ Suc n * 1"
unfolding power_Suc2 mult.assoc[symmetric]
by (rule mult_left_mono, fact less_imp_le[OF ‹x < 1›]) (auto simp: ‹0 ≤ x›)
thus "x ^ Suc (Suc n) ≤ x ^ Suc n" by auto
qed auto
from summable_Leibniz'(2,4)[OF ‹?a ⇢ 0› ‹⋀n. 0 ≤ ?a n›, OF ‹⋀n. ?a (Suc n) ≤ ?a n›, unfolded ln_eq]
show ?lb and ?ub
unfolding atLeast0LessThan by auto
qed
lemma ln_float_bounds:
assumes "0 ≤ real_of_float x"
and "real_of_float x < 1"
shows "x * lb_ln_horner prec (get_even n) 1 x ≤ ln (x + 1)" (is "?lb ≤ ?ln")
and "ln (x + 1) ≤ x * ub_ln_horner prec (get_odd n) 1 x" (is "?ln ≤ ?ub")
proof -
obtain ev where ev: "get_even n = 2 * ev" using get_even_double ..
obtain od where od: "get_odd n = 2 * od + 1" using get_odd_double ..
let "?s n" = "(- 1) ^ n * (1 / real (1 + n)) * (real_of_float x)^(Suc n)"
have "?lb ≤ sum ?s {0 ..< 2 * ev}"
unfolding power_Suc2 mult.assoc[symmetric] times_float.rep_eq sum_distrib_right[symmetric]
unfolding mult.commute[of "real_of_float x"] ev
using horner_bounds(1)[where G="λ i k. Suc k" and F="λx. x" and f="λx. x"
and lb="λn i k x. lb_ln_horner prec n k x"
and ub="λn i k x. ub_ln_horner prec n k x" and j'=1 and n="2*ev",
OF ‹0 ≤ real_of_float x› refl lb_ln_horner.simps ub_ln_horner.simps] ‹0 ≤ real_of_float x›
unfolding real_of_float_power
by (rule mult_right_mono)
also have "… ≤ ?ln"
using ln_bounds(1)[OF ‹0 ≤ real_of_float x› ‹real_of_float x < 1›] by auto
finally show "?lb ≤ ?ln" .
have "?ln ≤ sum ?s {0 ..< 2 * od + 1}"
using ln_bounds(2)[OF ‹0 ≤ real_of_float x› ‹real_of_float x < 1›] by auto
also have "… ≤ ?ub"
unfolding power_Suc2 mult.assoc[symmetric] times_float.rep_eq sum_distrib_right[symmetric]
unfolding mult.commute[of "real_of_float x"] od
using horner_bounds(2)[where G="λ i k. Suc k" and F="λx. x" and f="λx. x" and lb="λn i k x. lb_ln_horner prec n k x" and ub="λn i k x. ub_ln_horner prec n k x" and j'=1 and n="2 * od + 1",
OF ‹0 ≤ real_of_float x› refl lb_ln_horner.simps ub_ln_horner.simps] ‹0 ≤ real_of_float x›
unfolding real_of_float_power
by (rule mult_right_mono)
finally show "?ln ≤ ?ub" .
qed
lemma ln_add:
fixes x :: real
assumes "0 < x" and "0 < y"
shows "ln (x + y) = ln x + ln (1 + y / x)"
proof -
have "x ≠ 0" using assms by auto
have "x + y = x * (1 + y / x)"
unfolding distrib_left times_divide_eq_right nonzero_mult_div_cancel_left[OF ‹x ≠ 0›]
by auto
moreover
have "0 < y / x" using assms by auto
hence "0 < 1 + y / x" by auto
ultimately show ?thesis
using ln_mult assms by auto
qed
subsection "Compute the logarithm of 2"
definition ub_ln2 where "ub_ln2 prec = (let third = rapprox_rat (max prec 1) 1 3
in float_plus_up prec
((Float 1 (- 1) * ub_ln_horner prec (get_odd prec) 1 (Float 1 (- 1))))
(float_round_up prec (third * ub_ln_horner prec (get_odd prec) 1 third)))"
definition lb_ln2 where "lb_ln2 prec = (let third = lapprox_rat prec 1 3
in float_plus_down prec
((Float 1 (- 1) * lb_ln_horner prec (get_even prec) 1 (Float 1 (- 1))))
(float_round_down prec (third * lb_ln_horner prec (get_even prec) 1 third)))"
lemma ub_ln2: "ln 2 ≤ ub_ln2 prec" (is "?ub_ln2")
and lb_ln2: "lb_ln2 prec ≤ ln 2" (is "?lb_ln2")
proof -
let ?uthird = "rapprox_rat (max prec 1) 1 3"
let ?lthird = "lapprox_rat prec 1 3"
have ln2_sum: "ln 2 = ln (1/2 + 1) + ln (1 / 3 + 1::real)"
using ln_add[of "3 / 2" "1 / 2"] by auto
have lb3: "?lthird ≤ 1 / 3" using lapprox_rat[of prec 1 3] by auto
hence lb3_ub: "real_of_float ?lthird < 1" by auto
have lb3_lb: "0 ≤ real_of_float ?lthird" using lapprox_rat_nonneg[of 1 3] by auto
have ub3: "1 / 3 ≤ ?uthird" using rapprox_rat[of 1 3] by auto
hence ub3_lb: "0 ≤ real_of_float ?uthird" by auto
have lb2: "0 ≤ real_of_float (Float 1 (- 1))" and ub2: "real_of_float (Float 1 (- 1)) < 1"
unfolding Float_num by auto
have "0 ≤ (1::int)" and "0 < (3::int)" by auto
have ub3_ub: "real_of_float ?uthird < 1"
by (simp add: Float.compute_rapprox_rat Float.compute_lapprox_rat rapprox_posrat_less1)
have third_gt0: "(0 :: real) < 1 / 3 + 1" by auto
have uthird_gt0: "0 < real_of_float ?uthird + 1" using ub3_lb by auto
have lthird_gt0: "0 < real_of_float ?lthird + 1" using lb3_lb by auto
show ?ub_ln2
unfolding ub_ln2_def Let_def ln2_sum Float_num(4)[symmetric]
proof (rule float_plus_up_le, rule add_mono, fact ln_float_bounds(2)[OF lb2 ub2])
have "ln (1 / 3 + 1) ≤ ln (real_of_float ?uthird + 1)"
unfolding ln_le_cancel_iff[OF third_gt0 uthird_gt0] using ub3 by auto
also have "… ≤ ?uthird * ub_ln_horner prec (get_odd prec) 1 ?uthird"
using ln_float_bounds(2)[OF ub3_lb ub3_ub] .
also note float_round_up
finally show "ln (1 / 3 + 1) ≤ float_round_up prec (?uthird * ub_ln_horner prec (get_odd prec) 1 ?uthird)" .
qed
show ?lb_ln2
unfolding lb_ln2_def Let_def ln2_sum Float_num(4)[symmetric]
proof (rule float_plus_down_le, rule add_mono, fact ln_float_bounds(1)[OF lb2 ub2])
have "?lthird * lb_ln_horner prec (get_even prec) 1 ?lthird ≤ ln (real_of_float ?lthird + 1)"
using ln_float_bounds(1)[OF lb3_lb lb3_ub] .
note float_round_down_le[OF this]
also have "… ≤ ln (1 / 3 + 1)"
unfolding ln_le_cancel_iff[OF lthird_gt0 third_gt0]
using lb3 by auto
finally show "float_round_down prec (?lthird * lb_ln_horner prec (get_even prec) 1 ?lthird) ≤
ln (1 / 3 + 1)" .
qed
qed
subsection "Compute the logarithm in the entire domain"
function ub_ln :: "nat ⇒ float ⇒ float option" and lb_ln :: "nat ⇒ float ⇒ float option" where
"ub_ln prec x = (if x ≤ 0 then None
else if x < 1 then Some (- the (lb_ln prec (float_divl (max prec 1) 1 x)))
else let horner = λx. float_round_up prec (x * ub_ln_horner prec (get_odd prec) 1 x) in
if x ≤ Float 3 (- 1) then Some (horner (x - 1))
else if x < Float 1 1 then Some (float_round_up prec (horner (Float 1 (- 1)) + horner (x * rapprox_rat prec 2 3 - 1)))
else let l = bitlen (mantissa x) - 1 in
Some (float_plus_up prec (float_round_up prec (ub_ln2 prec * (Float (exponent x + l) 0))) (horner (Float (mantissa x) (- l) - 1))))" |
"lb_ln prec x = (if x ≤ 0 then None
else if x < 1 then Some (- the (ub_ln prec (float_divr prec 1 x)))
else let horner = λx. float_round_down prec (x * lb_ln_horner prec (get_even prec) 1 x) in
if x ≤ Float 3 (- 1) then Some (horner (x - 1))
else if x < Float 1 1 then Some (float_round_down prec (horner (Float 1 (- 1)) +
horner (max (x * lapprox_rat prec 2 3 - 1) 0)))
else let l = bitlen (mantissa x) - 1 in
Some (float_plus_down prec (float_round_down prec (lb_ln2 prec * (Float (exponent x + l) 0))) (horner (Float (mantissa x) (- l) - 1))))"
by pat_completeness auto
termination
proof (relation "measure (λ v. let (prec, x) = case_sum id id v in (if x < 1 then 1 else 0))", auto)
fix prec and x :: float
assume "¬ real_of_float x ≤ 0" and "real_of_float x < 1" and "real_of_float (float_divl (max prec (Suc 0)) 1 x) < 1"
hence "0 < real_of_float x" "1 ≤ max prec (Suc 0)" "real_of_float x < 1"
by auto
from float_divl_pos_less1_bound[OF ‹0 < real_of_float x› ‹real_of_float x < 1›[THEN less_imp_le] ‹1 ≤ max prec (Suc 0)›]
show False
using ‹real_of_float (float_divl (max prec (Suc 0)) 1 x) < 1› by auto
next
fix prec x
assume "¬ real_of_float x ≤ 0" and "real_of_float x < 1" and "real_of_float (float_divr prec 1 x) < 1"
hence "0 < x" by auto
from float_divr_pos_less1_lower_bound[OF ‹0 < x›, of prec] ‹real_of_float x < 1› show False
using ‹real_of_float (float_divr prec 1 x) < 1› by auto
qed
lemmas float_pos_eq_mantissa_pos = mantissa_pos_iff[symmetric]
lemma Float_pos_eq_mantissa_pos: "Float m e > 0 ⟷ m > 0"
using powr_gt_zero[of 2 "e"]
by (auto simp add: zero_less_mult_iff zero_float_def simp del: powr_gt_zero dest: less_zeroE)
lemma Float_representation_aux:
fixes m e
defines [THEN meta_eq_to_obj_eq]: "x ≡ Float m e"
assumes "x > 0"
shows "Float (exponent x + (bitlen (mantissa x) - 1)) 0 = Float (e + (bitlen m - 1)) 0" (is ?th1)
and "Float (mantissa x) (- (bitlen (mantissa x) - 1)) = Float m ( - (bitlen m - 1))" (is ?th2)
proof -
from assms have mantissa_pos: "m > 0" "mantissa x > 0"
using Float_pos_eq_mantissa_pos[of m e] float_pos_eq_mantissa_pos[of x] by simp_all
thus ?th1
using bitlen_Float[of m e] assms
by (auto simp add: zero_less_mult_iff intro!: arg_cong2[where f=Float])
have "x ≠ 0"
unfolding zero_float_def[symmetric] using ‹0 < x› by auto
from denormalize_shift[OF x_def this] obtain i where
i: "m = mantissa x * 2 ^ i" "e = exponent x - int i" .
have "2 powr (1 - (real_of_int (bitlen (mantissa x)) + real_of_int i)) =
2 powr (1 - (real_of_int (bitlen (mantissa x)))) * inverse (2 powr (real i))"
by (simp add: powr_minus[symmetric] powr_add[symmetric] field_simps)
hence "real_of_int (mantissa x) * 2 powr (1 - real_of_int (bitlen (mantissa x))) =
(real_of_int (mantissa x) * 2 ^ i) * 2 powr (1 - real_of_int (bitlen (mantissa x * 2 ^ i)))"
using ‹mantissa x > 0› by (simp add: powr_realpow)
then show ?th2
unfolding i
by (auto simp: real_of_float_eq)
qed
lemma compute_ln[code]:
fixes m e
defines "x ≡ Float m e"
shows "ub_ln prec x = (if x ≤ 0 then None
else if x < 1 then Some (- the (lb_ln prec (float_divl (max prec 1) 1 x)))
else let horner = λx. float_round_up prec (x * ub_ln_horner prec (get_odd prec) 1 x) in
if x ≤ Float 3 (- 1) then Some (horner (x - 1))
else if x < Float 1 1 then Some (float_round_up prec (horner (Float 1 (- 1)) + horner (x * rapprox_rat prec 2 3 - 1)))
else let l = bitlen m - 1 in
Some (float_plus_up prec (float_round_up prec (ub_ln2 prec * (Float (e + l) 0))) (horner (Float m (- l) - 1))))"
(is ?th1)
and "lb_ln prec x = (if x ≤ 0 then None
else if x < 1 then Some (- the (ub_ln prec (float_divr prec 1 x)))
else let horner = λx. float_round_down prec (x * lb_ln_horner prec (get_even prec) 1 x) in
if x ≤ Float 3 (- 1) then Some (horner (x - 1))
else if x < Float 1 1 then Some (float_round_down prec (horner (Float 1 (- 1)) +
horner (max (x * lapprox_rat prec 2 3 - 1) 0)))
else let l = bitlen m - 1 in
Some (float_plus_down prec (float_round_down prec (lb_ln2 prec * (Float (e + l) 0))) (horner (Float m (- l) - 1))))"
(is ?th2)
proof -
from assms Float_pos_eq_mantissa_pos have "x > 0 ⟹ m > 0"
by simp
thus ?th1 ?th2
using Float_representation_aux[of m e]
unfolding x_def[symmetric]
by (auto dest: not_le_imp_less)
qed
lemma ln_shifted_float:
assumes "0 < m"
shows "ln (Float m e) = ln 2 * (e + (bitlen m - 1)) + ln (Float m (- (bitlen m - 1)))"
proof -
let ?B = "2^nat (bitlen m - 1)"
define bl where "bl = bitlen m - 1"
have "0 < real_of_int m" and "⋀X. (0 :: real) < 2^X" and "0 < (2 :: real)" and "m ≠ 0"
using assms by auto
hence "0 ≤ bl" by (simp add: bitlen_alt_def bl_def)
show ?thesis
proof (cases "0 ≤ e")
case True
thus ?thesis
unfolding bl_def[symmetric] using ‹0 < real_of_int m› ‹0 ≤ bl›
apply (simp add: ln_mult)
apply (cases "e=0")
apply (cases "bl = 0", simp_all add: powr_minus ln_inverse ln_powr)
apply (cases "bl = 0", simp_all add: powr_minus ln_inverse ln_powr field_simps)
done
next
case False
hence "0 < -e" by auto
have lne: "ln (2 powr real_of_int e) = ln (inverse (2 powr - e))"
by (simp add: powr_minus)
hence pow_gt0: "(0::real) < 2^nat (-e)"
by auto
hence inv_gt0: "(0::real) < inverse (2^nat (-e))"
by auto
show ?thesis
using False unfolding bl_def[symmetric]
using ‹0 < real_of_int m› ‹0 ≤ bl›
by (auto simp add: lne ln_mult ln_powr ln_div field_simps)
qed
qed
lemma ub_ln_lb_ln_bounds':
assumes "1 ≤ x"
shows "the (lb_ln prec x) ≤ ln x ∧ ln x ≤ the (ub_ln prec x)"
(is "?lb ≤ ?ln ∧ ?ln ≤ ?ub")
proof (cases "x < Float 1 1")
case True
hence "real_of_float (x - 1) < 1" and "real_of_float x < 2" by auto
have "¬ x ≤ 0" and "¬ x < 1" using ‹1 ≤ x› by auto
hence "0 ≤ real_of_float (x - 1)" using ‹1 ≤ x› by auto
have [simp]: "(Float 3 (- 1)) = 3 / 2" by simp
show ?thesis
proof (cases "x ≤ Float 3 (- 1)")
case True
show ?thesis
unfolding lb_ln.simps
unfolding ub_ln.simps Let_def
using ln_float_bounds[OF ‹0 ≤ real_of_float (x - 1)› ‹real_of_float (x - 1) < 1›, of prec]
‹¬ x ≤ 0› ‹¬ x < 1› True
by (auto intro!: float_round_down_le float_round_up_le)
next
case False
hence *: "3 / 2 < x" by auto
with ln_add[of "3 / 2" "x - 3 / 2"]
have add: "ln x = ln (3 / 2) + ln (real_of_float x * 2 / 3)"
by (auto simp add: algebra_simps diff_divide_distrib)
let "?ub_horner x" = "float_round_up prec (x * ub_ln_horner prec (get_odd prec) 1 x)"
let "?lb_horner x" = "float_round_down prec (x * lb_ln_horner prec (get_even prec) 1 x)"
{ have up: "real_of_float (rapprox_rat prec 2 3) ≤ 1"
by (rule rapprox_rat_le1) simp_all
have low: "2 / 3 ≤ rapprox_rat prec 2 3"
by (rule order_trans[OF _ rapprox_rat]) simp
from mult_less_le_imp_less[OF * low] *
have pos: "0 < real_of_float (x * rapprox_rat prec 2 3 - 1)" by auto
have "ln (real_of_float x * 2/3)
≤ ln (real_of_float (x * rapprox_rat prec 2 3 - 1) + 1)"
proof (rule ln_le_cancel_iff[symmetric, THEN iffD1])
show "real_of_float x * 2 / 3 ≤ real_of_float (x * rapprox_rat prec 2 3 - 1) + 1"
using * low by auto
show "0 < real_of_float x * 2 / 3" using * by simp
show "0 < real_of_float (x * rapprox_rat prec 2 3 - 1) + 1" using pos by auto
qed
also have "… ≤ ?ub_horner (x * rapprox_rat prec 2 3 - 1)"
proof (rule float_round_up_le, rule ln_float_bounds(2))
from mult_less_le_imp_less[OF ‹real_of_float x < 2› up] low *
show "real_of_float (x * rapprox_rat prec 2 3 - 1) < 1" by auto
show "0 ≤ real_of_float (x * rapprox_rat prec 2 3 - 1)" using pos by auto
qed
finally have "ln x ≤ ?ub_horner (Float 1 (-1))
+ ?ub_horner ((x * rapprox_rat prec 2 3 - 1))"
using ln_float_bounds(2)[of "Float 1 (- 1)" prec prec] add
by (auto intro!: add_mono float_round_up_le)
note float_round_up_le[OF this, of prec]
}
moreover
{ let ?max = "max (x * lapprox_rat prec 2 3 - 1) 0"
have up: "lapprox_rat prec 2 3 ≤ 2/3"
by (rule order_trans[OF lapprox_rat], simp)
have low: "0 ≤ real_of_float (lapprox_rat prec 2 3)"
using lapprox_rat_nonneg[of 2 3 prec] by simp
have "?lb_horner ?max
≤ ln (real_of_float ?max + 1)"
proof (rule float_round_down_le, rule ln_float_bounds(1))
from mult_less_le_imp_less[OF ‹real_of_float x < 2› up] * low
show "real_of_float ?max < 1" by (cases "real_of_float (lapprox_rat prec 2 3) = 0",
auto simp add: real_of_float_max)
show "0 ≤ real_of_float ?max" by (auto simp add: real_of_float_max)
qed
also have "… ≤ ln (real_of_float x * 2/3)"
proof (rule ln_le_cancel_iff[symmetric, THEN iffD1])
show "0 < real_of_float ?max + 1" by (auto simp add: real_of_float_max)
show "0 < real_of_float x * 2/3" using * by auto
show "real_of_float ?max + 1 ≤ real_of_float x * 2/3" using * up
by (cases "0 < real_of_float x * real_of_float (lapprox_posrat prec 2 3) - 1",
auto simp add: max_def)
qed
finally have "?lb_horner (Float 1 (- 1)) + ?lb_horner ?max ≤ ln x"
using ln_float_bounds(1)[of "Float 1 (- 1)" prec prec] add
by (auto intro!: add_mono float_round_down_le)
note float_round_down_le[OF this, of prec]
}
ultimately
show ?thesis unfolding lb_ln.simps unfolding ub_ln.simps Let_def
using ‹¬ x ≤ 0› ‹¬ x < 1› True False by auto
qed
next
case False
hence "¬ x ≤ 0" and "¬ x < 1" "0 < x" "¬ x ≤ Float 3 (- 1)"
using ‹1 ≤ x› by auto
show ?thesis
proof -
define m where "m = mantissa x"
define e where "e = exponent x"
from Float_mantissa_exponent[of x] have Float: "x = Float m e"
by (simp add: m_def e_def)
let ?s = "Float (e + (bitlen m - 1)) 0"
let ?x = "Float m (- (bitlen m - 1))"
have "0 < m" and "m ≠ 0" using ‹0 < x› Float powr_gt_zero[of 2 e]
by (auto simp add: zero_less_mult_iff)
define bl where "bl = bitlen m - 1"
then have bitlen: "bitlen m = bl + 1"
by simp
hence "bl ≥ 0"
using ‹m > 0› by (auto simp add: bitlen_alt_def)
have "1 ≤ Float m e"
using ‹1 ≤ x› Float unfolding less_eq_float_def by auto
from bitlen_div[OF ‹0 < m›] float_gt1_scale[OF ‹1 ≤ Float m e›] ‹bl ≥ 0›
have x_bnds: "0 ≤ real_of_float (?x - 1)" "real_of_float (?x - 1) < 1"
using abs_real_le_2_powr_bitlen [of m] ‹m > 0›
by (simp_all add: bitlen powr_realpow [symmetric] powr_minus powr_add field_simps)
{
have "float_round_down prec (lb_ln2 prec * ?s) ≤ ln 2 * (e + (bitlen m - 1))"
(is "real_of_float ?lb2 ≤ _")
apply (rule float_round_down_le)
unfolding nat_0 power_0 mult_1_right times_float.rep_eq
using lb_ln2[of prec]
proof (rule mult_mono)
from float_gt1_scale[OF ‹1 ≤ Float m e›]
show "0 ≤ real_of_float (Float (e + (bitlen m - 1)) 0)" by simp
qed auto
moreover
from ln_float_bounds(1)[OF x_bnds]
have "float_round_down prec ((?x - 1) * lb_ln_horner prec (get_even prec) 1 (?x - 1)) ≤ ln ?x" (is "real_of_float ?lb_horner ≤ _")
by (auto intro!: float_round_down_le)
ultimately have "float_plus_down prec ?lb2 ?lb_horner ≤ ln x"
unfolding Float ln_shifted_float[OF ‹0 < m›, of e] by (auto intro!: float_plus_down_le)
}
moreover
{
from ln_float_bounds(2)[OF x_bnds]
have "ln ?x ≤ float_round_up prec ((?x - 1) * ub_ln_horner prec (get_odd prec) 1 (?x - 1))"
(is "_ ≤ real_of_float ?ub_horner")
by (auto intro!: float_round_up_le)
moreover
have "ln 2 * (e + (bitlen m - 1)) ≤ float_round_up prec (ub_ln2 prec * ?s)"
(is "_ ≤ real_of_float ?ub2")
apply (rule float_round_up_le)
unfolding nat_0 power_0 mult_1_right times_float.rep_eq
using ub_ln2[of prec]
proof (rule mult_mono)
from float_gt1_scale[OF ‹1 ≤ Float m e›]
show "0 ≤ real_of_int (e + (bitlen m - 1))" by auto
have "0 ≤ ln (2 :: real)" by simp
thus "0 ≤ real_of_float (ub_ln2 prec)" using ub_ln2[of prec] by arith
qed auto
ultimately have "ln x ≤ float_plus_up prec ?ub2 ?ub_horner"
unfolding Float ln_shifted_float[OF ‹0 < m›, of e]
by (auto intro!: float_plus_up_le)
}
ultimately show ?thesis
unfolding lb_ln.simps
unfolding ub_ln.simps
unfolding if_not_P[OF ‹¬ x ≤ 0›] if_not_P[OF ‹¬ x < 1›]
if_not_P[OF False] if_not_P[OF ‹¬ x ≤ Float 3 (- 1)›] Let_def
unfolding plus_float.rep_eq e_def[symmetric] m_def[symmetric]
by simp
qed
qed
lemma ub_ln_lb_ln_bounds:
assumes "0 < x"
shows "the (lb_ln prec x) ≤ ln x ∧ ln x ≤ the (ub_ln prec x)"
(is "?lb ≤ ?ln ∧ ?ln ≤ ?ub")
proof (cases "x < 1")
case False
hence "1 ≤ x"
unfolding less_float_def less_eq_float_def by auto
show ?thesis
using ub_ln_lb_ln_bounds'[OF ‹1 ≤ x›] .
next
case True
have "¬ x ≤ 0" using ‹0 < x› by auto
from True have "real_of_float x ≤ 1" "x ≤ 1"
by simp_all
have "0 < real_of_float x" and "real_of_float x ≠ 0"
using ‹0 < x› by auto
hence A: "0 < 1 / real_of_float x" by auto
{
let ?divl = "float_divl (max prec 1) 1 x"
have A': "1 ≤ ?divl" using float_divl_pos_less1_bound[OF ‹0 < real_of_float x› ‹real_of_float x ≤ 1›] by auto
hence B: "0 < real_of_float ?divl" by auto
have "ln ?divl ≤ ln (1 / x)" unfolding ln_le_cancel_iff[OF B A] using float_divl[of _ 1 x] by auto
hence "ln x ≤ - ln ?divl" unfolding nonzero_inverse_eq_divide[OF ‹real_of_float x ≠ 0›, symmetric] ln_inverse[OF ‹0 < real_of_float x›] by auto
from this ub_ln_lb_ln_bounds'[OF A', THEN conjunct1, THEN le_imp_neg_le]
have "?ln ≤ - the (lb_ln prec ?divl)" unfolding uminus_float.rep_eq by (rule order_trans)
} moreover
{
let ?divr = "float_divr prec 1 x"
have A': "1 ≤ ?divr" using float_divr_pos_less1_lower_bound[OF ‹0 < x› ‹x ≤ 1›] unfolding less_eq_float_def less_float_def by auto
hence B: "0 < real_of_float ?divr" by auto
have "ln (1 / x) ≤ ln ?divr" unfolding ln_le_cancel_iff[OF A B] using float_divr[of 1 x] by auto
hence "- ln ?divr ≤ ln x" unfolding nonzero_inverse_eq_divide[OF ‹real_of_float x ≠ 0›, symmetric] ln_inverse[OF ‹0 < real_of_float x›] by auto
from ub_ln_lb_ln_bounds'[OF A', THEN conjunct2, THEN le_imp_neg_le] this
have "- the (ub_ln prec ?divr) ≤ ?ln" unfolding uminus_float.rep_eq by (rule order_trans)
}
ultimately show ?thesis unfolding lb_ln.simps[where x=x] ub_ln.simps[where x=x]
unfolding if_not_P[OF ‹¬ x ≤ 0›] if_P[OF True] by auto
qed
lemma lb_ln:
assumes "Some y = lb_ln prec x"
shows "y ≤ ln x" and "0 < real_of_float x"
proof -
have "0 < x"
proof (rule ccontr)
assume "¬ 0 < x"
hence "x ≤ 0"
unfolding less_eq_float_def less_float_def by auto
thus False
using assms by auto
qed
thus "0 < real_of_float x" by auto
have "the (lb_ln prec x) ≤ ln x"
using ub_ln_lb_ln_bounds[OF ‹0 < x›] ..
thus "y ≤ ln x"
unfolding assms[symmetric] by auto
qed
lemma ub_ln:
assumes "Some y = ub_ln prec x"
shows "ln x ≤ y" and "0 < real_of_float x"
proof -
have "0 < x"
proof (rule ccontr)
assume "¬ 0 < x"
hence "x ≤ 0" by auto
thus False
using assms by auto
qed
thus "0 < real_of_float x" by auto
have "ln x ≤ the (ub_ln prec x)"
using ub_ln_lb_ln_bounds[OF ‹0 < x›] ..
thus "ln x ≤ y"
unfolding assms[symmetric] by auto
qed
lemma bnds_ln: "∀(x::real) lx ux. (Some l, Some u) =
(lb_ln prec lx, ub_ln prec ux) ∧ x ∈ {lx .. ux} ⟶ l ≤ ln x ∧ ln x ≤ u"
proof (rule allI, rule allI, rule allI, rule impI)
fix x :: real
fix lx ux
assume "(Some l, Some u) = (lb_ln prec lx, ub_ln prec ux) ∧ x ∈ {lx .. ux}"
hence l: "Some l = lb_ln prec lx " and u: "Some u = ub_ln prec ux" and x: "x ∈ {lx .. ux}"
by auto
have "ln ux ≤ u" and "0 < real_of_float ux"
using ub_ln u by auto
have "l ≤ ln lx" and "0 < real_of_float lx" and "0 < x"
using lb_ln[OF l] x by auto
from ln_le_cancel_iff[OF ‹0 < real_of_float lx› ‹0 < x›] ‹l ≤ ln lx›
have "l ≤ ln x"
using x unfolding atLeastAtMost_iff by auto
moreover
from ln_le_cancel_iff[OF ‹0 < x› ‹0 < real_of_float ux›] ‹ln ux ≤ real_of_float u›
have "ln x ≤ u"
using x unfolding atLeastAtMost_iff by auto
ultimately show "l ≤ ln x ∧ ln x ≤ u" ..
qed
lemmas [simp del] = lb_ln.simps ub_ln.simps
lemma lb_lnD:
"y ≤ ln x ∧ 0 < real_of_float x" if "lb_ln prec x = Some y"
using lb_ln[OF that[symmetric]] by auto
lemma ub_lnD:
"ln x ≤ y∧ 0 < real_of_float x" if "ub_ln prec x = Some y"
using ub_ln[OF that[symmetric]] by auto
lift_definition(code_dt) ln_float_interval :: "nat ⇒ float interval ⇒ float interval option"
is "λprec. λ(lx, ux).
Option.bind (lb_ln prec lx) (λl.
Option.bind (ub_ln prec ux) (λu. Some (l, u)))"
by (auto simp: pred_option_def bind_eq_Some_conv ln_le_cancel_iff[symmetric]
simp del: ln_le_cancel_iff dest!: lb_lnD ub_lnD)
lemma ln_float_interval_eq_Some_conv:
"ln_float_interval p x = Some y ⟷
lb_ln p (lower x) = Some (lower y) ∧ ub_ln p (upper x) = Some (upper y)"
by transfer (auto simp: bind_eq_Some_conv)
lemma ln_float_interval: "ln ` set_of (real_interval x) ⊆ set_of (real_interval y)"
if "ln_float_interval p x = Some y"
using that lb_ln[of "lower y" p "lower x"]
ub_ln[of "lower y" p "lower x"]
apply (auto simp add: set_of_eq ln_float_interval_eq_Some_conv ln_le_cancel_iff)
apply (meson less_le_trans ln_less_cancel_iff not_le)
by (meson less_le_trans ln_less_cancel_iff not_le ub_lnD)
lemma ln_float_intervalI:
"ln x ∈ set_of' (ln_float_interval p X)" if "x ∈⇩r X"
using ln_float_interval[of p X] that
by (auto simp: set_of'_def split: option.splits)
lemma ln_float_interval_eqI:
"ln x ∈⇩r IVL" if "ln_float_interval p X = Some IVL" "x ∈⇩r X"
using ln_float_intervalI[of x X p] that
by (auto simp: set_of'_def split: option.splits)
section ‹Real power function›
definition bnds_powr :: "nat ⇒ float ⇒ float ⇒ float ⇒ float ⇒ (float × float) option" where
"bnds_powr prec l1 u1 l2 u2 = (
if l1 = 0 ∧ u1 = 0 then
Some (0, 0)
else if l1 = 0 ∧ l2 ≥ 1 then
let uln = the (ub_ln prec u1)
in Some (0, ub_exp prec (float_round_up prec (uln * (if uln ≥ 0 then u2 else l2))))
else if l1 ≤ 0 then
None
else
Some (map_bnds lb_exp ub_exp prec
(bnds_mult prec (the (lb_ln prec l1)) (the (ub_ln prec u1)) l2 u2)))"
lemma mono_exp_real: "mono (exp :: real ⇒ real)"
by (auto simp: mono_def)
lemma ub_exp_nonneg: "real_of_float (ub_exp prec x) ≥ 0"
proof -
have "0 ≤ exp (real_of_float x)" by simp
also from exp_boundaries[of x prec]
have "… ≤ real_of_float (ub_exp prec x)" by simp
finally show ?thesis .
qed
lemma bnds_powr:
assumes lu: "Some (l, u) = bnds_powr prec l1 u1 l2 u2"
assumes x: "x ∈ {real_of_float l1..real_of_float u1}"
assumes y: "y ∈ {real_of_float l2..real_of_float u2}"
shows "x powr y ∈ {real_of_float l..real_of_float u}"
proof -
consider "l1 = 0" "u1 = 0" | "l1 = 0" "u1 ≠ 0" "l2 ≥ 1" |
"l1 ≤ 0" "¬(l1 = 0 ∧ (u1 = 0 ∨ l2 ≥ 1))" | "l1 > 0" by force
thus ?thesis
proof cases
assume "l1 = 0" "u1 = 0"
with x lu show ?thesis by (auto simp: bnds_powr_def)
next
assume A: "l1 = 0" "u1 ≠ 0" "l2 ≥ 1"
define uln where "uln = the (ub_ln prec u1)"
show ?thesis
proof (cases "x = 0")
case False
with A x y have "x powr y = exp (ln x * y)" by (simp add: powr_def)
also {
from A x False have "ln x ≤ ln (real_of_float u1)" by simp
also from ub_ln_lb_ln_bounds[of u1 prec] A y x False
have "ln (real_of_float u1) ≤ real_of_float uln" by (simp add: uln_def del: lb_ln.simps)
also from A x y have "… * y ≤ real_of_float uln * (if uln ≥ 0 then u2 else l2)"
by (auto intro: mult_left_mono mult_left_mono_neg)
also have "… ≤ real_of_float (float_round_up prec (uln * (if uln ≥ 0 then u2 else l2)))"
by (simp add: float_round_up_le)
finally have "ln x * y ≤ …" using A y by - simp
}
also have "exp (real_of_float (float_round_up prec (uln * (if uln ≥ 0 then u2 else l2)))) ≤
real_of_float (ub_exp prec (float_round_up prec
(uln * (if uln ≥ 0 then u2 else l2))))"
using exp_boundaries by simp
finally show ?thesis using A x y lu
by (simp add: bnds_powr_def uln_def Let_def del: lb_ln.simps ub_ln.simps)
qed (insert x y lu A, simp_all add: bnds_powr_def Let_def ub_exp_nonneg
del: lb_ln.simps ub_ln.simps)
next
assume "l1 ≤ 0" "¬(l1 = 0 ∧ (u1 = 0 ∨ l2 ≥ 1))"
with lu show ?thesis by (simp add: bnds_powr_def split: if_split_asm)
next
assume l1: "l1 > 0"
obtain lm um where lmum:
"(lm, um) = bnds_mult prec (the (lb_ln prec l1)) (the (ub_ln prec u1)) l2 u2"
by (cases "bnds_mult prec (the (lb_ln prec l1)) (the (ub_ln prec u1)) l2 u2") simp
with l1 have "(l, u) = map_bnds lb_exp ub_exp prec (lm, um)"
using lu by (simp add: bnds_powr_def del: lb_ln.simps ub_ln.simps split: if_split_asm)
hence "exp (ln x * y) ∈ {real_of_float l..real_of_float u}"
proof (rule map_bnds[OF _ mono_exp_real], goal_cases)
case 1
let ?lln = "the (lb_ln prec l1)" and ?uln = "the (ub_ln prec u1)"
from ub_ln_lb_ln_bounds[of l1 prec] ub_ln_lb_ln_bounds[of u1 prec] x l1
have "real_of_float ?lln ≤ ln (real_of_float l1) ∧
ln (real_of_float u1) ≤ real_of_float ?uln"
by (auto simp del: lb_ln.simps ub_ln.simps)
moreover from l1 x have "ln (real_of_float l1) ≤ ln x ∧ ln x ≤ ln (real_of_float u1)"
by auto
ultimately have ln: "real_of_float ?lln ≤ ln x ∧ ln x ≤ real_of_float ?uln" by simp
from lmum show ?case
by (rule bnds_mult) (insert y ln, simp_all)
qed (insert exp_boundaries[of lm prec] exp_boundaries[of um prec], simp_all)
with x l1 show ?thesis
by (simp add: powr_def mult_ac)
qed
qed
lift_definition(code_dt) powr_float_interval :: "nat ⇒ float interval ⇒ float interval ⇒ float interval option"
is "λprec. λ(l1, u1). λ(l2, u2). bnds_powr prec l1 u1 l2 u2"
by (auto simp: pred_option_def dest!: bnds_powr[OF sym])
lemma powr_float_interval:
"{x powr y | x y. x ∈ set_of (real_interval X) ∧ y ∈ set_of (real_interval Y)}
⊆ set_of (real_interval R)"
if "powr_float_interval prec X Y = Some R"
using that
by transfer (auto dest!: bnds_powr[OF sym])
lemma powr_float_intervalI:
"x powr y ∈ set_of' (powr_float_interval p X Y)"
if "x ∈⇩r X" "y ∈⇩r Y"
using powr_float_interval[of p X Y] that
by (auto simp: set_of'_def split: option.splits)
lemma powr_float_interval_eqI:
"x powr y ∈⇩r IVL"
if "powr_float_interval p X Y = Some IVL" "x ∈⇩r X" "y ∈⇩r Y"
using powr_float_intervalI[of x X y Y p] that
by (auto simp: set_of'_def split: option.splits)
end
end