Datasets:
Tasks:
Text Generation
Modalities:
Text
Sub-tasks:
language-modeling
Languages:
English
Size:
100K - 1M
License:
File size: 42,114 Bytes
4365a98 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 |
(*
Authors: Jose Divasón
Sebastiaan Joosten
René Thiemann
Akihisa Yamada
*)
subsection \<open>Mahler Measure\<close>
text \<open>This part contains a definition of the Mahler measure, it contains Landau's inequality and
the Graeffe-transformation. We also assemble a heuristic to approximate the Mahler's measure.\<close>
theory Mahler_Measure
imports
Sqrt_Babylonian.Sqrt_Babylonian
Poly_Mod_Finite_Field_Record_Based (* stuff about polynomials *)
Polynomial_Factorization.Fundamental_Theorem_Algebra_Factorized
Polynomial_Factorization.Missing_Multiset
begin
context comm_monoid_list begin
lemma induct_gen_abs:
assumes "\<And> a r. a\<in>set lst \<Longrightarrow> P (f (h a) r) (f (g a) r)"
"\<And> x y z. P x y \<Longrightarrow> P y z \<Longrightarrow> P x z"
"P (F (map g lst)) (F (map g lst))"
shows "P (F (map h lst)) (F (map g lst)) "
using assms proof(induct lst arbitrary:P)
case (Cons a as P)
have inl:"a\<in>set (a#as)" by auto
let ?uf = "\<lambda> v w. P (f (g a) v) (f (g a) w)"
have p_suc:"?uf (F (map g as)) (F (map g as))"
using Cons.prems(3) by auto
{ fix r aa assume "aa \<in> set as" hence ins:"aa \<in> set (a#as)" by auto
have "P (f (g a) (f (h aa) r)) (f (g a) (f (g aa) r))"
using Cons.prems(1)[of aa "f r (g a)",OF ins]
by (auto simp: assoc commute left_commute)
} note h = this
from Cons.hyps(1)[of ?uf, OF h Cons.prems(2)[simplified] p_suc]
have e1:"P (f (g a) (F (map h as))) (f (g a) (F (map g as)))" by simp
have e2:"P (f (h a) (F (map h as))) (f (g a) (F (map h as)))"
using Cons.prems(1)[OF inl] by blast
from Cons(3)[OF e2 e1] show ?case by auto next
qed auto
end
lemma prod_induct_gen:
assumes "\<And> a r. f (h a * r :: 'a :: {comm_monoid_mult}) = f (g a * r)"
shows "f (\<Prod>v\<leftarrow>lst. h v) = f (\<Prod>v\<leftarrow>lst. g v)"
proof - let "?P x y" = "f x = f y"
show ?thesis using comm_monoid_mult_class.prod_list.induct_gen_abs[of _ ?P,OF assms] by auto
qed
abbreviation complex_of_int::"int \<Rightarrow> complex" where
"complex_of_int \<equiv> of_int"
definition l2norm_list :: "int list \<Rightarrow> int" where
"l2norm_list lst = \<lfloor>sqrt (sum_list (map (\<lambda> a. a * a) lst))\<rfloor>"
abbreviation l2norm :: "int poly \<Rightarrow> int" where
"l2norm p \<equiv> l2norm_list (coeffs p)"
abbreviation "norm2 p \<equiv> \<Sum>a\<leftarrow>coeffs p. (cmod a)\<^sup>2" (* the square of the Euclidean/l2-norm *)
abbreviation l2norm_complex where
"l2norm_complex p \<equiv> sqrt (norm2 p)"
abbreviation height :: "int poly \<Rightarrow> int" where
"height p \<equiv> max_list (map (nat \<circ> abs) (coeffs p))"
definition complex_roots_complex where
"complex_roots_complex (p::complex poly) = (SOME as. smult (coeff p (degree p)) (\<Prod>a\<leftarrow>as. [:- a, 1:]) = p \<and> length as = degree p)"
lemma complex_roots:
"smult (lead_coeff p) (\<Prod>a\<leftarrow>complex_roots_complex p. [:- a, 1:]) = p"
"length (complex_roots_complex p) = degree p"
using someI_ex[OF fundamental_theorem_algebra_factorized]
unfolding complex_roots_complex_def by simp_all
lemma complex_roots_c [simp]:
"complex_roots_complex [:c:] = []"
using complex_roots(2) [of "[:c:]"] by simp
declare complex_roots(2)[simp]
lemma complex_roots_1 [simp]:
"complex_roots_complex 1 = []"
using complex_roots_c [of 1] by (simp add: pCons_one)
lemma linear_term_irreducible\<^sub>d[simp]: "irreducible\<^sub>d [: a, 1:]"
by (rule linear_irreducible\<^sub>d, simp)
definition complex_roots_int where
"complex_roots_int (p::int poly) = complex_roots_complex (map_poly of_int p)"
lemma complex_roots_int:
"smult (lead_coeff p) (\<Prod>a\<leftarrow>complex_roots_int p. [:- a, 1:]) = map_poly of_int p"
"length (complex_roots_int p) = degree p"
proof -
show "smult (lead_coeff p) (\<Prod>a\<leftarrow>complex_roots_int p. [:- a, 1:]) = map_poly of_int p"
"length (complex_roots_int p) = degree p"
using complex_roots[of "map_poly of_int p"] unfolding complex_roots_int_def by auto
qed
text \<open>The measure for polynomials, after K. Mahler\<close>
definition mahler_measure_poly where
"mahler_measure_poly p = cmod (lead_coeff p) * (\<Prod>a\<leftarrow>complex_roots_complex p. (max 1 (cmod a)))"
definition mahler_measure where
"mahler_measure p = mahler_measure_poly (map_poly complex_of_int p)"
definition mahler_measure_monic where
"mahler_measure_monic p = (\<Prod>a\<leftarrow>complex_roots_complex p. (max 1 (cmod a)))"
lemma mahler_measure_poly_via_monic :
"mahler_measure_poly p = cmod (lead_coeff p) * mahler_measure_monic p"
unfolding mahler_measure_poly_def mahler_measure_monic_def by simp
lemma smult_inj[simp]: assumes "(a::'a::idom) \<noteq> 0" shows "inj (smult a)"
proof-
interpret map_poly_inj_zero_hom "(*) a" using assms by (unfold_locales, auto)
show ?thesis unfolding smult_as_map_poly by (rule inj_f)
qed
definition reconstruct_poly::"'a::idom \<Rightarrow> 'a list \<Rightarrow> 'a poly" where
"reconstruct_poly c roots = smult c (\<Prod>a\<leftarrow>roots. [:- a, 1:])"
lemma reconstruct_is_original_poly:
"reconstruct_poly (lead_coeff p) (complex_roots_complex p) = p"
using complex_roots(1) by (simp add: reconstruct_poly_def)
lemma reconstruct_with_type_conversion:
"smult (lead_coeff (map_poly of_int f)) (prod_list (map (\<lambda> a. [:- a, 1:]) (complex_roots_int f)))
= map_poly of_int f"
unfolding complex_roots_int_def complex_roots(1) by simp
lemma reconstruct_prod:
shows "reconstruct_poly (a::complex) as * reconstruct_poly b bs
= reconstruct_poly (a * b) (as @ bs)"
unfolding reconstruct_poly_def by auto
lemma linear_term_inj[simplified,simp]: "inj (\<lambda> a. [:- a, 1::'a::idom:])"
unfolding inj_on_def by simp
lemma reconstruct_poly_monic_defines_mset:
assumes "(\<Prod>a\<leftarrow>as. [:- a, 1:]) = (\<Prod>a\<leftarrow>bs. [:- a, 1::'a::field:])"
shows "mset as = mset bs"
proof -
let ?as = "mset (map (\<lambda> a. [:- a, 1:]) as)"
let ?bs = "mset (map (\<lambda> a. [:- a, 1:]) bs)"
have eq_smult:"prod_mset ?as = prod_mset ?bs" using assms by (metis prod_mset_prod_list)
have irr:"\<And> as::'a list. set_mset (mset (map (\<lambda> a. [:- a, 1:]) as)) \<subseteq> {q. irreducible q \<and> monic q}"
by (auto intro!: linear_term_irreducible\<^sub>d[of "-_::'a", simplified])
from monic_factorization_unique_mset[OF eq_smult irr irr]
show ?thesis apply (subst inj_eq[OF multiset.inj_map,symmetric]) by auto
qed
lemma reconstruct_poly_defines_mset_of_argument:
assumes "(a::'a::field) \<noteq> 0"
"reconstruct_poly a as = reconstruct_poly a bs"
shows "mset as = mset bs"
proof -
have eq_smult:"smult a (\<Prod>a\<leftarrow>as. [:- a, 1:]) = smult a (\<Prod>a\<leftarrow>bs. [:- a, 1:])"
using assms(2) by (auto simp:reconstruct_poly_def)
from reconstruct_poly_monic_defines_mset[OF Fun.injD[OF smult_inj[OF assms(1)] eq_smult]]
show ?thesis by simp
qed
lemma complex_roots_complex_prod [simp]:
assumes "f \<noteq> 0" "g \<noteq> 0"
shows "mset (complex_roots_complex (f * g))
= mset (complex_roots_complex f) + mset (complex_roots_complex g)"
proof -
let ?p = "f * g"
let "?lc v" = "(lead_coeff (v:: complex poly))"
have nonzero_prod:"?lc ?p \<noteq> 0" using assms by auto
from reconstruct_prod[of "?lc f" "complex_roots_complex f" "?lc g" "complex_roots_complex g"]
have "reconstruct_poly (?lc ?p) (complex_roots_complex ?p)
= reconstruct_poly (?lc ?p) (complex_roots_complex f @ complex_roots_complex g)"
unfolding lead_coeff_mult[symmetric] reconstruct_is_original_poly by auto
from reconstruct_poly_defines_mset_of_argument[OF nonzero_prod this]
show ?thesis by simp
qed
lemma mset_mult_add:
assumes "mset (a::'a::field list) = mset b + mset c"
shows "prod_list a = prod_list b * prod_list c"
unfolding prod_mset_prod_list[symmetric]
using prod_mset_Un[of "mset b" "mset c",unfolded assms[symmetric]].
lemma mset_mult_add_2:
assumes "mset a = mset b + mset c"
shows "prod_list (map i a::'b::field list) = prod_list (map i b) * prod_list (map i c)"
proof -
have r:"mset (map i a) = mset (map i b) + mset (map i c) " using assms
by (metis map_append mset_append mset_map)
show ?thesis using mset_mult_add[OF r] by auto
qed
lemma measure_mono_eq_prod:
assumes "f \<noteq> 0" "g \<noteq> 0"
shows "mahler_measure_monic (f * g) = mahler_measure_monic f * mahler_measure_monic g"
unfolding mahler_measure_monic_def
using mset_mult_add_2[OF complex_roots_complex_prod[OF assms],of "\<lambda> a. max 1 (cmod a)"] by simp
lemma mahler_measure_poly_0[simp]: "mahler_measure_poly 0 = 0" unfolding mahler_measure_poly_via_monic by auto
lemma measure_eq_prod: (* Remark 10.2 *)
"mahler_measure_poly (f * g) = mahler_measure_poly f * mahler_measure_poly g"
proof -
consider "f = 0" | "g = 0" | (both) "f \<noteq> 0" "g \<noteq> 0" by auto
thus ?thesis proof(cases)
case both show ?thesis unfolding mahler_measure_poly_via_monic norm_mult lead_coeff_mult
by (auto simp: measure_mono_eq_prod[OF both])
qed (simp_all)
qed
lemma prod_cmod[simp]:
"cmod (\<Prod>a\<leftarrow>lst. f a) = (\<Prod>a\<leftarrow>lst. cmod (f a))"
by(induct lst,auto simp:real_normed_div_algebra_class.norm_mult)
lemma lead_coeff_of_prod[simp]:
"lead_coeff (\<Prod>a\<leftarrow>lst. f a::'a::idom poly) = (\<Prod>a\<leftarrow>lst. lead_coeff (f a))"
by(induct lst,auto simp:lead_coeff_mult)
lemma ineq_about_squares:assumes "x \<le> (y::real)" shows "x \<le> c^2 + y" using assms
by (simp add: add.commute add_increasing2)
lemma first_coeff_le_tail:"(cmod (lead_coeff g))^2 \<le> (\<Sum>a\<leftarrow>coeffs g. (cmod a)^2)"
proof(induct g)
case (pCons a p)
thus ?case proof(cases "p = 0") case False
show ?thesis using pCons unfolding lead_coeff_pCons(1)[OF False]
by(cases "a = 0",simp_all add:ineq_about_squares)
qed simp
qed simp
lemma square_prod_cmod[simp]:
"(cmod (a * b))^2 = cmod a ^ 2 * cmod b ^ 2"
by (simp add: norm_mult power_mult_distrib)
lemma sum_coeffs_smult_cmod:
"(\<Sum>a\<leftarrow>coeffs (smult v p). (cmod a)^2) = (cmod v)^2 * (\<Sum>a\<leftarrow>coeffs p. (cmod a)^2)"
(is "?l = ?r")
proof -
have "?l = (\<Sum>a\<leftarrow>coeffs p. (cmod v)^2 * (cmod a)^2)" by(cases "v=0";induct p,auto)
thus ?thesis by (auto simp:sum_list_const_mult)
qed
abbreviation "linH a \<equiv> if (cmod a > 1) then [:- 1,cnj a:] else [:- a,1:]"
lemma coeffs_cong_1[simp]: "cCons a v = cCons b v \<longleftrightarrow> a = b" unfolding cCons_def by auto
lemma strip_while_singleton[simp]:
"strip_while ((=) 0) [v * a] = cCons (v * a) []" unfolding cCons_def strip_while_def by auto
lemma coeffs_times_linterm:
shows "coeffs (pCons 0 (smult a p) + smult b p) = strip_while (HOL.eq (0::'a::{comm_ring_1}))
(map (\<lambda>(c,d).b*d+c*a) (zip (0 # coeffs p) (coeffs p @ [0])))" proof -
{fix v
have "coeffs (smult b p + pCons (a* v) (smult a p)) = strip_while (HOL.eq 0) (map (\<lambda>(c,d).b*d+c*a) (zip ([v] @ coeffs p) (coeffs p @ [0])))"
proof(induct p arbitrary:v) case (pCons pa ps) thus ?case by auto qed auto (* just putting ;auto does not work *)
}
from this[of 0] show ?thesis by (simp add: add.commute)
qed
lemma filter_distr_rev[simp]:
shows "filter f (rev lst) = rev (filter f lst)"
by(induct lst;auto)
lemma strip_while_filter:
shows "filter ((\<noteq>) 0) (strip_while ((=) 0) (lst::'a::zero list)) = filter ((\<noteq>) 0) lst"
proof - {fix lst::"'a list"
have "filter ((\<noteq>) 0) (dropWhile ((=) 0) lst) = filter ((\<noteq>) 0) lst" by (induct lst;auto)
hence "(filter ((\<noteq>) 0) (strip_while ((=) 0) (rev lst))) = filter ((\<noteq>) 0) (rev lst)"
unfolding strip_while_def by(simp)}
from this[of "rev lst"] show ?thesis by simp
qed
lemma sum_stripwhile[simp]:
assumes "f 0 = 0"
shows "(\<Sum>a\<leftarrow>strip_while ((=) 0) lst. f a) = (\<Sum>a\<leftarrow>lst. f a)"
proof -
{fix lst
have "(\<Sum>a\<leftarrow>filter ((\<noteq>) 0) lst. f a) = (\<Sum>a\<leftarrow>lst. f a)" by(induct lst,auto simp:assms)}
note f=this
have "sum_list (map f (filter ((\<noteq>) 0) (strip_while ((=) 0) lst)))
= sum_list (map f (filter ((\<noteq>) 0) lst))"
using strip_while_filter[of lst] by(simp)
thus ?thesis unfolding f.
qed
lemma complex_split : "Complex a b = c \<longleftrightarrow> (a = Re c \<and> b = Im c)"
using complex_surj by auto
lemma norm_times_const:"(\<Sum>y\<leftarrow>lst. (cmod (a * y))\<^sup>2) = (cmod a)\<^sup>2 * (\<Sum>y\<leftarrow>lst. (cmod y)\<^sup>2)"
by(induct lst,auto simp:ring_distribs)
fun bisumTail where (* Used for Landau's lemma *)
"bisumTail f (Cons a (Cons b bs)) = f a b + bisumTail f (Cons b bs)" |
"bisumTail f (Cons a Nil) = f a 0" |
"bisumTail f Nil = f 1 0" (* never called, not used in proofs *)
fun bisum where
"bisum f (Cons a as) = f 0 a + bisumTail f (Cons a as)" |
"bisum f Nil = f 0 0"
lemma bisumTail_is_map_zip:
"(\<Sum>x\<leftarrow>zip (v # l1) (l1 @ [0]). f x) = bisumTail (\<lambda>x y .f (x,y)) (v#l1)"
by(induct l1 arbitrary:v,auto)
(* converting to and from bisum *)
lemma bisum_is_map_zip:
"(\<Sum>x\<leftarrow>zip (0 # l1) (l1 @ [0]). f x) = bisum (\<lambda>x y. f (x,y)) l1"
using bisumTail_is_map_zip[of f "hd l1" "tl l1"] by(cases l1,auto)
lemma map_zip_is_bisum:
"bisum f l1 = (\<Sum>(x,y)\<leftarrow>zip (0 # l1) (l1 @ [0]). f x y)"
using bisum_is_map_zip[of "\<lambda>(x,y). f x y"] by auto
lemma bisum_outside :
"(bisum (\<lambda> x y. f1 x - f2 x y + f3 y) lst :: 'a :: field)
= sum_list (map f1 lst) + f1 0 - bisum f2 lst + sum_list (map f3 lst) + f3 0"
proof(cases lst)
case (Cons a lst) show ?thesis unfolding map_zip_is_bisum Cons by(induct lst arbitrary:a,auto)
qed auto
lemma Landau_lemma:
"(\<Sum>a\<leftarrow>coeffs (\<Prod>a\<leftarrow>lst. [:- a, 1:]). (cmod a)\<^sup>2) = (\<Sum>a\<leftarrow>coeffs (\<Prod>a\<leftarrow>lst. linH a). (cmod a)\<^sup>2)"
(is "norm2 ?l = norm2 ?r")
proof -
have a:"\<And> a. (cmod a)\<^sup>2 = Re (a * cnj a) " using complex_norm_square
unfolding complex_split complex_of_real_def by simp
have b:"\<And> x a y. (cmod (x - a * y))^2
= (cmod x)\<^sup>2 - Re (a * y * cnj x + x * cnj (a * y)) + (cmod (a * y))^2"
unfolding left_diff_distrib right_diff_distrib a complex_cnj_diff by simp
have c:"\<And> y a x. (cmod (cnj a * x - y))\<^sup>2
= (cmod (a * x))\<^sup>2 - Re (a * y * cnj x + x * cnj (a * y)) + (cmod y)^2"
unfolding left_diff_distrib right_diff_distrib a complex_cnj_diff
by (simp add: mult.assoc mult.left_commute)
{ fix f1 a
have "norm2 ([:- a, 1 :] * f1) = bisum (\<lambda>x y. cmod (x - a * y)^2) (coeffs f1)"
by(simp add: bisum_is_map_zip[of _ "coeffs f1"] coeffs_times_linterm[of 1 _ "-a",simplified])
also have "\<dots> = norm2 f1 + cmod a^2*norm2 f1
- bisum (\<lambda>x y. Re (a * y * cnj x + x * cnj (a * y))) (coeffs f1)"
unfolding b bisum_outside norm_times_const by simp
also have "\<dots> = bisum (\<lambda>x y. cmod (cnj a * x - y)^2) (coeffs f1)"
unfolding c bisum_outside norm_times_const by auto
also have "\<dots> = norm2 ([:- 1, cnj a :] * f1)"
using coeffs_times_linterm[of "cnj a" _ "-1"]
by(simp add: bisum_is_map_zip[of _ "coeffs f1"] mult.commute)
finally have "norm2 ([:- a, 1 :] * f1) = \<dots>".}
hence h:"\<And> a f1. norm2 ([:- a, 1 :] * f1) = norm2 (linH a * f1)" by auto
show ?thesis by(rule prod_induct_gen[OF h])
qed
lemma Landau_inequality:
"mahler_measure_poly f \<le> l2norm_complex f"
proof -
let ?f = "reconstruct_poly (lead_coeff f) (complex_roots_complex f)"
let ?roots = "(complex_roots_complex f)"
let ?g = "\<Prod>a\<leftarrow>?roots. linH a"
(* g is chosen such that lead_coeff_g holds, and its l2 norm is equal to f's l2 norm *)
have max:"\<And>a. cmod (if 1 < cmod a then cnj a else 1) = max 1 (cmod a)"
by simp
have "\<And>a. 1 < cmod a \<Longrightarrow> a \<noteq> 0" by auto
hence "\<And>a. lead_coeff (linH a) = (if (cmod a > 1) then cnj a else 1)" by(auto simp:if_split)
hence lead_coeff_g:"cmod (lead_coeff ?g) = (\<Prod>a\<leftarrow>?roots. max 1 (cmod a))" by(auto simp:max)
have "norm2 f = (\<Sum>a\<leftarrow>coeffs ?f. (cmod a)^2)" unfolding reconstruct_is_original_poly..
also have "\<dots> = cmod (lead_coeff f)^2 * (\<Sum>a\<leftarrow>coeffs (\<Prod>a\<leftarrow>?roots. [:- a, 1:]). (cmod a)\<^sup>2)"
unfolding reconstruct_poly_def using sum_coeffs_smult_cmod.
finally have fg_norm:"norm2 f = cmod (lead_coeff f)^2 * (\<Sum>a\<leftarrow>coeffs ?g. (cmod a)^2)"
unfolding Landau_lemma by auto
have "(cmod (lead_coeff ?g))^2 \<le> (\<Sum>a\<leftarrow>coeffs ?g. (cmod a)^2)"
using first_coeff_le_tail by blast
from ordered_comm_semiring_class.comm_mult_left_mono[OF this]
have "(cmod (lead_coeff f) * cmod (lead_coeff ?g))^2 \<le> (\<Sum>a\<leftarrow>coeffs f. (cmod a)^2)"
unfolding fg_norm by (simp add:power_mult_distrib)
hence "cmod (lead_coeff f) * (\<Prod>a\<leftarrow>?roots. max 1 (cmod a)) \<le> sqrt (norm2 f)"
using NthRoot.real_le_rsqrt lead_coeff_g by auto
thus "mahler_measure_poly f \<le> sqrt (norm2 f)"
using reconstruct_with_type_conversion[unfolded complex_roots_int_def]
by (simp add: mahler_measure_poly_via_monic mahler_measure_monic_def complex_roots_int_def)
qed
lemma prod_list_ge1:
assumes "Ball (set x) (\<lambda> (a::real). a \<ge> 1)"
shows "prod_list x \<ge> 1"
using assms proof(induct x)
case (Cons a as)
have "\<forall>a\<in>set as. 1 \<le> a" "1 \<le> a" using Cons(2) by auto
thus ?case using Cons.hyps mult_mono' by fastforce
qed auto
lemma mahler_measure_monic_ge_1: "mahler_measure_monic p \<ge> 1"
unfolding mahler_measure_monic_def by(rule prod_list_ge1,simp)
lemma mahler_measure_monic_ge_0: "mahler_measure_monic p \<ge> 0"
using mahler_measure_monic_ge_1 le_numeral_extra(1) order_trans by blast
lemma mahler_measure_ge_0: "0 \<le> mahler_measure h" unfolding mahler_measure_def mahler_measure_poly_via_monic
by (simp add: mahler_measure_monic_ge_0)
lemma mahler_measure_constant[simp]: "mahler_measure_poly [:c:] = cmod c"
proof -
have main: "complex_roots_complex [:c:] = []" unfolding complex_roots_complex_def
by (rule some_equality, auto)
show ?thesis unfolding mahler_measure_poly_def main by auto
qed
lemma mahler_measure_factor[simplified,simp]: "mahler_measure_poly [:- a, 1:] = max 1 (cmod a)"
proof -
have main: "complex_roots_complex [:- a, 1:] = [a]" unfolding complex_roots_complex_def
proof (rule some_equality, auto, goal_cases)
case (1 as)
thus ?case by (cases as, auto)
qed
show ?thesis unfolding mahler_measure_poly_def main by auto
qed
lemma mahler_measure_poly_explicit: "mahler_measure_poly (smult c (\<Prod>a\<leftarrow>as. [:- a, 1:]))
= cmod c * (\<Prod>a\<leftarrow>as. (max 1 (cmod a)))"
proof (cases "c = 0")
case True
thus ?thesis by auto
next
case False note c = this
show ?thesis
proof (induct as)
case (Cons a as)
have "mahler_measure_poly (smult c (\<Prod>a\<leftarrow>a # as. [:- a, 1:]))
= mahler_measure_poly (smult c (\<Prod>a\<leftarrow>as. [:- a, 1:]) * [: -a, 1 :])"
by (rule arg_cong[of _ _ mahler_measure_poly], unfold list.simps prod_list.Cons mult_smult_left, simp)
also have "\<dots> = mahler_measure_poly (smult c (\<Prod>a\<leftarrow>as. [:- a, 1:])) * mahler_measure_poly ([:- a, 1:])"
(is "_ = ?l * ?r") by (rule measure_eq_prod)
also have "?l = cmod c * (\<Prod>a\<leftarrow>as. max 1 (cmod a))" unfolding Cons by simp
also have "?r = max 1 (cmod a)" by simp
finally show ?case by simp
next
case Nil
show ?case by simp
qed
qed
lemma mahler_measure_poly_ge_1:
assumes "h \<noteq> 0"
shows "(1::real) \<le> mahler_measure h"
proof -
have rc: "\<bar>real_of_int i\<bar> = of_int \<bar>i\<bar>" for i by simp
from assms have "cmod (lead_coeff (map_poly complex_of_int h)) > 0" by simp
hence "cmod (lead_coeff (map_poly complex_of_int h)) \<ge> 1"
by(cases "lead_coeff h = 0", auto simp del: leading_coeff_0_iff)
from mult_mono[OF this mahler_measure_monic_ge_1 norm_ge_zero]
show ?thesis unfolding mahler_measure_def mahler_measure_poly_via_monic
by auto
qed
lemma mahler_measure_dvd: assumes "f \<noteq> 0" and "h dvd f"
shows "mahler_measure h \<le> mahler_measure f"
proof -
from assms obtain g where f: "f = g * h" unfolding dvd_def by auto
from f assms have g0: "g \<noteq> 0" by auto
hence mg: "mahler_measure g \<ge> 1" by (rule mahler_measure_poly_ge_1)
have "1 * mahler_measure h \<le> mahler_measure f"
unfolding mahler_measure_def f measure_eq_prod
of_int_poly_hom.hom_mult unfolding mahler_measure_def[symmetric]
by (rule mult_right_mono[OF mg mahler_measure_ge_0])
thus ?thesis by simp
qed
definition graeffe_poly :: "'a \<Rightarrow> 'a :: comm_ring_1 list \<Rightarrow> nat \<Rightarrow> 'a poly" where
"graeffe_poly c as m = smult (c ^ (2^m)) (\<Prod>a\<leftarrow>as. [:- (a ^ (2^m)), 1:])"
context
fixes f :: "complex poly" and c as
assumes f: "f = smult c (\<Prod>a\<leftarrow>as. [:- a, 1:])"
begin
lemma mahler_graeffe: "mahler_measure_poly (graeffe_poly c as m) = (mahler_measure_poly f)^(2^m)"
proof -
have graeffe: "graeffe_poly c as m = smult (c ^ 2 ^ m) (\<Prod>a\<leftarrow>(map (\<lambda> a. a ^ 2 ^ m) as). [:- a, 1:])"
unfolding graeffe_poly_def
by (rule arg_cong[of _ _ "smult (c ^ 2 ^ m)"], induct as, auto)
{
fix n :: nat
assume n: "n > 0"
have id: "max 1 (cmod a ^ n) = max 1 (cmod a) ^ n" for a
proof (cases "cmod a \<le> 1")
case True
hence "cmod a ^ n \<le> 1" by (simp add: power_le_one)
with True show ?thesis by (simp add: max_def)
qed (auto simp: max_def)
have "(\<Prod>x\<leftarrow>as. max 1 (cmod x ^ n)) = (\<Prod>a\<leftarrow>as. max 1 (cmod a)) ^ n"
by (induct as, auto simp: field_simps n id)
}
thus ?thesis unfolding f mahler_measure_poly_explicit graeffe
by (auto simp: o_def field_simps norm_power)
qed
end
fun drop_half :: "'a list \<Rightarrow> 'a list" where
"drop_half (x # y # ys) = x # drop_half ys"
| "drop_half xs = xs"
fun alternate :: "'a list \<Rightarrow> 'a list \<times> 'a list" where
"alternate (x # y # ys) = (case alternate ys of (evn, od) \<Rightarrow> (x # evn, y # od))"
| "alternate xs = (xs,[])"
definition poly_square_subst :: "'a :: comm_ring_1 poly \<Rightarrow> 'a poly" where
"poly_square_subst f = poly_of_list (drop_half (coeffs f))"
definition poly_even_odd :: "'a :: comm_ring_1 poly \<Rightarrow> 'a poly \<times> 'a poly" where
"poly_even_odd f = (case alternate (coeffs f) of (evn,od) \<Rightarrow> (poly_of_list evn, poly_of_list od))"
lemma poly_square_subst_coeff: "coeff (poly_square_subst f) i = coeff f (2 * i)"
proof -
have id: "coeff f (2 * i) = coeff (Poly (coeffs f)) (2 * i)" by simp
obtain xs where xs: "coeffs f = xs" by auto
show ?thesis unfolding poly_square_subst_def poly_of_list_def coeff_Poly_eq id xs
proof (induct xs arbitrary: i rule: drop_half.induct)
case (1 x y ys i) thus ?case by (cases i, auto)
next
case ("2_2" x i) thus ?case by (cases i, auto)
qed auto
qed
lemma poly_even_odd_coeff: assumes "poly_even_odd f = (ev,od)"
shows "coeff ev i = coeff f (2 * i)" "coeff od i = coeff f (2 * i + 1)"
proof -
have id: "\<And> i. coeff f i = coeff (Poly (coeffs f)) i" by simp
obtain xs where xs: "coeffs f = xs" by auto
from assms[unfolded poly_even_odd_def]
have ev_od: "ev = Poly (fst (alternate xs))" "od = Poly (snd (alternate xs))"
by (auto simp: xs split: prod.splits)
have "coeff ev i = coeff f (2 * i) \<and> coeff od i = coeff f (2 * i + 1)"
unfolding poly_of_list_def coeff_Poly_eq id xs ev_od
proof (induct xs arbitrary: i rule: alternate.induct)
case (1 x y ys i) thus ?case by (cases "alternate ys"; cases i, auto)
next
case ("2_2" x i) thus ?case by (cases i, auto)
qed auto
thus "coeff ev i = coeff f (2 * i)" "coeff od i = coeff f (2 * i + 1)" by auto
qed
lemma poly_square_subst: "poly_square_subst (f \<circ>\<^sub>p (monom 1 2)) = f"
by (rule poly_eqI, unfold poly_square_subst_coeff, subst coeff_pcompose_x_pow_n, auto)
lemma poly_even_odd: assumes "poly_even_odd f = (g,h)"
shows "f = g \<circ>\<^sub>p monom 1 2 + monom 1 1 * (h \<circ>\<^sub>p monom 1 2)"
proof -
note id = poly_even_odd_coeff[OF assms]
show ?thesis
proof (rule poly_eqI, unfold coeff_add coeff_monom_mult)
fix n :: nat
obtain m i where mi: "m = n div 2" "i = n mod 2" by auto
have nmi: "n = 2 * m + i" "i < 2" "0 < (2 :: nat)" "1 < (2 :: nat)" unfolding mi by auto
have "(2 :: nat) \<noteq> 0" by auto
show "coeff f n = coeff (g \<circ>\<^sub>p monom 1 2) n + (if 1 \<le> n then 1 * coeff (h \<circ>\<^sub>p monom 1 2) (n - 1) else 0)"
proof (cases "i = 1")
case True
hence id1: "2 * m + i - 1 = 2 * m + 0" by auto
show ?thesis unfolding nmi id id1 coeff_pcompose_monom[OF nmi(2)] coeff_pcompose_monom[OF nmi(3)]
unfolding True by auto
next
case False
with nmi have i0: "i = 0" by auto
show ?thesis
proof (cases m)
case (Suc k)
hence id1: "2 * m + i - 1 = 2 * k + 1" using i0 by auto
show ?thesis unfolding nmi id coeff_pcompose_monom[OF nmi(2)]
coeff_pcompose_monom[OF nmi(4)] id1 unfolding Suc i0 by auto
next
case 0
show ?thesis unfolding nmi id coeff_pcompose_monom[OF nmi(2)] unfolding i0 0 by auto
qed
qed
qed
qed
context
fixes f :: "'a :: idom poly"
begin
lemma graeffe_0: "f = smult c (\<Prod>a\<leftarrow>as. [:- a, 1:]) \<Longrightarrow> graeffe_poly c as 0 = f"
unfolding graeffe_poly_def by auto
lemma graeffe_recursion: assumes "graeffe_poly c as m = f"
shows "graeffe_poly c as (Suc m) = smult ((-1)^(degree f)) (poly_square_subst (f * f \<circ>\<^sub>p [:0,-1:]))"
proof -
let ?g = "graeffe_poly c as m"
have "f * f \<circ>\<^sub>p [:0,-1:] = ?g * ?g \<circ>\<^sub>p [:0,-1:]" unfolding assms by simp
also have "?g \<circ>\<^sub>p [:0,-1:] = smult ((- 1) ^ length as) (smult (c ^ 2 ^ m) (\<Prod>a\<leftarrow>as. [:a ^ 2 ^ m, 1:]))"
unfolding graeffe_poly_def
proof (induct as)
case (Cons a as)
have "?case = ((smult (c ^ 2 ^ m) ([:- (a ^ 2 ^ m), 1:] \<circ>\<^sub>p [:0, - 1:] * (\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ m), 1:]) \<circ>\<^sub>p [:0, - 1:]) =
smult (-1 * (- 1) ^ length as)
(smult (c ^ 2 ^ m) ([: a ^ 2 ^ m, 1:] * (\<Prod>a\<leftarrow>as. [:a ^ 2 ^ m, 1:])))))"
unfolding list.simps prod_list.Cons pcompose_smult pcompose_mult by simp
also have "smult (c ^ 2 ^ m) ([:- (a ^ 2 ^ m), 1:] \<circ>\<^sub>p [:0, - 1:] * (\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ m), 1:]) \<circ>\<^sub>p [:0, - 1:])
= smult (c ^ 2 ^ m) ((\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ m), 1:]) \<circ>\<^sub>p [:0, - 1:]) * [:- (a ^ 2 ^ m), 1:] \<circ>\<^sub>p [:0, - 1:]"
unfolding mult_smult_left by simp
also have "smult (c ^ 2 ^ m) ((\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ m), 1:]) \<circ>\<^sub>p [:0, - 1:]) =
smult ((- 1) ^ length as) (smult (c ^ 2 ^ m) (\<Prod>a\<leftarrow>as. [:a ^ 2 ^ m, 1:]))"
unfolding pcompose_smult[symmetric] Cons ..
also have "[:- (a ^ 2 ^ m), 1:] \<circ>\<^sub>p [:0, - 1:] = smult (-1) [: a^2^m, 1:]" by simp
finally have id: "?case = (smult ((- 1) ^ length as) (smult (c ^ 2 ^ m) (\<Prod>a\<leftarrow>as. [:a ^ 2 ^ m, 1:])) * smult (- 1) [:a ^ 2 ^ m, 1:] =
smult (- 1 * (- 1) ^ length as) (smult (c ^ 2 ^ m) ([:a ^ 2 ^ m, 1:] * (\<Prod>a\<leftarrow>as. [:a ^ 2 ^ m, 1:]))))" by simp
obtain c d where id': "(\<Prod>a\<leftarrow>as. [:a ^ 2 ^ m, 1:]) = c" "[:a ^ 2 ^ m, 1:] = d" by auto
show ?case unfolding id unfolding id' by (simp add: ac_simps)
qed simp
finally have "f * f \<circ>\<^sub>p [:0, - 1:] =
smult ((- 1) ^ length as * (c ^ 2 ^ m * c ^ 2 ^ m))
((\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ m), 1:]) * (\<Prod>a\<leftarrow>as. [:a ^ 2 ^ m, 1:]))"
unfolding graeffe_poly_def by (simp add: ac_simps)
also have "c ^ 2 ^ m * c ^ 2 ^ m = c ^ 2 ^ (Suc m)" by (simp add: semiring_normalization_rules(36))
also have "(\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ m), 1:]) * (\<Prod>a\<leftarrow>as. [:a ^ 2 ^ m, 1:]) =
(\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ (Suc m)), 1:]) \<circ>\<^sub>p monom 1 2"
proof (induct as)
case (Cons a as)
have id: "(monom 1 2 :: 'a poly) = [:0,0,1:]"
by (metis monom_altdef pCons_0_as_mult power2_eq_square smult_1_left)
have "(\<Prod>a\<leftarrow>a # as. [:- (a ^ 2 ^ m), 1:]) * (\<Prod>a\<leftarrow>a # as. [:a ^ 2 ^ m, 1:])
= ([:- (a ^ 2 ^ m), 1:] * [: a ^ 2 ^ m, 1:]) * ((\<Prod>a\<leftarrow> as. [:- (a ^ 2 ^ m), 1:]) * (\<Prod>a\<leftarrow> as. [:a ^ 2 ^ m, 1:]))"
(is "_ = ?a * ?b")
unfolding list.simps prod_list.Cons by (simp only: ac_simps)
also have "?b = (\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ Suc m), 1:]) \<circ>\<^sub>p monom 1 2" unfolding Cons by simp
also have "?a = [: - (a ^ 2 ^ (Suc m)), 0 , 1:]" by (simp add: semiring_normalization_rules(36))
also have "\<dots> = [: - (a ^ 2 ^ (Suc m)), 1:] \<circ>\<^sub>p monom 1 2" by (simp add: id)
also have "[: - (a ^ 2 ^ (Suc m)), 1:] \<circ>\<^sub>p monom 1 2 * (\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ Suc m), 1:]) \<circ>\<^sub>p monom 1 2 =
(\<Prod>a\<leftarrow>a # as. [:- (a ^ 2 ^ Suc m), 1:]) \<circ>\<^sub>p monom 1 2" unfolding pcompose_mult[symmetric] by simp
finally show ?case .
qed simp
finally have "f * f \<circ>\<^sub>p [:0, - 1:] = (smult ((- 1) ^ length as) (graeffe_poly c as (Suc m)) \<circ>\<^sub>p monom 1 2)"
unfolding graeffe_poly_def pcompose_smult by simp
from arg_cong[OF this, of "\<lambda> f. smult ((- 1) ^ length as) (poly_square_subst f)", unfolded poly_square_subst]
have "graeffe_poly c as (Suc m) = smult ((- 1) ^ length as) (poly_square_subst (f * f \<circ>\<^sub>p [:0, - 1:]))" by simp
also have "\<dots> = smult ((- 1) ^ degree f) (poly_square_subst (f * f \<circ>\<^sub>p [:0, - 1:]))"
proof (cases "f = 0")
case True
thus ?thesis by (auto simp: poly_square_subst_def)
next
case False
with assms have c0: "c \<noteq> 0" unfolding graeffe_poly_def by auto
from arg_cong[OF assms, of degree]
have "degree f = degree (smult (c ^ 2 ^ m) (\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ m), 1:]))" unfolding graeffe_poly_def by auto
also have "\<dots> = degree (\<Prod>a\<leftarrow>as. [:- (a ^ 2 ^ m), 1:])" unfolding degree_smult_eq using c0 by auto
also have "\<dots> = length as" unfolding degree_linear_factors by simp
finally show ?thesis by simp
qed
finally show ?thesis .
qed
end
definition graeffe_one_step :: "'a \<Rightarrow> 'a :: idom poly \<Rightarrow> 'a poly" where
"graeffe_one_step c f = smult c (poly_square_subst (f * f \<circ>\<^sub>p [:0,-1:]))"
lemma graeffe_one_step_code[code]: fixes c :: "'a :: idom"
shows "graeffe_one_step c f = (case poly_even_odd f of (g,h)
\<Rightarrow> smult c (g * g - monom 1 1 * h * h))"
proof -
obtain g h where eo: "poly_even_odd f = (g,h)" by force
from poly_even_odd[OF eo] have fgh: "f = g \<circ>\<^sub>p monom 1 2 + monom 1 1 * h \<circ>\<^sub>p monom 1 2 " by auto
have m2: "monom (1 :: 'a) 2 = [:0,0,1:]" "monom (1 :: 'a) 1 = [:0,1:]"
unfolding coeffs_eq_iff coeffs_monom
by (auto simp add: numeral_2_eq_2)
show ?thesis unfolding eo split graeffe_one_step_def
proof (rule arg_cong[of _ _ "smult c"])
let ?g = "g \<circ>\<^sub>p monom 1 2"
let ?h = "h \<circ>\<^sub>p monom 1 2"
let ?x = "monom (1 :: 'a) 1"
have 2: "2 = Suc (Suc 0)" by simp
have "f * f \<circ>\<^sub>p [:0, - 1:] = (g \<circ>\<^sub>p monom 1 2 + monom 1 1 * h \<circ>\<^sub>p monom 1 2) *
(g \<circ>\<^sub>p monom 1 2 + monom 1 1 * h \<circ>\<^sub>p monom 1 2) \<circ>\<^sub>p [:0, - 1:]" unfolding fgh by simp
also have "(g \<circ>\<^sub>p monom 1 2 + monom 1 1 * h \<circ>\<^sub>p monom 1 2) \<circ>\<^sub>p [:0, - 1:]
= g \<circ>\<^sub>p (monom 1 2 \<circ>\<^sub>p [:0, - 1:]) + monom 1 1 \<circ>\<^sub>p [:0, - 1:] * h \<circ>\<^sub>p (monom 1 2 \<circ>\<^sub>p [:0, - 1:])"
unfolding pcompose_add pcompose_mult pcompose_assoc by simp
also have "monom (1 :: 'a) 2 \<circ>\<^sub>p [:0, - 1:] = monom 1 2" unfolding m2 by auto
also have "?x \<circ>\<^sub>p [:0, - 1:] = [:0, -1:]" unfolding m2 by auto
also have "[:0, - 1:] * h \<circ>\<^sub>p monom 1 2 = (-?x) * ?h" unfolding m2 by simp
also have "(?g + ?x * ?h) * (?g + (- ?x) * ?h) = (?g * ?g - (?x * ?x) * ?h * ?h)"
by (auto simp: field_simps)
also have "?x * ?x = ?x \<circ>\<^sub>p monom 1 2" unfolding mult_monom by (insert m2, simp add: 2)
also have "(?g * ?g - \<dots> * ?h * ?h) = (g * g - ?x * h * h) \<circ>\<^sub>p monom 1 2"
unfolding pcompose_diff pcompose_mult by auto
finally have "poly_square_subst (f * f \<circ>\<^sub>p [:0, - 1:])
= poly_square_subst ((g * g - ?x * h * h) \<circ>\<^sub>p monom 1 2)" by simp
also have "\<dots> = g * g - ?x * h * h" unfolding poly_square_subst by simp
finally show "poly_square_subst (f * f \<circ>\<^sub>p [:0, - 1:]) = g * g - ?x * h * h" .
qed
qed
fun graeffe_poly_impl_main :: "'a \<Rightarrow> 'a :: idom poly \<Rightarrow> nat \<Rightarrow> 'a poly" where
"graeffe_poly_impl_main c f 0 = f"
| "graeffe_poly_impl_main c f (Suc m) = graeffe_one_step c (graeffe_poly_impl_main c f m)"
lemma graeffe_poly_impl_main: assumes "f = smult c (\<Prod>a\<leftarrow>as. [:- a, 1:])"
shows "graeffe_poly_impl_main ((-1)^degree f) f m = graeffe_poly c as m"
proof (induct m)
case 0
show ?case using graeffe_0[OF assms] by simp
next
case (Suc m)
have [simp]: "degree (graeffe_poly c as m) = degree f" unfolding graeffe_poly_def degree_smult_eq assms
degree_linear_factors by auto
from arg_cong[OF Suc, of degree]
show ?case unfolding graeffe_recursion[OF Suc[symmetric]]
by (simp add: graeffe_one_step_def)
qed
definition graeffe_poly_impl :: "'a :: idom poly \<Rightarrow> nat \<Rightarrow> 'a poly" where
"graeffe_poly_impl f = graeffe_poly_impl_main ((-1)^(degree f)) f"
lemma graeffe_poly_impl: assumes "f = smult c (\<Prod>a\<leftarrow>as. [:- a, 1:])"
shows "graeffe_poly_impl f m = graeffe_poly c as m"
using graeffe_poly_impl_main[OF assms] unfolding graeffe_poly_impl_def .
lemma drop_half_map: "drop_half (map f xs) = map f (drop_half xs)"
by (induct xs rule: drop_half.induct, auto)
lemma (in inj_comm_ring_hom) map_poly_poly_square_subst:
"map_poly hom (poly_square_subst f) = poly_square_subst (map_poly hom f)"
unfolding poly_square_subst_def coeffs_map_poly_hom drop_half_map poly_of_list_def
by (rule poly_eqI, auto simp: nth_default_map_eq)
context inj_idom_hom
begin
lemma graeffe_poly_impl_hom:
"map_poly hom (graeffe_poly_impl f m) = graeffe_poly_impl (map_poly hom f) m"
proof -
interpret mh: map_poly_inj_idom_hom..
obtain c where c: "(((- 1) ^ degree f) :: 'a) = c" by auto
have c': "(((- 1) ^ degree f) :: 'b) = hom c" unfolding c[symmetric] by (simp add:hom_distribs)
show ?thesis unfolding graeffe_poly_impl_def degree_map_poly_hom c c'
apply (induct m arbitrary: f; simp)
by (unfold graeffe_one_step_def hom_distribs map_poly_poly_square_subst map_poly_pcompose,simp)
qed
end
lemma graeffe_poly_impl_mahler: "mahler_measure (graeffe_poly_impl f m) = mahler_measure f ^ 2 ^ m"
proof -
let ?c = "complex_of_int"
let ?cc = "map_poly ?c"
let ?f = "?cc f"
note eq = complex_roots(1)[of ?f]
interpret inj_idom_hom complex_of_int by (standard, auto)
show ?thesis
unfolding mahler_measure_def mahler_graeffe[OF eq[symmetric], symmetric]
graeffe_poly_impl[OF eq[symmetric], symmetric] by (simp add: of_int_hom.graeffe_poly_impl_hom)
qed
definition mahler_landau_graeffe_approximation :: "nat \<Rightarrow> nat \<Rightarrow> int poly \<Rightarrow> int" where
"mahler_landau_graeffe_approximation kk dd f = (let
no = sum_list (map (\<lambda> a. a * a) (coeffs f))
in root_int_floor kk (dd * no))"
lemma mahler_landau_graeffe_approximation_core:
assumes g: "g = graeffe_poly_impl f k"
shows "mahler_measure f \<le> root (2 ^ Suc k) (real_of_int (\<Sum>a\<leftarrow>coeffs g. a * a))"
proof -
have "mahler_measure f = root (2^k) (mahler_measure f ^ (2^k))"
by (simp add: real_root_power_cancel mahler_measure_ge_0)
also have "\<dots> = root (2^k) (mahler_measure g)"
unfolding graeffe_poly_impl_mahler g by simp
also have "\<dots> = root (2^k) (root 2 (((mahler_measure g)^2)))"
by (simp add: real_root_power_cancel mahler_measure_ge_0)
also have "\<dots> = root (2^Suc k) (((mahler_measure g)^2))"
by (metis power_Suc2 real_root_mult_exp)
also have "\<dots> \<le> root (2 ^ Suc k) (real_of_int (\<Sum>a\<leftarrow>coeffs g. a * a))"
proof (rule real_root_le_mono, force)
have square_mono: "0 \<le> (x :: real) \<Longrightarrow> x \<le> y \<Longrightarrow> x * x \<le> y * y" for x y
by (simp add: mult_mono')
obtain gs where gs: "coeffs g = gs" by auto
have "(mahler_measure g)\<^sup>2 \<le> real_of_int \<bar>\<Sum>a\<leftarrow>coeffs g. a * a\<bar>"
using square_mono[OF mahler_measure_ge_0 Landau_inequality[of "of_int_poly g", folded mahler_measure_def]]
by (auto simp: power2_eq_square coeffs_map_poly o_def of_int_hom.hom_sum_list)
also have "\<bar>\<Sum>a\<leftarrow>coeffs g. a * a\<bar> = (\<Sum>a\<leftarrow>coeffs g. a * a)" unfolding gs
by (induct gs, auto)
finally show "(mahler_measure g)\<^sup>2 \<le> real_of_int (\<Sum>a\<leftarrow>coeffs g. a * a)" .
qed
finally show "mahler_measure f \<le> root (2 ^ Suc k) (real_of_int (\<Sum>a\<leftarrow>coeffs g. a * a))" .
qed
lemma Landau_inequality_mahler_measure: "mahler_measure f \<le> sqrt (real_of_int (\<Sum>a\<leftarrow>coeffs f. a * a))"
by (rule order.trans[OF mahler_landau_graeffe_approximation_core[OF refl, of _ 0]],
auto simp: graeffe_poly_impl_def sqrt_def)
lemma mahler_landau_graeffe_approximation:
assumes g: "g = graeffe_poly_impl f k" "dd = d^(2^(Suc k))" "kk = 2^(Suc k)"
shows "\<lfloor>real d * mahler_measure f\<rfloor> \<le> mahler_landau_graeffe_approximation kk dd g"
proof -
have id1: "real_of_int (int (d ^ 2 ^ Suc k)) = (real d) ^ 2 ^ Suc k" by simp
have id2: "root (2 ^ Suc k) (real d ^ 2 ^ Suc k) = real d"
by (simp add: real_root_power_cancel)
show ?thesis unfolding mahler_landau_graeffe_approximation_def Let_def root_int_floor of_int_mult g(2-3)
by (rule floor_mono, unfold real_root_mult id1 id2, rule mult_left_mono,
rule mahler_landau_graeffe_approximation_core[OF g(1)], auto)
qed
context
fixes bnd :: nat
begin
(* "dd = d^(2^(Suc k))" "kk = 2^(Suc k)" *)
function mahler_approximation_main :: "nat \<Rightarrow> int \<Rightarrow> int poly \<Rightarrow> int \<Rightarrow> nat \<Rightarrow> nat \<Rightarrow> int" where
"mahler_approximation_main dd c g mm k kk = (let mmm = mahler_landau_graeffe_approximation kk dd g;
new_mm = (if k = 0 then mmm else min mm mmm)
in (if k \<ge> bnd then new_mm else
\<comment> \<open>abort after \<open>bnd\<close> iterations of Graeffe transformation\<close>
mahler_approximation_main (dd * dd) c (graeffe_one_step c g) new_mm (Suc k) (2 * kk)))"
by pat_completeness auto
termination by (relation "measure (\<lambda> (dd,c,f,mm,k,kk). Suc bnd - k)", auto)
declare mahler_approximation_main.simps[simp del]
lemma mahler_approximation_main: assumes "k \<noteq> 0 \<Longrightarrow> \<lfloor>real d * mahler_measure f\<rfloor> \<le> mm"
and "c = (-1)^(degree f)"
and "g = graeffe_poly_impl_main c f k" "dd = d^(2^(Suc k))" "kk = 2^(Suc k)"
shows "\<lfloor>real d * mahler_measure f\<rfloor> \<le> mahler_approximation_main dd c g mm k kk"
using assms
proof (induct c g mm k kk rule: mahler_approximation_main.induct)
case (1 dd c g mm k kk)
let ?df = "\<lfloor>real d * mahler_measure f\<rfloor>"
note dd = 1(5)
note kk = 1(6)
note g = 1(4)
note c = 1(3)
note mm = 1(2)
note IH = 1(1)
note mahl = mahler_approximation_main.simps[of dd c g mm k kk]
define mmm where "mmm = mahler_landau_graeffe_approximation kk dd g"
define new_mm where "new_mm = (if k = 0 then mmm else min mm mmm)"
let ?cond = "bnd \<le> k"
have id: "mahler_approximation_main dd c g mm k kk = (if ?cond then new_mm
else mahler_approximation_main (dd * dd) c (graeffe_one_step c g) new_mm (Suc k) (2 * kk))"
unfolding mahl mmm_def[symmetric] Let_def new_mm_def[symmetric] by simp
have gg: "g = (graeffe_poly_impl f k)" unfolding g graeffe_poly_impl_def c ..
from mahler_landau_graeffe_approximation[OF gg dd kk, folded mmm_def]
have mmm: "?df \<le> mmm" .
with mm have new_mm: "?df \<le> new_mm" unfolding new_mm_def by auto
show ?case
proof (cases ?cond)
case True
show ?thesis unfolding id using True new_mm by auto
next
case False
hence id: "mahler_approximation_main dd c g mm k kk =
mahler_approximation_main (dd * dd) c (graeffe_one_step c g) new_mm (Suc k) (2 * kk)"
unfolding id by auto
have id': "graeffe_one_step c g = graeffe_poly_impl_main c f (Suc k)"
unfolding g by simp
have "dd * dd = d ^ 2 ^ Suc (Suc k)" "2 * kk = 2 ^ Suc (Suc k)" unfolding dd kk
semiring_normalization_rules(26) by auto
from IH[OF mmm_def new_mm_def False new_mm c id' this]
show ?thesis unfolding id .
qed
qed
definition mahler_approximation :: "nat \<Rightarrow> int poly \<Rightarrow> int" where
"mahler_approximation d f = mahler_approximation_main (d * d) ((-1)^(degree f)) f (-1) 0 2"
lemma mahler_approximation: "\<lfloor>real d * mahler_measure f\<rfloor> \<le> mahler_approximation d f"
unfolding mahler_approximation_def
by (rule mahler_approximation_main, auto simp: semiring_normalization_rules(29))
end
end
|