| author | wenzelm | 
| Sat, 25 Nov 2023 20:18:44 +0100 | |
| changeset 79063 | ad7f485195df | 
| parent 74101 | d804e93ae9ff | 
| permissions | -rw-r--r-- | 
| 51240 | 1 | (* Title: HOL/ex/Reflection_Examples.thy | 
| 20319 | 2 | Author: Amine Chaieb, TU Muenchen | 
| 3 | *) | |
| 4 | ||
| 61343 | 5 | section \<open>Examples for generic reflection and reification\<close> | 
| 29650 | 6 | |
| 51093 | 7 | theory Reflection_Examples | 
| 66453 
cc19f7ca2ed6
session-qualified theory imports: isabelle imports -U -i -d '~~/src/Benchmarks' -a;
 wenzelm parents: 
61933diff
changeset | 8 | imports Complex_Main "HOL-Library.Reflection" | 
| 20319 | 9 | begin | 
| 10 | ||
| 61343 | 11 | text \<open>This theory presents two methods: reify and reflection\<close> | 
| 51093 | 12 | |
| 61343 | 13 | text \<open> | 
| 61933 | 14 | Consider an HOL type \<open>\<sigma>\<close>, the structure of which is not recongnisable | 
| 69597 | 15 | on the theory level. This is the case of \<^typ>\<open>bool\<close>, arithmetical terms such as \<^typ>\<open>int\<close>, | 
| 16 | \<^typ>\<open>real\<close> etc \dots In order to implement a simplification on terms of type \<open>\<sigma>\<close> we | |
| 51093 | 17 | often need its structure. Traditionnaly such simplifications are written in ML, | 
| 18 | proofs are synthesized. | |
| 19 | ||
| 61933 | 20 | An other strategy is to declare an HOL datatype \<open>\<tau>\<close> and an HOL function (the | 
| 21 | interpretation) that maps elements of \<open>\<tau>\<close> to elements of \<open>\<sigma>\<close>. | |
| 51093 | 22 | |
| 61933 | 23 | The functionality of \<open>reify\<close> then is, given a term \<open>t\<close> of type \<open>\<sigma>\<close>, | 
| 24 | to compute a term \<open>s\<close> of type \<open>\<tau>\<close>. For this it needs equations for the | |
| 51093 | 25 | interpretation. | 
| 20319 | 26 | |
| 61933 | 27 | N.B: All the interpretations supported by \<open>reify\<close> must have the type | 
| 28 | \<open>'a list \<Rightarrow> \<tau> \<Rightarrow> \<sigma>\<close>. The method \<open>reify\<close> can also be told which subterm | |
| 29 | of the current subgoal should be reified. The general call for \<open>reify\<close> is | |
| 30 | \<open>reify eqs (t)\<close>, where \<open>eqs\<close> are the defining equations of the interpretation | |
| 31 | and \<open>(t)\<close> is an optional parameter which specifies the subterm to which reification | |
| 32 | should be applied to. If \<open>(t)\<close> is abscent, \<open>reify\<close> tries to reify the whole | |
| 51093 | 33 | subgoal. | 
| 20337 | 34 | |
| 61933 | 35 | The method \<open>reflection\<close> uses \<open>reify\<close> and has a very similar signature: | 
| 36 | \<open>reflection corr_thm eqs (t)\<close>. Here again \<open>eqs\<close> and \<open>(t)\<close> | |
| 37 | are as described above and \<open>corr_thm\<close> is a theorem proving | |
| 69597 | 38 | \<^prop>\<open>I vs (f t) = I vs t\<close>. We assume that \<open>I\<close> is the interpretation | 
| 61933 | 39 | and \<open>f\<close> is some useful and executable simplification of type \<open>\<tau> \<Rightarrow> \<tau>\<close>. | 
| 69597 | 40 | The method \<open>reflection\<close> applies reification and hence the theorem \<^prop>\<open>t = I xs s\<close> | 
| 41 | and hence using \<open>corr_thm\<close> derives \<^prop>\<open>t = I xs (f s)\<close>. It then uses | |
| 42 | normalization by equational rewriting to prove \<^prop>\<open>f s = s'\<close> which almost finishes | |
| 43 | the proof of \<^prop>\<open>t = t'\<close> where \<^prop>\<open>I xs s' = t'\<close>. | |
| 61343 | 44 | \<close> | 
| 20337 | 45 | |
| 61343 | 46 | text \<open>Example 1 : Propositional formulae and NNF.\<close> | 
| 61933 | 47 | text \<open>The type \<open>fm\<close> represents simple propositional formulae:\<close> | 
| 20337 | 48 | |
| 58310 | 49 | datatype form = TrueF | FalseF | Less nat nat | 
| 51093 | 50 | | And form form | Or form form | Neg form | ExQ form | 
| 23547 | 51 | |
| 51093 | 52 | primrec interp :: "form \<Rightarrow> ('a::ord) list \<Rightarrow> bool"
 | 
| 53 | where | |
| 54 | "interp TrueF vs \<longleftrightarrow> True" | |
| 55 | | "interp FalseF vs \<longleftrightarrow> False" | |
| 56 | | "interp (Less i j) vs \<longleftrightarrow> vs ! i < vs ! j" | |
| 57 | | "interp (And f1 f2) vs \<longleftrightarrow> interp f1 vs \<and> interp f2 vs" | |
| 58 | | "interp (Or f1 f2) vs \<longleftrightarrow> interp f1 vs \<or> interp f2 vs" | |
| 59 | | "interp (Neg f) vs \<longleftrightarrow> \<not> interp f vs" | |
| 60 | | "interp (ExQ f) vs \<longleftrightarrow> (\<exists>v. interp f (v # vs))" | |
| 23547 | 61 | |
| 23608 | 62 | lemmas interp_reify_eqs = interp.simps | 
| 51093 | 63 | declare interp_reify_eqs [reify] | 
| 23547 | 64 | |
| 51093 | 65 | lemma "\<exists>x. x < y \<and> x < z" | 
| 66 | apply reify | |
| 23608 | 67 | oops | 
| 23547 | 68 | |
| 74101 | 69 | datatype fm = And fm fm | Or fm fm | Imp fm fm | Iff fm fm | Not fm | At nat | 
| 20319 | 70 | |
| 51093 | 71 | primrec Ifm :: "fm \<Rightarrow> bool list \<Rightarrow> bool" | 
| 72 | where | |
| 73 | "Ifm (At n) vs \<longleftrightarrow> vs ! n" | |
| 74 | | "Ifm (And p q) vs \<longleftrightarrow> Ifm p vs \<and> Ifm q vs" | |
| 75 | | "Ifm (Or p q) vs \<longleftrightarrow> Ifm p vs \<or> Ifm q vs" | |
| 76 | | "Ifm (Imp p q) vs \<longleftrightarrow> Ifm p vs \<longrightarrow> Ifm q vs" | |
| 77 | | "Ifm (Iff p q) vs \<longleftrightarrow> Ifm p vs = Ifm q vs" | |
| 74101 | 78 | | "Ifm (Not p) vs \<longleftrightarrow> \<not> Ifm p vs" | 
| 20319 | 79 | |
| 51093 | 80 | lemma "Q \<longrightarrow> (D \<and> F \<and> ((\<not> D) \<and> (\<not> F)))" | 
| 81 | apply (reify Ifm.simps) | |
| 23547 | 82 | oops | 
| 83 | ||
| 69597 | 84 | text \<open>Method \<open>reify\<close> maps a \<^typ>\<open>bool\<close> to an \<^typ>\<open>fm\<close>. For this it needs the | 
| 61933 | 85 | semantics of \<open>fm\<close>, i.e.\ the rewrite rules in \<open>Ifm.simps\<close>.\<close> | 
| 23547 | 86 | |
| 61343 | 87 | text \<open>You can also just pick up a subterm to reify.\<close> | 
| 51093 | 88 | lemma "Q \<longrightarrow> (D \<and> F \<and> ((\<not> D) \<and> (\<not> F)))" | 
| 89 |   apply (reify Ifm.simps ("((\<not> D) \<and> (\<not> F))"))
 | |
| 23547 | 90 | oops | 
| 91 | ||
| 61343 | 92 | text \<open>Let's perform NNF. This is a version that tends to generate disjunctions\<close> | 
| 51093 | 93 | primrec fmsize :: "fm \<Rightarrow> nat" | 
| 94 | where | |
| 20319 | 95 | "fmsize (At n) = 1" | 
| 74101 | 96 | | "fmsize (Not p) = 1 + fmsize p" | 
| 35419 | 97 | | "fmsize (And p q) = 1 + fmsize p + fmsize q" | 
| 98 | | "fmsize (Or p q) = 1 + fmsize p + fmsize q" | |
| 99 | | "fmsize (Imp p q) = 2 + fmsize p + fmsize q" | |
| 100 | | "fmsize (Iff p q) = 2 + 2* fmsize p + 2* fmsize q" | |
| 101 | ||
| 102 | lemma [measure_function]: "is_measure fmsize" .. | |
| 20319 | 103 | |
| 35419 | 104 | fun nnf :: "fm \<Rightarrow> fm" | 
| 105 | where | |
| 20319 | 106 | "nnf (At n) = At n" | 
| 35419 | 107 | | "nnf (And p q) = And (nnf p) (nnf q)" | 
| 108 | | "nnf (Or p q) = Or (nnf p) (nnf q)" | |
| 74101 | 109 | | "nnf (Imp p q) = Or (nnf (Not p)) (nnf q)" | 
| 110 | | "nnf (Iff p q) = Or (And (nnf p) (nnf q)) (And (nnf (Not p)) (nnf (Not q)))" | |
| 111 | | "nnf (Not (And p q)) = Or (nnf (Not p)) (nnf (Not q))" | |
| 112 | | "nnf (Not (Or p q)) = And (nnf (Not p)) (nnf (Not q))" | |
| 113 | | "nnf (Not (Imp p q)) = And (nnf p) (nnf (Not q))" | |
| 114 | | "nnf (Not (Iff p q)) = Or (And (nnf p) (nnf (Not q))) (And (nnf (Not p)) (nnf q))" | |
| 115 | | "nnf (Not (Not p)) = nnf p" | |
| 116 | | "nnf (Not p) = Not p" | |
| 20319 | 117 | |
| 69597 | 118 | text \<open>The correctness theorem of \<^const>\<open>nnf\<close>: it preserves the semantics of \<^typ>\<open>fm\<close>\<close> | 
| 51093 | 119 | lemma nnf [reflection]: | 
| 120 | "Ifm (nnf p) vs = Ifm p vs" | |
| 20319 | 121 | by (induct p rule: nnf.induct) auto | 
| 122 | ||
| 69597 | 123 | text \<open>Now let's perform NNF using our \<^const>\<open>nnf\<close> function defined above. First to the | 
| 61343 | 124 | whole subgoal.\<close> | 
| 51093 | 125 | lemma "A \<noteq> B \<and> (B \<longrightarrow> A \<noteq> (B \<or> C \<and> (B \<longrightarrow> A \<or> D))) \<longrightarrow> A \<or> B \<and> D" | 
| 126 | apply (reflection Ifm.simps) | |
| 20319 | 127 | oops | 
| 128 | ||
| 61343 | 129 | text \<open>Now we specify on which subterm it should be applied\<close> | 
| 51093 | 130 | lemma "A \<noteq> B \<and> (B \<longrightarrow> A \<noteq> (B \<or> C \<and> (B \<longrightarrow> A \<or> D))) \<longrightarrow> A \<or> B \<and> D" | 
| 131 | apply (reflection Ifm.simps only: "B \<or> C \<and> (B \<longrightarrow> A \<or> D)") | |
| 20319 | 132 | oops | 
| 133 | ||
| 134 | ||
| 61343 | 135 | text \<open>Example 2: Simple arithmetic formulae\<close> | 
| 20319 | 136 | |
| 61933 | 137 | text \<open>The type \<open>num\<close> reflects linear expressions over natural number\<close> | 
| 58310 | 138 | datatype num = C nat | Add num num | Mul nat num | Var nat | CN nat nat num | 
| 20319 | 139 | |
| 61343 | 140 | text \<open>This is just technical to make recursive definitions easier.\<close> | 
| 35419 | 141 | primrec num_size :: "num \<Rightarrow> nat" | 
| 142 | where | |
| 20319 | 143 | "num_size (C c) = 1" | 
| 35419 | 144 | | "num_size (Var n) = 1" | 
| 145 | | "num_size (Add a b) = 1 + num_size a + num_size b" | |
| 146 | | "num_size (Mul c a) = 1 + num_size a" | |
| 147 | | "num_size (CN n c a) = 4 + num_size a " | |
| 20319 | 148 | |
| 51093 | 149 | lemma [measure_function]: "is_measure num_size" .. | 
| 150 | ||
| 61343 | 151 | text \<open>The semantics of num\<close> | 
| 35419 | 152 | primrec Inum:: "num \<Rightarrow> nat list \<Rightarrow> nat" | 
| 153 | where | |
| 23624 
82091387f6d7
The order for parameter for interpretation is now inversted:
 chaieb parents: 
23608diff
changeset | 154 | Inum_C : "Inum (C i) vs = i" | 
| 35419 | 155 | | Inum_Var: "Inum (Var n) vs = vs!n" | 
| 156 | | Inum_Add: "Inum (Add s t) vs = Inum s vs + Inum t vs " | |
| 157 | | Inum_Mul: "Inum (Mul c t) vs = c * Inum t vs " | |
| 158 | | Inum_CN : "Inum (CN n c t) vs = c*(vs!n) + Inum t vs " | |
| 20319 | 159 | |
| 61343 | 160 | text \<open>Let's reify some nat expressions \dots\<close> | 
| 51093 | 161 | lemma "4 * (2 * x + (y::nat)) + f a \<noteq> 0" | 
| 162 |   apply (reify Inum.simps ("4 * (2 * x + (y::nat)) + f a"))
 | |
| 20319 | 163 | oops | 
| 61933 | 164 | text \<open>We're in a bad situation! \<open>x\<close>, \<open>y\<close> and \<open>f\<close> have been recongnized | 
| 51093 | 165 | as constants, which is correct but does not correspond to our intuition of the constructor C. | 
| 61343 | 166 | It should encapsulate constants, i.e. numbers, i.e. numerals.\<close> | 
| 20337 | 167 | |
| 61933 | 168 | text \<open>So let's leave the \<open>Inum_C\<close> equation at the end and see what happens \dots\<close> | 
| 51093 | 169 | lemma "4 * (2 * x + (y::nat)) \<noteq> 0" | 
| 170 |   apply (reify Inum_Var Inum_Add Inum_Mul Inum_CN Inum_C ("4 * (2 * x + (y::nat))"))
 | |
| 20319 | 171 | oops | 
| 61933 | 172 | text \<open>Hm, let's specialize \<open>Inum_C\<close> with numerals.\<close> | 
| 20319 | 173 | |
| 47108 
2a1953f0d20d
merged fork with new numeral representation (see NEWS)
 huffman parents: 
41413diff
changeset | 174 | lemma Inum_number: "Inum (C (numeral t)) vs = numeral t" by simp | 
| 20319 | 175 | lemmas Inum_eqs = Inum_Var Inum_Add Inum_Mul Inum_CN Inum_number | 
| 176 | ||
| 61343 | 177 | text \<open>Second attempt\<close> | 
| 51093 | 178 | lemma "1 * (2 * x + (y::nat)) \<noteq> 0" | 
| 179 |   apply (reify Inum_eqs ("1 * (2 * x + (y::nat))"))
 | |
| 20319 | 180 | oops | 
| 51093 | 181 | |
| 61343 | 182 | text\<open>That was fine, so let's try another one \dots\<close> | 
| 20319 | 183 | |
| 51093 | 184 | lemma "1 * (2 * x + (y::nat) + 0 + 1) \<noteq> 0" | 
| 185 |   apply (reify Inum_eqs ("1 * (2 * x + (y::nat) + 0 + 1)"))
 | |
| 20319 | 186 | oops | 
| 51093 | 187 | |
| 61933 | 188 | text \<open>Oh!! 0 is not a variable \dots\ Oh! 0 is not a \<open>numeral\<close> \dots\ thing. | 
| 61343 | 189 | The same for 1. So let's add those equations, too.\<close> | 
| 20319 | 190 | |
| 23624 
82091387f6d7
The order for parameter for interpretation is now inversted:
 chaieb parents: 
23608diff
changeset | 191 | lemma Inum_01: "Inum (C 0) vs = 0" "Inum (C 1) vs = 1" "Inum (C(Suc n)) vs = Suc n" | 
| 51093 | 192 | by simp_all | 
| 20337 | 193 | |
| 20319 | 194 | lemmas Inum_eqs'= Inum_eqs Inum_01 | 
| 20337 | 195 | |
| 61343 | 196 | text\<open>Third attempt:\<close> | 
| 20319 | 197 | |
| 51093 | 198 | lemma "1 * (2 * x + (y::nat) + 0 + 1) \<noteq> 0" | 
| 199 |   apply (reify Inum_eqs' ("1 * (2 * x + (y::nat) + 0 + 1)"))
 | |
| 20319 | 200 | oops | 
| 51093 | 201 | |
| 69597 | 202 | text \<open>Okay, let's try reflection. Some simplifications on \<^typ>\<open>num\<close> follow. You can | 
| 61933 | 203 | skim until the main theorem \<open>linum\<close>.\<close> | 
| 51093 | 204 | |
| 35419 | 205 | fun lin_add :: "num \<Rightarrow> num \<Rightarrow> num" | 
| 206 | where | |
| 207 | "lin_add (CN n1 c1 r1) (CN n2 c2 r2) = | |
| 51093 | 208 | (if n1 = n2 then | 
| 209 | (let c = c1 + c2 | |
| 210 | in (if c = 0 then lin_add r1 r2 else CN n1 c (lin_add r1 r2))) | |
| 211 | else if n1 \<le> n2 then (CN n1 c1 (lin_add r1 (CN n2 c2 r2))) | |
| 212 | else (CN n2 c2 (lin_add (CN n1 c1 r1) r2)))" | |
| 35419 | 213 | | "lin_add (CN n1 c1 r1) t = CN n1 c1 (lin_add r1 t)" | 
| 214 | | "lin_add t (CN n2 c2 r2) = CN n2 c2 (lin_add t r2)" | |
| 51093 | 215 | | "lin_add (C b1) (C b2) = C (b1 + b2)" | 
| 35419 | 216 | | "lin_add a b = Add a b" | 
| 51093 | 217 | |
| 218 | lemma lin_add: | |
| 219 | "Inum (lin_add t s) bs = Inum (Add t s) bs" | |
| 220 | apply (induct t s rule: lin_add.induct, simp_all add: Let_def) | |
| 221 | apply (case_tac "c1+c2 = 0",case_tac "n1 \<le> n2", simp_all) | |
| 222 | apply (case_tac "n1 = n2", simp_all add: algebra_simps) | |
| 223 | done | |
| 20319 | 224 | |
| 35419 | 225 | fun lin_mul :: "num \<Rightarrow> nat \<Rightarrow> num" | 
| 226 | where | |
| 51093 | 227 | "lin_mul (C j) i = C (i * j)" | 
| 228 | | "lin_mul (CN n c a) i = (if i=0 then (C 0) else CN n (i * c) (lin_mul a i))" | |
| 35419 | 229 | | "lin_mul t i = (Mul i t)" | 
| 20319 | 230 | |
| 51093 | 231 | lemma lin_mul: | 
| 232 | "Inum (lin_mul t i) bs = Inum (Mul i t) bs" | |
| 233 | by (induct t i rule: lin_mul.induct) (auto simp add: algebra_simps) | |
| 35419 | 234 | |
| 235 | fun linum:: "num \<Rightarrow> num" | |
| 236 | where | |
| 20319 | 237 | "linum (C b) = C b" | 
| 35419 | 238 | | "linum (Var n) = CN n 1 (C 0)" | 
| 239 | | "linum (Add t s) = lin_add (linum t) (linum s)" | |
| 240 | | "linum (Mul c t) = lin_mul (linum t) c" | |
| 241 | | "linum (CN n c t) = lin_add (linum (Mul c (Var n))) (linum t)" | |
| 20319 | 242 | |
| 51093 | 243 | lemma linum [reflection]: | 
| 244 | "Inum (linum t) bs = Inum t bs" | |
| 245 | by (induct t rule: linum.induct) (simp_all add: lin_mul lin_add) | |
| 20319 | 246 | |
| 61343 | 247 | text \<open>Now we can use linum to simplify nat terms using reflection\<close> | 
| 51093 | 248 | |
| 249 | lemma "Suc (Suc 1) * (x + Suc 1 * y) = 3 * x + 6 * y" | |
| 250 | apply (reflection Inum_eqs' only: "Suc (Suc 1) * (x + Suc 1 * y)") | |
| 20319 | 251 | oops | 
| 252 | ||
| 61343 | 253 | text \<open>Let's lift this to formulae and see what happens\<close> | 
| 20319 | 254 | |
| 58310 | 255 | datatype aform = Lt num num | Eq num num | Ge num num | NEq num num | | 
| 20319 | 256 | Conj aform aform | Disj aform aform | NEG aform | T | F | 
| 35419 | 257 | |
| 258 | primrec linaformsize:: "aform \<Rightarrow> nat" | |
| 259 | where | |
| 20319 | 260 | "linaformsize T = 1" | 
| 35419 | 261 | | "linaformsize F = 1" | 
| 262 | | "linaformsize (Lt a b) = 1" | |
| 263 | | "linaformsize (Ge a b) = 1" | |
| 264 | | "linaformsize (Eq a b) = 1" | |
| 265 | | "linaformsize (NEq a b) = 1" | |
| 266 | | "linaformsize (NEG p) = 2 + linaformsize p" | |
| 267 | | "linaformsize (Conj p q) = 1 + linaformsize p + linaformsize q" | |
| 268 | | "linaformsize (Disj p q) = 1 + linaformsize p + linaformsize q" | |
| 20319 | 269 | |
| 35419 | 270 | lemma [measure_function]: "is_measure linaformsize" .. | 
| 20319 | 271 | |
| 35419 | 272 | primrec is_aform :: "aform => nat list => bool" | 
| 273 | where | |
| 28866 
30cd9d89a0fb
reactivated some dead theories (based on hints by Mark Hillebrand);
 wenzelm parents: 
23650diff
changeset | 274 | "is_aform T vs = True" | 
| 35419 | 275 | | "is_aform F vs = False" | 
| 276 | | "is_aform (Lt a b) vs = (Inum a vs < Inum b vs)" | |
| 277 | | "is_aform (Eq a b) vs = (Inum a vs = Inum b vs)" | |
| 278 | | "is_aform (Ge a b) vs = (Inum a vs \<ge> Inum b vs)" | |
| 279 | | "is_aform (NEq a b) vs = (Inum a vs \<noteq> Inum b vs)" | |
| 280 | | "is_aform (NEG p) vs = (\<not> (is_aform p vs))" | |
| 281 | | "is_aform (Conj p q) vs = (is_aform p vs \<and> is_aform q vs)" | |
| 282 | | "is_aform (Disj p q) vs = (is_aform p vs \<or> is_aform q vs)" | |
| 20319 | 283 | |
| 61343 | 284 | text\<open>Let's reify and do reflection\<close> | 
| 51093 | 285 | lemma "(3::nat) * x + t < 0 \<and> (2 * x + y \<noteq> 17)" | 
| 286 | apply (reify Inum_eqs' is_aform.simps) | |
| 20319 | 287 | oops | 
| 288 | ||
| 61343 | 289 | text \<open>Note that reification handles several interpretations at the same time\<close> | 
| 51093 | 290 | lemma "(3::nat) * x + t < 0 \<and> x * x + t * x + 3 + 1 = z * t * 4 * z \<or> x + x + 1 < 0" | 
| 291 | apply (reflection Inum_eqs' is_aform.simps only: "x + x + 1") | |
| 20319 | 292 | oops | 
| 293 | ||
| 61343 | 294 | text \<open>For reflection we now define a simple transformation on aform: NNF + linum on atoms\<close> | 
| 35419 | 295 | |
| 296 | fun linaform:: "aform \<Rightarrow> aform" | |
| 297 | where | |
| 20319 | 298 | "linaform (Lt s t) = Lt (linum s) (linum t)" | 
| 35419 | 299 | | "linaform (Eq s t) = Eq (linum s) (linum t)" | 
| 300 | | "linaform (Ge s t) = Ge (linum s) (linum t)" | |
| 301 | | "linaform (NEq s t) = NEq (linum s) (linum t)" | |
| 302 | | "linaform (Conj p q) = Conj (linaform p) (linaform q)" | |
| 303 | | "linaform (Disj p q) = Disj (linaform p) (linaform q)" | |
| 304 | | "linaform (NEG T) = F" | |
| 305 | | "linaform (NEG F) = T" | |
| 306 | | "linaform (NEG (Lt a b)) = Ge a b" | |
| 307 | | "linaform (NEG (Ge a b)) = Lt a b" | |
| 308 | | "linaform (NEG (Eq a b)) = NEq a b" | |
| 309 | | "linaform (NEG (NEq a b)) = Eq a b" | |
| 310 | | "linaform (NEG (NEG p)) = linaform p" | |
| 311 | | "linaform (NEG (Conj p q)) = Disj (linaform (NEG p)) (linaform (NEG q))" | |
| 312 | | "linaform (NEG (Disj p q)) = Conj (linaform (NEG p)) (linaform (NEG q))" | |
| 313 | | "linaform p = p" | |
| 20319 | 314 | |
| 28866 
30cd9d89a0fb
reactivated some dead theories (based on hints by Mark Hillebrand);
 wenzelm parents: 
23650diff
changeset | 315 | lemma linaform: "is_aform (linaform p) vs = is_aform p vs" | 
| 
30cd9d89a0fb
reactivated some dead theories (based on hints by Mark Hillebrand);
 wenzelm parents: 
23650diff
changeset | 316 | by (induct p rule: linaform.induct) (auto simp add: linum) | 
| 20319 | 317 | |
| 51093 | 318 | lemma "(Suc (Suc (Suc 0)) * ((x::nat) + Suc (Suc 0)) + Suc (Suc (Suc 0)) * | 
| 319 | (Suc (Suc (Suc 0))) * ((x::nat) + Suc (Suc 0))) < 0 \<and> Suc 0 + Suc 0 < 0" | |
| 320 | apply (reflection Inum_eqs' is_aform.simps rules: linaform) | |
| 23650 
0a6a719d24d5
Show the use of reflection attribute; fixed comments about the order of absorbing equations : f (C x) = x ; now automaically tried as last;
 chaieb parents: 
23624diff
changeset | 321 | oops | 
| 
0a6a719d24d5
Show the use of reflection attribute; fixed comments about the order of absorbing equations : f (C x) = x ; now automaically tried as last;
 chaieb parents: 
23624diff
changeset | 322 | |
| 51093 | 323 | declare linaform [reflection] | 
| 324 | ||
| 325 | lemma "(Suc (Suc (Suc 0)) * ((x::nat) + Suc (Suc 0)) + Suc (Suc (Suc 0)) * | |
| 326 | (Suc (Suc (Suc 0))) * ((x::nat) + Suc (Suc 0))) < 0 \<and> Suc 0 + Suc 0 < 0" | |
| 327 | apply (reflection Inum_eqs' is_aform.simps) | |
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 328 | oops | 
| 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 329 | |
| 61343 | 330 | text \<open>We now give an example where interpretaions have zero or more than only | 
| 51093 | 331 | one envornement of different types and show that automatic reification also deals with | 
| 61343 | 332 | bindings\<close> | 
| 51093 | 333 | |
| 58310 | 334 | datatype rb = BC bool | BAnd rb rb | BOr rb rb | 
| 51093 | 335 | |
| 35419 | 336 | primrec Irb :: "rb \<Rightarrow> bool" | 
| 337 | where | |
| 51093 | 338 | "Irb (BC p) \<longleftrightarrow> p" | 
| 339 | | "Irb (BAnd s t) \<longleftrightarrow> Irb s \<and> Irb t" | |
| 340 | | "Irb (BOr s t) \<longleftrightarrow> Irb s \<or> Irb t" | |
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 341 | |
| 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 342 | lemma "A \<and> (B \<or> D \<and> B) \<and> A \<and> (B \<or> D \<and> B) \<or> A \<and> (B \<or> D \<and> B) \<or> A \<and> (B \<or> D \<and> B)" | 
| 51093 | 343 | apply (reify Irb.simps) | 
| 20319 | 344 | oops | 
| 345 | ||
| 58310 | 346 | datatype rint = IC int | IVar nat | IAdd rint rint | IMult rint rint | 
| 51093 | 347 | | INeg rint | ISub rint rint | 
| 23650 
0a6a719d24d5
Show the use of reflection attribute; fixed comments about the order of absorbing equations : f (C x) = x ; now automaically tried as last;
 chaieb parents: 
23624diff
changeset | 348 | |
| 35419 | 349 | primrec Irint :: "rint \<Rightarrow> int list \<Rightarrow> int" | 
| 350 | where | |
| 51093 | 351 | Irint_Var: "Irint (IVar n) vs = vs ! n" | 
| 35419 | 352 | | Irint_Neg: "Irint (INeg t) vs = - Irint t vs" | 
| 353 | | Irint_Add: "Irint (IAdd s t) vs = Irint s vs + Irint t vs" | |
| 354 | | Irint_Sub: "Irint (ISub s t) vs = Irint s vs - Irint t vs" | |
| 355 | | Irint_Mult: "Irint (IMult s t) vs = Irint s vs * Irint t vs" | |
| 356 | | Irint_C: "Irint (IC i) vs = i" | |
| 51093 | 357 | |
| 23624 
82091387f6d7
The order for parameter for interpretation is now inversted:
 chaieb parents: 
23608diff
changeset | 358 | lemma Irint_C0: "Irint (IC 0) vs = 0" | 
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 359 | by simp | 
| 51093 | 360 | |
| 23624 
82091387f6d7
The order for parameter for interpretation is now inversted:
 chaieb parents: 
23608diff
changeset | 361 | lemma Irint_C1: "Irint (IC 1) vs = 1" | 
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 362 | by simp | 
| 51093 | 363 | |
| 47108 
2a1953f0d20d
merged fork with new numeral representation (see NEWS)
 huffman parents: 
41413diff
changeset | 364 | lemma Irint_Cnumeral: "Irint (IC (numeral x)) vs = numeral x" | 
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 365 | by simp | 
| 51093 | 366 | |
| 47108 
2a1953f0d20d
merged fork with new numeral representation (see NEWS)
 huffman parents: 
41413diff
changeset | 367 | lemmas Irint_simps = Irint_Var Irint_Neg Irint_Add Irint_Sub Irint_Mult Irint_C0 Irint_C1 Irint_Cnumeral | 
| 51093 | 368 | |
| 369 | lemma "(3::int) * x + y * y - 9 + (- z) = 0" | |
| 370 |   apply (reify Irint_simps ("(3::int) * x + y * y - 9 + (- z)"))
 | |
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 371 | oops | 
| 51093 | 372 | |
| 58310 | 373 | datatype rlist = LVar nat | LEmpty | LCons rint rlist | LAppend rlist rlist | 
| 51093 | 374 | |
| 375 | primrec Irlist :: "rlist \<Rightarrow> int list \<Rightarrow> int list list \<Rightarrow> int list" | |
| 35419 | 376 | where | 
| 23624 
82091387f6d7
The order for parameter for interpretation is now inversted:
 chaieb parents: 
23608diff
changeset | 377 | "Irlist (LEmpty) is vs = []" | 
| 51093 | 378 | | "Irlist (LVar n) is vs = vs ! n" | 
| 379 | | "Irlist (LCons i t) is vs = Irint i is # Irlist t is vs" | |
| 380 | | "Irlist (LAppend s t) is vs = Irlist s is vs @ Irlist t is vs" | |
| 381 | ||
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 382 | lemma "[(1::int)] = []" | 
| 51093 | 383 |   apply (reify Irlist.simps Irint_simps ("[1] :: int list"))
 | 
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 384 | oops | 
| 20374 | 385 | |
| 51093 | 386 | lemma "([(3::int) * x + y * y - 9 + (- z)] @ []) @ xs = [y * y - z - 9 + (3::int) * x]" | 
| 387 |   apply (reify Irlist.simps Irint_simps ("([(3::int) * x + y * y - 9 + (- z)] @ []) @ xs"))
 | |
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 388 | oops | 
| 20374 | 389 | |
| 58310 | 390 | datatype rnat = NC nat| NVar nat| NSuc rnat | NAdd rnat rnat | NMult rnat rnat | 
| 51093 | 391 | | NNeg rnat | NSub rnat rnat | Nlgth rlist | 
| 392 | ||
| 393 | primrec Irnat :: "rnat \<Rightarrow> int list \<Rightarrow> int list list \<Rightarrow> nat list \<Rightarrow> nat" | |
| 35419 | 394 | where | 
| 395 | Irnat_Suc: "Irnat (NSuc t) is ls vs = Suc (Irnat t is ls vs)" | |
| 51093 | 396 | | Irnat_Var: "Irnat (NVar n) is ls vs = vs ! n" | 
| 35419 | 397 | | Irnat_Neg: "Irnat (NNeg t) is ls vs = 0" | 
| 398 | | Irnat_Add: "Irnat (NAdd s t) is ls vs = Irnat s is ls vs + Irnat t is ls vs" | |
| 399 | | Irnat_Sub: "Irnat (NSub s t) is ls vs = Irnat s is ls vs - Irnat t is ls vs" | |
| 400 | | Irnat_Mult: "Irnat (NMult s t) is ls vs = Irnat s is ls vs * Irnat t is ls vs" | |
| 401 | | Irnat_lgth: "Irnat (Nlgth rxs) is ls vs = length (Irlist rxs is ls)" | |
| 402 | | Irnat_C: "Irnat (NC i) is ls vs = i" | |
| 51093 | 403 | |
| 23624 
82091387f6d7
The order for parameter for interpretation is now inversted:
 chaieb parents: 
23608diff
changeset | 404 | lemma Irnat_C0: "Irnat (NC 0) is ls vs = 0" | 
| 51093 | 405 | by simp | 
| 406 | ||
| 23624 
82091387f6d7
The order for parameter for interpretation is now inversted:
 chaieb parents: 
23608diff
changeset | 407 | lemma Irnat_C1: "Irnat (NC 1) is ls vs = 1" | 
| 51093 | 408 | by simp | 
| 409 | ||
| 47108 
2a1953f0d20d
merged fork with new numeral representation (see NEWS)
 huffman parents: 
41413diff
changeset | 410 | lemma Irnat_Cnumeral: "Irnat (NC (numeral x)) is ls vs = numeral x" | 
| 51093 | 411 | by simp | 
| 412 | ||
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 413 | lemmas Irnat_simps = Irnat_Suc Irnat_Var Irnat_Neg Irnat_Add Irnat_Sub Irnat_Mult Irnat_lgth | 
| 47108 
2a1953f0d20d
merged fork with new numeral representation (see NEWS)
 huffman parents: 
41413diff
changeset | 414 | Irnat_C0 Irnat_C1 Irnat_Cnumeral | 
| 51093 | 415 | |
| 416 | lemma "Suc n * length (([(3::int) * x + y * y - 9 + (- z)] @ []) @ xs) = length xs" | |
| 417 | apply (reify Irnat_simps Irlist.simps Irint_simps | |
| 418 |      ("Suc n * length (([(3::int) * x + y * y - 9 + (- z)] @ []) @ xs)"))
 | |
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 419 | oops | 
| 51093 | 420 | |
| 58310 | 421 | datatype rifm = RT | RF | RVar nat | 
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 422 | | RNLT rnat rnat | RNILT rnat rint | RNEQ rnat rnat | 
| 51093 | 423 | | RAnd rifm rifm | ROr rifm rifm | RImp rifm rifm| RIff rifm rifm | 
| 424 | | RNEX rifm | RIEX rifm | RLEX rifm | RNALL rifm | RIALL rifm | RLALL rifm | |
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 425 | | RBEX rifm | RBALL rifm | 
| 23624 
82091387f6d7
The order for parameter for interpretation is now inversted:
 chaieb parents: 
23608diff
changeset | 426 | |
| 35419 | 427 | primrec Irifm :: "rifm \<Rightarrow> bool list \<Rightarrow> int list \<Rightarrow> (int list) list \<Rightarrow> nat list \<Rightarrow> bool" | 
| 428 | where | |
| 51093 | 429 | "Irifm RT ps is ls ns \<longleftrightarrow> True" | 
| 430 | | "Irifm RF ps is ls ns \<longleftrightarrow> False" | |
| 431 | | "Irifm (RVar n) ps is ls ns \<longleftrightarrow> ps ! n" | |
| 432 | | "Irifm (RNLT s t) ps is ls ns \<longleftrightarrow> Irnat s is ls ns < Irnat t is ls ns" | |
| 433 | | "Irifm (RNILT s t) ps is ls ns \<longleftrightarrow> int (Irnat s is ls ns) < Irint t is" | |
| 434 | | "Irifm (RNEQ s t) ps is ls ns \<longleftrightarrow> Irnat s is ls ns = Irnat t is ls ns" | |
| 435 | | "Irifm (RAnd p q) ps is ls ns \<longleftrightarrow> Irifm p ps is ls ns \<and> Irifm q ps is ls ns" | |
| 436 | | "Irifm (ROr p q) ps is ls ns \<longleftrightarrow> Irifm p ps is ls ns \<or> Irifm q ps is ls ns" | |
| 437 | | "Irifm (RImp p q) ps is ls ns \<longleftrightarrow> Irifm p ps is ls ns \<longrightarrow> Irifm q ps is ls ns" | |
| 438 | | "Irifm (RIff p q) ps is ls ns \<longleftrightarrow> Irifm p ps is ls ns = Irifm q ps is ls ns" | |
| 439 | | "Irifm (RNEX p) ps is ls ns \<longleftrightarrow> (\<exists>x. Irifm p ps is ls (x # ns))" | |
| 440 | | "Irifm (RIEX p) ps is ls ns \<longleftrightarrow> (\<exists>x. Irifm p ps (x # is) ls ns)" | |
| 441 | | "Irifm (RLEX p) ps is ls ns \<longleftrightarrow> (\<exists>x. Irifm p ps is (x # ls) ns)" | |
| 442 | | "Irifm (RBEX p) ps is ls ns \<longleftrightarrow> (\<exists>x. Irifm p (x # ps) is ls ns)" | |
| 443 | | "Irifm (RNALL p) ps is ls ns \<longleftrightarrow> (\<forall>x. Irifm p ps is ls (x#ns))" | |
| 444 | | "Irifm (RIALL p) ps is ls ns \<longleftrightarrow> (\<forall>x. Irifm p ps (x # is) ls ns)" | |
| 445 | | "Irifm (RLALL p) ps is ls ns \<longleftrightarrow> (\<forall>x. Irifm p ps is (x#ls) ns)" | |
| 446 | | "Irifm (RBALL p) ps is ls ns \<longleftrightarrow> (\<forall>x. Irifm p (x # ps) is ls ns)" | |
| 20374 | 447 | |
| 51093 | 448 | lemma " \<forall>x. \<exists>n. ((Suc n) * length (([(3::int) * x + f t * y - 9 + (- z)] @ []) @ xs) = length xs) \<and> m < 5*n - length (xs @ [2,3,4,x*z + 8 - y]) \<longrightarrow> (\<exists>p. \<forall>q. p \<and> q \<longrightarrow> r)" | 
| 20564 
6857bd9f1a79
Reifiaction now deals with Interpretations with an arbtrary number of parameters. It deals with binding. The Atomic cases can be I ... = f (xs!n)
 chaieb parents: 
20503diff
changeset | 449 | apply (reify Irifm.simps Irnat_simps Irlist.simps Irint_simps) | 
| 20374 | 450 | oops | 
| 451 | ||
| 61343 | 452 | text \<open>An example for equations containing type variables\<close> | 
| 22199 
b617ddd200eb
Now deals with simples cases where the input equations contain type variables
 chaieb parents: 
20564diff
changeset | 453 | |
| 58310 | 454 | datatype prod = Zero | One | Var nat | Mul prod prod | 
| 22199 
b617ddd200eb
Now deals with simples cases where the input equations contain type variables
 chaieb parents: 
20564diff
changeset | 455 | | Pw prod nat | PNM nat nat prod | 
| 51093 | 456 | |
| 457 | primrec Iprod :: " prod \<Rightarrow> ('a::linordered_idom) list \<Rightarrow>'a" 
 | |
| 35419 | 458 | where | 
| 23624 
82091387f6d7
The order for parameter for interpretation is now inversted:
 chaieb parents: 
23608diff
changeset | 459 | "Iprod Zero vs = 0" | 
| 35419 | 460 | | "Iprod One vs = 1" | 
| 51093 | 461 | | "Iprod (Var n) vs = vs ! n" | 
| 462 | | "Iprod (Mul a b) vs = Iprod a vs * Iprod b vs" | |
| 463 | | "Iprod (Pw a n) vs = Iprod a vs ^ n" | |
| 464 | | "Iprod (PNM n k t) vs = (vs ! n) ^ k * Iprod t vs" | |
| 39246 | 465 | |
| 58310 | 466 | datatype sgn = Pos prod | Neg prod | ZeroEq prod | NZeroEq prod | Tr | F | 
| 22199 
b617ddd200eb
Now deals with simples cases where the input equations contain type variables
 chaieb parents: 
20564diff
changeset | 467 | | Or sgn sgn | And sgn sgn | 
| 
b617ddd200eb
Now deals with simples cases where the input equations contain type variables
 chaieb parents: 
20564diff
changeset | 468 | |
| 51093 | 469 | primrec Isgn :: "sgn \<Rightarrow> ('a::linordered_idom) list \<Rightarrow> bool"
 | 
| 35419 | 470 | where | 
| 51093 | 471 | "Isgn Tr vs \<longleftrightarrow> True" | 
| 472 | | "Isgn F vs \<longleftrightarrow> False" | |
| 473 | | "Isgn (ZeroEq t) vs \<longleftrightarrow> Iprod t vs = 0" | |
| 474 | | "Isgn (NZeroEq t) vs \<longleftrightarrow> Iprod t vs \<noteq> 0" | |
| 475 | | "Isgn (Pos t) vs \<longleftrightarrow> Iprod t vs > 0" | |
| 476 | | "Isgn (Neg t) vs \<longleftrightarrow> Iprod t vs < 0" | |
| 477 | | "Isgn (And p q) vs \<longleftrightarrow> Isgn p vs \<and> Isgn q vs" | |
| 478 | | "Isgn (Or p q) vs \<longleftrightarrow> Isgn p vs \<or> Isgn q vs" | |
| 22199 
b617ddd200eb
Now deals with simples cases where the input equations contain type variables
 chaieb parents: 
20564diff
changeset | 479 | |
| 
b617ddd200eb
Now deals with simples cases where the input equations contain type variables
 chaieb parents: 
20564diff
changeset | 480 | lemmas eqs = Isgn.simps Iprod.simps | 
| 
b617ddd200eb
Now deals with simples cases where the input equations contain type variables
 chaieb parents: 
20564diff
changeset | 481 | |
| 51093 | 482 | lemma "(x::'a::{linordered_idom}) ^ 4 * y * z * y ^ 2 * z ^ 23 > 0"
 | 
| 22199 
b617ddd200eb
Now deals with simples cases where the input equations contain type variables
 chaieb parents: 
20564diff
changeset | 483 | apply (reify eqs) | 
| 
b617ddd200eb
Now deals with simples cases where the input equations contain type variables
 chaieb parents: 
20564diff
changeset | 484 | oops | 
| 
b617ddd200eb
Now deals with simples cases where the input equations contain type variables
 chaieb parents: 
20564diff
changeset | 485 | |
| 20319 | 486 | end | 
| 51093 | 487 |