ebook img

RCF2: Evaluation and Consistency PDF

0.51 MB·
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview RCF2: Evaluation and Consistency

RCF 2 Evaluation and Consistency∗ ε&C ∗ π R ∗ π• R O O Michael Pfender† 9 0 0 July 2008‡ 2 n a J Abstract: Weconstructhereaniterative evaluation ofall(coded)PRmaps: 0 progress of this iteration can be measured by descending complexity, within 3 Ordinal O := N[ω], of polynomials in one indeterminate, called “ω”. As (well) ] T order on this Ordinal we choose the lexicographical one. Non-infinit descent C of such iterations is added as a mild additional axiom schema (π ) to Theory O . PR = PR + (abstr) of Primitive Recursion with predicate abstraction, out h A t of foregoing part RFC 1. This then gives (correct) on-termination of iterative a m evaluation of argumented deduction trees as well: for theories PR and π R = A O [ PR + (π ). By means of this constructive evaluation the Main Theorem A O is proved, on Termination-conditioned (Inner) Soundness for Theories π R, O 2 O v extending N[ω]. As a consequence we get in fact Self-Consistency for theories 1 π R, namely π R-derivability of π R’s own free-variable Consistency formula 8 O O O 8 Con = Con (k) = ¬Prov (k, pfalseq ) : N → 2, k ∈ N free. 3 πOR πOR def πOR . Here PR predicate Prov (k,u) says, for an arithmetical theory T : number 9 T 0 k ∈ N is a T-Proof code proving internally T-formula code u, arithmetised 8 Proof in G¨odel’s sense. 0 : As to expect from classiccal setting, Self-Consistency of π R gives (uncon- v O i ditioned) Objective Soundness. Eventually we show Termination-Conditioned X Soundness “already” for PR . But it turns out that present derivation of Self- A r a Consistency, and already that of Consistency formula of PR from this con- A ditioned Soundness “needs” schema (π˜) of non-infinit descent in Ordinal N[ω], which is presumably not derived by PR itself. A 0 LegendofLOGO:εforConstructiveevaluation,C forSelf-Consistency tobe derivedfor suitable theories πOR, πO•R strengthening in a “mild” way the (categorical) Free-Variables Theory PRA of Primitive Recursion with predicate abstraction ∗Consideration of implicational version (π•) of Descent axiom added O †TU Berlin, Mathematik, [email protected] ‡last revised January 30, 2009 1 1 Summary G¨odel’s first Incompleteness Theorem for Principia Mathematica and “ver- wandte Systeme”, on which in particular is based the second one, on non- provability of PM’s own Consistency formula Con , exhibits a (closed) PM PM formula ϕ with property that PM ⊢ [ϕ ⇐⇒ ¬(∃k ∈ N)Prov (k, pϕq )], in words: PM Theory PM derives ϕ to be equivalent to its “own” coded, arithmetised non- Provability. Since this equivalence needs already for its statement “full” formal, “not testable” quantification, theConsistency Provability issue isnotsettledforFree- Variables Primitive Recursive Arithmetic and its strengthenings – Theories T which express (formalised, “internal”) Consistency as free-variable formula Con = Con (k) = ¬Prov (k, pfalseq ) : N → 2 : T T T “No k ∈ N is a Proof code proving pfalseq .” This is the point of depart for investigation of “suitable” strengthenings π R = PR +(π )ofcategoricalTheoryPR ofPrimitiveRecursion, enriched O A O A with predicate abstraction Objects {A|χ} = {a ∈ A|χ(a)} : Plausibel axiom schema (π ), more presisely: its contraposition π˜ , states “weak” impossibility O O of infinite descending chains in any Ordinal O extending polynomial semiring N[ω], with its canonical, lexicographical order. Central Non-Infinite Descent Schema, Descent Schema for short: We need an axiom-schema for expressing – in free variables – Finite de- scent (endo-driven) chains, descending in complexity value out of Ordinal O (cid:23) N[ω], a schema called (π ), which gives the “name” to Descent1 Theory O π R = PR +(π ) : This theory is a pure strengthening of PR , it has the O A O A same language. Easier to interprete logically is (π )’s equivalent, Free-Variables contraposi- O 0extended Poster Abstract “Arithmetical Consistency via Constructive evaluation”, Con- ference celebrating Kurt G¨odel’s 100th birthday, Vienna april 28, 29, 2006 1notion added 2 JAN 2009 2 tion, on “absurdity” of infinite descending chains, namely: c = c(a) : A → O PR (complexity), p = p(a) : A → A PR (predecessor endo), PR ⊢ c(a) > 0 =⇒ cp(a) < c(a) (descent), A . O . PR ⊢ c(a) = 0 =⇒ p(a) = a (stationarity at zero) A O ψ = ψ(a) : A → 2 absurdity test predicate, PR ⊢ ψ(a) =⇒ cpn(a) > 0 , A O with quantifier decoration: PR ⊢ ∀a[ψ(a) =⇒ ∀ncpn(a) > 0 ] A O the latter statement: “infinit descent”, is felt absurd, and “therefore” so “must be”, by axiom, condition ψ implying this “absurdity”: (π˜ ) O . π R ⊢ ψ(a) = false : A → 2, intuitively: O π R ⊢ ∀a ¬ψ(a). O [The first four lines of the antecedent constitute (p,c) as (the data of) a CCI : of a Complexity Controlled Iteration, with (stepwise) descending order O values inOrdinal O.Central example: General Recursive, Acckermann type PR-code evaluation ε will be resolved into such a CCI , O := N[ω] ⊂ N.] O My Thesis then is that these theories π R, weaker than PM, set theories O and even Peano Arithmetic PA (when given its quantified form), derive their own internal (Free-Variable) Consistency formula Con (k) : N → 2, see πOR above. Notions and Arguments for Self-Consistency of π R : In order to O obtain constructive Theories – candidates for self-Consistency – we introduce first, into fundamental Theory PR of (categorical) Free-Variables Primitive Recursion, predicate abstraction of PR maps χ = χ(a) : A → 2 (A a finite power of NNO N), into defined Objects {A|χ}, and then strengthen Theory PR obtained this way, by a free-variables, (inferential) schema (π ) of “on”- A O terminating descent,intoTheorie(s)π R,on-terminatingdescent ofComplexity O Controlled Iterations (CCI ’s, see above), with (descending) complexity values O in Ordinal O (cid:23) N[ω]. Strengthened Theory π R = PR +(π ), with its language equal to that O A O of PR , is asserted to derive the (Free-Variable) formula Con (k) which A πOR expresses internally: within π R itself, Consistency of Theory π R, see above. O O Proof is by CCIN[ω] (descent) property of a suitable, atomic PR evaluation step e applied to PR-map-code/argument pairs (u,x) ∈ PR × . A X [Here ⊂ N denotes the Universal Object of all (codes of) singletons and X (nested) pairs of natural numbers, enriched by a shymbol ⊥ equally coded in N, to designate undefined values, of defined partially defined PR maps. Objects A of PR , π R admit a natural embedding A ⊏ into this this universal A O X Object.] 3 Iteration ε, of step e, is in fact controlled by a syntactic complexity c (u) ∈ PR N[ω], descending with each application of e as long as minimum complexity 0 = c ( pidq ) is not “yet” reached. PR Strengthening of PR by schema (π ) – cf. its free-variables contraposition A O (π˜ ) above – into Theory π R = PR+(π ), is “just” to allow for a so to say O O O sound, canonical evaluation “algorithm” for π R : O On one hand it is proved straight forward that evaluation ε above has the expected recursive properties of an evaluation, this within (categorical, Free- Variables) Theory µR of µ-Recursion. Onthe other hand, π R has the same Language asPR ,so that this ε is a O A natural candidate for likewise – sound – evaluation of internal version of theory π R, and for being totally defined in a suitable Free-Variables sense, techni- O cally: to on-terminate, this just by its property to be a Complexity Controlled Iteration, with order values in N[ω]. In fact, by schema (π ) itself (O extending N[ω]), ε preserves the extra O equation instances inserted by internalisation of (π ). O Dangerous bound: is there a good reason that this evaluation is not a self-evaluation for Theory π R? O Answer: ε is – by definition – not PR: If you take the diagonal diag(n) = ε(enum (n),cantor (n)) : N → N, def PR X enum an internal PR count of all PR map codes, and cantor : N −∼=→ PR “the” Cantor’s count of ⊂ N, then you get Ackermann’s origXinal diagonXal X function2 which grows faster than any PR function: but π R has only PR maps O as its maps, it is a (pure) strengthening of PR . A Ontheotherhand,εisintuitively total,since,intuitively, complexitycem(u,x) “must” reach 0 in finitely many e-steps. The latter intuition can be, in free variables (!), expressed formally by π R’s schema (π˜ ) : Free-Variables con- O O traposition of (π ). Schema (π˜ ) says that a condition which implies infinite O O descent of such a chain (on all x), must be false (on all x), “absurd”. Complexity Controlled Iteration ε of e extends canonically into a Com- plexity Controlled evaluation ε , of argumented deduction trees, ε again d d defined by CCIN[ω] : this time by iteration of a tree evaluation step ed suitably extending basic evaluation step e to argumented deduction trees. Deduction-tree evaluation starts on trees of form dtree /x, obtained as fol- k lowsfromk andx :Calldtree the(first)deduction tree which(internally)proves k kth internal equation u=ˇ v of theory π R, enumeration of proved equations k O being (lexicographically) by code of (first) Proof. This argument-free deduction tree dtree then is provided – node-wise top down from given x ∈ – with k its spread down arguments in = ∪˙ {(cid:3)} = ∪˙ {hi} ⊂ N; (eXmpty list (cid:3) def X X X (cid:3) = hi refers to a not yet known argument, not “yet” ata given time ofstepwise evaluation e .) d 2 for a two-parameter,simple genuine Ackermann function cf. Eilenberg/Elgot1970 4 Spreading down arguments this way eventually converts argument-free kth deduction tree dtree into (partially non-dummy) argumented deduction tree k dtree /x. k Iteration ε , of tree evaluation step e , again is Complexity Controlled de- d d scending in Ordinal N[ω],when controlled by deduction tree complexity c . This d complexity is defined essentially as the (polynomial) sum of all (syntactical) complexities c (u) of map codes appearing in the deduction tree. PR So, as it does to basic evaluation ε, schema π˜ applies to complexity con- N[ω] trolled evaluation ε of argumented deduction-trees as well, and gives d Deduction-Tree Evaluation non-infinit Descent: Infinit strict descent of endo map e – with respect to complexity c – is absurd. d d This deduction-tree evaluation ε externalises, as far as terminating, kth in- d . ternalequation u=ˇ v oftheory π Rinto complete evaluation ε(u,x) = ε(v,x) : k O Termination-Conditioned Inner Soundness, our Main Theorem. For a given PR predicate χ = χ(x) : → 2, the Main Theorem reads: X Theoryπ Rderives: If fork ∈ Nandforx ∈ r{⊥}given,Prov (k, pχq ) O X πOR “holds”, and if argumented π R deduction tree dtree /x admits complete eval- O k uation by m (“say”) deduction-tree evaluation-steps e , d Then the pair (k,x) is a Soundness-Instance, i.e. then kth given (inter- nal) π R-Provability Prov (k, pχq ) implies χ(x), for the given argument O πOR x ∈ r{⊥}. All this within Theory π R itself. O X Corollary: Self-Consistency Derivability for Theory π R : O π R ⊢ Con , i.e. “necessarily” in Free-Variables form: O πOR π R ⊢ ¬Prov (k, pfalseq ) : N → 2, i.e. equationally: O πOR π R ⊢ ¬[ pfalseq =ˇ ptrueq ] : N → 2, k ∈ N free : O k Theory π R derives that no k ∈ N is the internal π R-Proof for pfalseq . O O Proof of this Corollary to Termination-Conditioned Soundness: By the last assertion of the Theorem, with χ = χ(x) := false (x) : → 2, X X and x := h0i ∈ , we get: X Evaluation-effective internal inconsistency of π R, i.e. availability of an O evaluation-terminating internal deduction tree of pfalseq , implies false : . π R ⊢ pfalseq =ˇ ptrueq ∧ c em(dtree /h0i) = 0 =⇒ false (h0i). O k d d k X Contraposition to this, still with k,m ∈ N free: π R ⊢ true (h0i) =⇒ ¬[ pfalseq =ˇ ptrueq ] ∨ c em(dtree /h0i) > 0, O k d d k X i.e. by Free-Variables (Boolean) tautology: π R ⊢ pfalseq =ˇ ptrueq =⇒ c em(dtree /h0i) > 0 : N2 → 2. O k d d k 5 This πOR derivative invites to apply schema (π˜N[ω]) of πOR : “infinite endo-driven descent with order values in N[ω] is absurd.” We apply this schema to deduction tree evaluation ε given by step e and d d complexity c which descends – this is Argumented-Tree Evaluation Descent – d with each application of e , as long as complexity 0 is not (“yet”) reached. We d combine this with choice of “overall” absurdity condition ψ = ψ(k) := [ pfalseq =ˇ ptrueq ] : N → 2, k ∈ N free (!) k and get, by schema (π˜N[ω]), overall negation of this (overall) “absurd” predicate ψ, namely π R ⊢ ¬[ pfalseq =ˇ ptrueq ] : N → 2, k ∈ N free. O k This is π R-derivation of the free-variable Consistency Formula of π R itself. O O From this Self-Consistency of Theorie(s) π R, which is equivalent to in- O jectivity of (special) internal numeralisation ν : 2 ֌ [ ,2] , we get im- mediately injectivity of all these numeralisatio2ns ν = ν1(a)πO:RA ֌ [ ,A] = A A 1 ⌈ ,A⌉/=ˇ , and from this, with naturality of this family, “full” objective Sound- 1 ness of Theory π R which reads: O Formalised π R-Provability of (code of) PR predicate χ : → 2 implies – O X within Theory π R – “validity” χ(x) of χ at “each” of χ’s arguments x ∈ . O X But for derivation ofSelf-Consistency fromTermination-conditioned Sound- ness, a suitable strengthening of PR , here by schema (π˜) = (π˜ ), stating A N[ω] absurdity of infinite descent in Ordinal N[ω], seems to be necessary: my guess is that Theories PRA as well as PR and hence PR , are not strong enough A to derive their own (internal) Consistency. On the other hand, we know from G¨odel’s work that Principia Mathematica “und verwandte Systeme” are too strong for being self-consistent. This is true for any (formally) quantified Arith- metical Theory Q, in particular for the (classical, quantified) version PA of Peano Arithmetic: Such theory Q has all ingredients for G¨odel’s Proof of his two Incompleteness Theorems. In section 7 We discuss3 a formally stronger, implicational, “local” variant (π•) of inferential Descent axiom (π ), with respect to Self-Consistency and O O (Objective) Soundness: In particular, Self-Consistency Proof becomes techni- cally easier for corresponding theory π•R. O The final section 84 gives a proof of (Objective) Consistency for Theorie(s) π•R (hence π R) relative to basic Theory PR of Primitive Recursion and O O A hence relative to fundamental Theory PR of Primitive Recursion “itself”. For proof of this (relative) Consistency, we use a schema, (ρ ), of recur- O sive reduction for predicate validity, reduction along a Complexity Controlled Iteration (CCI ), admitted by Theory PR (and its strengthenings.) O A 3 insertion ? JAN 2009 4inserted 2 JAN 2009 6 2 Iterative Evaluation of PR Map Codes Object- andmaptermsofallour theoriesarecodedstraight ahead, inparticular since formally we have no (individual) variables on the Object Language level: We code all our terms just as prime power products “over” the LATEXsource codes describing these terms, this externally in naive numbers, out of N as well as into the NNO N of the (categorical) arithmetical theory itself. Equality Enumeration: As “any” theories, fundamental Theory PR of Primitive Recursion as well as basic Theory PR = PR+(abstr), definitional A enrichement of PR by the schema of predicate abstraction: hχ : A → 2i 7→ {A|χ}, a “virtual”, abstracted Object in PR , admit an (external) primitive A recursive enumeration of their respective theorems, ordered by length (more precisely: by lexicographical order) of the first proofs of these (equational) Theorems, here: =PR (k) : N → PR×PR ⊂ N×N and =PRA (k) : N → PR ×PR ⊂ N×N A A respectively. BythePR Representation Theorem5.3of Roma`n1989,theseenumerations give rise to their internal versions =ˇPR : N → PR×PR ⊂ N2 and k =ˇPRA : N → PR ×PR ⊂ N2, k A A with internalisation (representation) property PR ⊢ =ˇ = num(=PR) : → PR×PR ⊂ N2 and num(k) k 1 PR ⊢ =ˇ = num(=PRA) : → PR ×PR ⊂ N2. num(k) k 1 A A Here (external) numeralisation is given externally PR as num(n) = sn : −→0 N −→s ... −→s N, 1 num(m,n) = (num(m),num(n)) : → N×N, m,n (“meta”) free in N, 1 PR = {N|PR}isthepredicative, PR decidablesubset ofN“ofallPRcodes” (a PR -Object), internalisation of PR ⊂ N of all PR-terms on Object Language A level. Analogeous meaning for internalisation PR ⊂ N of PR ⊂ N. A A For discussion of “constructive” evaluation, we need representation of all PR maps within one PR endo map monoid, namely within PR( , ), A ⊥ ⊥ where ⊂ N, = {N| : N → 2} is the (predicative) UniversalXObjeXct of N-singlXetons {hnXi|n ∈ N}X, possibly nested N-pairs {ha;bi|a,b ∈ }, and X . = ∪˙ {⊥} = (a) ∨˙ a = ⊥ : N → 2 ⊥ def X X X 7 isX augmentedbysymbol(code)⊥ : → N,⊥ takingcareofdefinedundefined 1 arguments of defined partial maps.5 Here we view (formally) = (a), = (a) : N → N as PR- ⊥ ⊥ predicates, not “yet” as abstrXacted XObjectXs = X{N| }, = {N| }, ⊥ ⊥ X X X X of Theory PR = PR+(abstr). A . We allow us to write “a ∈ ” instead of (a) = true : N → N, and . X X “a ∈ ” for (a) = true, and similarly for other predicates. ⊥ ⊥ X X This way we introduce – `a la Reiter – “Object” 2 just as target for predi- cates χ : A → 2, meaning χ : A → N to be a predicate in the exact sense that χ : → N satisfies A χ◦sign = χ◦¬◦¬ = χ : N −χ→ N −si→gn N, “still” A fundamental. bydef We define, within endo map set PR(N,N) a subTheory PR externally PR X as follows, by mimikry of schema (abstr) for the special case of predicate = (a) : N → N, but without introduction of a coarser notion of equality, aXs in X case of schema of abstraction constituting Theory PR = PR+(abstr). A So Theory PR ⊂ PR(N,N) comes in, by external PR enumeration of its X Object and map terms as follows: Objects of PR are predicates χ : → 2, i.e. PR-predicates χ : N → 2 X X such that PR ⊢ χ(a) =⇒ (a) : N → 2, i.e. such that X PR ⊢ χ(a) =⇒ (a) ∧ a 6= ⊥ : N → 2. ⊥ X PR -maps in PR (χ,ψ) are PR-maps f : N → N such that X X . ¬ (a) =⇒ f(a) = ⊥, and χ(a) =⇒ ψ ◦f(a) : N → 2, X observe the “truncated” parallelism to definition of PR -maps A f : {A|χ} → {B|ψ}. ⊏ Then “assignment” I : PR −→ PR is defined as follows externally PR: X I = ˙ = {h0i} : N ⊃ ⊃ → 2, def ⊥ 1 1 X X IN = N˙ = hNi = {hni|n ∈ N} : N ⊃ ⊃ → 2, def def ⊥ X X and further recursively: I(A×B) = hA×Bi = {ha;bi|(a,b) ∈ (A×B)} : N ⊃ → 2, def def X Functorial definition of I on PR maps: I ˙ PR(A,B) ∋ f 7→ If = f ∈ PR X 5 cf. Ch. 1, final section X 8 then is “canonical”, by external PR on the structure of PR-map f : A → B, in particualar by mapping all “arguments” in NrA˙ = NrIA into ⊥ ∈ : ⊥ X one waste basket outside all Objects of PR .6 X Interesting now is that we can extend embedding I above into an embedding I : PR −→ PR , by the following A X Definition: For a (general) PR Object, of form {A|χ}, define A I{A|χ} = {A˙ |χ˙} = {IA|Iχ} def bydef . = {a ∈ IA|Iχ(a) = htruei} : N ⊃ → 2. bydef ⊥ X We replace here “don’t-worry arguments” in the complement ¬χ of PR - A Object {A|χ} by cutting them out in the definition of replacing PR -Object I{A|χ} = {A˙ |χ˙}. “Coarser” notion =PRA (coarser then =PR) isXthen re- placed by original notion of equality, =PR itself, notion of map-equality of roof PR “⊂”PR(N,N) : This formal “sameness” of PR equality was the goal of X the considerations above: The new version PR replacing PR isomorphically, XA A is a subTheory of PR with notion of equality – objectively as well as (then) internally – inherited from fundamental Theory PR. Universal Embedding Theorem:7 (i) I : PR −→ PR ⊂ PR(N,N) above is an embedding which preserves X composition. ∼ = (ii) (Enumerative) Restriction I : PR −→ PR = I[PR] of this embed- X def ding to its (enumerated) Image defines an isomorphism of categories. It is defined above as hf : A → Bi −→I hf˙ : A˙ → B˙ i, by the “natural” (primitive) recursion on the structure of f as a map in fundamental Theory PR of (Cartesian) Primitive Recursion. (iii) PR embedding I “canonically” extends into an embedding (!) I : PR −→ PR(N,N) A of Theory PR = PR+(abstr) – Theory PR with abstraction of predi- A cates into (“new”, “virtual”) Objects {A|χ : A → 2} – to the Set of PR endomaps of N, of which – by the way – PR ( , ) is (formally) a A ⊥ ⊥ X X SubQuotient. [Equality =PRA of (distinguished) PR endo maps when viewed as PR endo maps on = {N| : N → 2}, is embedded to A ⊥ ⊥ PR - (PR-)equalityXby I : PRX −→ PR “⊂”PR(N,N).] A X X 6for the details see Ch. 1, final section . 7from Ch. 1, final section X X 9 (iv) Mainassertion: Embedding Iabovedefinesanisomorphismofcategories ∼ = I : PR −→ PR A XA ontoa “naturallychoosen” (emumerated) categoryPR ofPRpredicates XA on Universal Object (PR-predicate) : N → N, with canonical maps ⊥ X in between (see above), and whith composition inherited from that of PR(N,N). This isomorphism is defined (naturally) by ˙ ˙ I(f : {A|χ} → {B|ψ}) = hf : χ˙ → ψi, χ˙ : N ⊃ ⊃ ⊃ A˙ → 2, ⊥ X X ψ˙ : N ⊃ ⊃ ⊃ B˙ → 2, ⊥ X X f˙ = I (f) : N ⊃ A˙ → B˙ ⊂ N above. bydef PR Bythis isomorphism of categories, PR inherits fromcategoryPR allof XA A its (categorically described) structure: the isomorphism transports Carte- sian PR structure, equality predicates on all Objects, schema of predicate abstraction, equalisers, and – trivially – the whole algebraic, logic and order structure on NNO N and truth Object 2. We have furthermore: ˙ (v) For each fundamental Object A, embedded Object A = IA ⊂ comes ⊥ X with a retraction retr : → A˙ ∪˙ {⊥}, defined by retr (a) = a for XA X⊥ XA def ˙ a ∈ A, retr (a) = ⊥ otherwise. XA def This family of retractions clearly extends to a retraction family retr : → {A˙ |χ˙}∪˙ {⊥} = I{A|χ}∪˙ {⊥} X{A|χ} X⊥ for all PR -Objects {A|χ} : This is what ⊥ ∈ is good for. A ⊥ X (vi) ForeachObject{A|χ}ofPR ,inparticularforeachfundamental Object A A ≡ {A|true }, PR comes with the characteristic (predicative) subset A A χ˙ : I{A|χ} : → 2 of defined PR above, isomorphic to {A|χ} ⊥ ⊥ X X within PR (!) via “canonical” PR -isomorphism A A ∼= ˙ iso : {A|χ} −→ I{A|χ} = {A|χ˙}, X{A|χ} the PR -isomorphism defined PR on the “structure” of {A|χ}, as re- A striction of iso : A → IA for fundamental Object A, in turn (exter- XA nally/internally) PR defined by iso (0) = h0i : → I ⊂ , X def ⊥ 1 1 1 X iso (0) = h0i : → [ I ⊂ ] IN ⊂ , XN def ⊥ 1 1 X further externally PR: ∼= iso (a,b) = hiso (a);iso (b)i : A×B −→ I(A×B) ⊂ . X(A×B) def XA XB X⊥ ∼= We name the inverse isomorphism jso : I{A|χ} −→ {A|χ}. X{A|χ} 10

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.