Concepts locaux et globaux. Deuxième partie: Théorie ‚fonctorielle‘ Guerino Mazzola U & ETH Zürich Internet Institute for Music Science [email protected] www.encyclospace.org.
Download ReportTranscript Concepts locaux et globaux. Deuxième partie: Théorie ‚fonctorielle‘ Guerino Mazzola U & ETH Zürich Internet Institute for Music Science [email protected] www.encyclospace.org.
Concepts locaux et globaux. Deuxième partie: Théorie ‚fonctorielle‘ Guerino Mazzola U & ETH Zürich Internet Institute for Music Science [email protected] www.encyclospace.org Contents • Where we are... • What is a topos? • The topos of presheaves • Functorial local compositions • Concept modeling over topoi Where we are... C Ÿ12 (chords) M — 2 (motives) Ambient space Ÿ12= finite ->enumeration, Pólya & de Bruijn —2 = infinite -> ?? Where we are... K B B set module B @0Ÿ@B A Ÿ@B K 0 • A = Ÿn: sequences (b0,b1,…,bn) • A = B: self-addressed tones Need general addresses A Where we are... B M A@B B A@B = eB.Lin(A,B) A=R R@B = eB.Lin(R,B) B2 Where we are... Ÿ12 S A@B = eB.Lin(A,B) R = Ÿ, A = Ÿ11, B= Ÿ12 Series: S Ÿ11 @ Ÿ12 = e Ÿ12.Lin(Ÿ11, Ÿ12) Ÿ12 12 Where we are... I II III IV V VI VII Where we are... The class nerve cn(K) of global composition is not classifying 2 5 II 2 10 5 10 2 10 6 2 V VII 10 2 VI IV 2 6 2 2 5 15 2 2 10 I 2 6 5 2 2 2 10 III Where we are... Motivic strip of Zig-Zag (15) 5 6 4 (19) (19) 7 3 (2) 1 (11) 8 (20) 2 9 (10) (15) (16) Where we are... M Ÿ@B B = „EH“ —2 H E E Where we are... Have universal construction of a „resolution of KI“ res: ADn* KI It is determined only by the KI address A and the nerve n* of the covering atlas I. res ADn* KI Where we are... 3 6 1 4 5 0Dn* res 2 4 6 2 2 1 5 1 4 d c a b 4 5 KI 6 3 1 3 6 3 2 5 Where we are... The category ObLocomA of local objective A-addressed compositions has as objects the couples (K, A@C) of sets K of affine morphisms in A@C and as morphisms f: (K, A@C) L, A@D) set maps f: K L which are naturally induced by affine morphism F in C@D The category ObGlocomA of global objective A-addressed compositions has as objects KI coverings of sets K by atlases I of local objective A-addressed compositions with manifold gluing conditions and manifold morphisms ff: KI LJ, including and compatible with atlas morphisms f: I J What is a Topos? Sets cartesian products X x Y disjoint sums X Y powersets XY characteristic maps c:X —> 2 no „algebra“ Mod@ F: Mod —> Sets presheaves have all these properties Mod direct products A≈B has „algebra“ no powersets no characteristic maps What is a Topos? A category E is a topos iff it • has terminal object 1 and products A B • has initial object 0 and coproducts A +B • has exponentials XY • has a subobject classifier 1 W Our examples: 1) E = Sets 2) E = Mod@ sets presheaves over the category Mod of modules What is a Topos? Example: E = Sets A B = cartesian product A¥B B b (a,b) A a 1 = {} is terminal: There is a unique !:X 1: What is a Topos? Example: E = Sets A +B = disjoint union, 0 = A A+B B 0 = is initial: There is a unique !:0 X:?? What is a Topos? Example: E = Sets XY = {maps f:Y X} Hom(Z, XY) ≈ Hom(Z Y, X) g: Z XY ~> g*: Z Y X g*(z,y) = g(z)(y) What is a Topos? Example: E = Sets subobject classifier 1 W X W = 2 = {0,1} 1 2: 0 ~> 0 c)=0 iff X Y c ! 1 2 X Y Subobjects(Y) ≈ Hom(Y,W) {0,1} What is a Topos? Counterexample: E = ModR with R-linear maps There is no subobject classifier here! X Y c ! W 0-module X = Ker(c) Y/X ≈ Im(c) W absurd! What is a Topos? More generally, take the category Mod of modules over any rings, together with (di)affine morphism „A@B“. This is not only not a topos, it has other not very agreable properties: • Have no module M+N for the property k A@M or k A@N iff k A@(M+N) • Have no module P(M) for the property K A@M iff K A@P(M) Presheaves Problem: When replacing M by the set A@M, we loose all information about M. Solution: Replace a module M by the system of sets @M: Mod Sets: A ~> A@M „set of all perspectives of M, as viewed from A“ B u A g 1A@M = 1A@M g.u M @M = presheaf of M u@M: A@M B@M u.v:C B A u.v@M = [email protected]@M Presheaves Presheaves: A@F F: Mod Sets: A ~> F(A) Together with the transition maps u@F : A@F B@F for u:B A with the properties 1A@F = 1A@F u.v: C B A u.v@F = [email protected]@F Mod@ = category of presheaves on Mod Presheaves Example 1 S = set, @S: Mod Sets: A ~> A@S = S Transition maps, u: B A, u@S = 1S : S S Sets „small topos within a large topos“ @Sets Mod@ Presheaves Example 2 M = module, 2@M: Mod Sets: A ~> A@ 2@M = 2A@M Transition maps, u: B A, u@2@M : A@2@M B@2@M u@M: A@M B@M K A@M K ~> u@M(K) B@M u@M(K) Presheaves Example 2* F = presheaf, 2F: Mod Sets: A ~> A@2F = 2A@F Transition maps, u: B A, u@2F : A@2F B@2F u@F: A@F B@F K A@F K ~> u@F(K) B@F u@F(K) Presheaves Example 3 M, N = modules, @M+@N: Mod Sets: A ~> A@M + A@N Transition maps, u: B A A@M A@N B@M B@N Presheaves Example 3* F, G = presheaves, F+G: Mod Sets: A ~> A@F + A@G Transition maps, u: B A A@F A@G B@F B@G Presheaves Why are presheaves a solution? Yoneda Lemma The functorial map @: Mod Mod@ is fully faithfull. M @M • M@N ≈ Hom(@M,@N) • M ≈ N iff @M ≈ @N • M@F ≈ Hom(@M,F) Mod@ @Mod Mod Presheaves Example: E = Mod@ F G = pointwise cartesian product A@(F G G) = A@F A@ G F¥ A@G G g (f,g) A@F F f 1 = {} is terminal: A@1 Unique A@!: A@X !:X 1: Presheaves Example: E = Mod@ F +G = pointwise disjoint union A@(G H) = A@G + A@ H G +H A@G G H A@H 0 = is initial: A@0 Unique !:0 X:?? A@!: A@0 A@X: Presheaves Example: E = Mod@ A@XY Hom(@A, XY) Hom(@A Y, X) (Yoneda!) (axiom) Define: A@XY = Hom(@A Y, X) Presheaves Example: E = Mod@ subobject classifier 1 W A@W= {subpresheaves of @A} = {sieves in @A} 1 W : 0 ~> @A X Y c ! Subpresheaves(Y) Hom(Y,W) 1 W Functorial Locs In Mod@ replace 2F by WF Understand the musical meaning of the difference! A@2F = 2A@F ={subsets of A@F} = {A-addressed local objective compositions in F} ? ObLocomA, but F = presheaf, not only module! A@ WF ≈ Hom(@A,WF) ≈ Hom(@A F,W) ≈ Subpresheaves(@A F) = {A-addressed local functorial compositions in F} Functorial Locs ^: A@2F A@WF K A@F ~> K^ @A F X@K^ X@A X@F = {(f,x.f), f:X A, x K} f@F: A@F X@F: x ~> x.f F K f@K^ @A 1A f:X A Functorial Locs K Ÿ @F F = @EH @—2 H E f10: 0Ÿ Ÿ: 0 ~> 10 Functorial Locs Ÿ12 S series S Ÿ11 @ Ÿ12 K = {S} More general: set of k sequences of pitch classes of length t+1 K = {S1,S2,...,Sk} This is a „polyphonic“ local composition Ÿ12 K Ÿt @ Ÿ12 S1 Sk Functorial Locs Ÿ12 S1 Sk s ≤ t, define morphism f: Ÿs Ÿt e0 ~> ei(0) e1 ~> ei(1) ................. es ~> ei(s) f@K^ Ÿ12 S1.f Sk.f es e0 e1 Ÿs Functorial Locs The „functorial“ change K ~> K^ has dramatic consequences for the global theory! II A = 0Ÿ I II III IV V VI V VII VI I VII A = Ÿ12 IV X Ÿ12 ~> III X* = End*(X) Ÿ12@Ÿ12 Functorial Locs IIII* VI VI* VV* IV* IV II* VII* VII ToM, ch. 25 III* III I* I* II* = Ÿ12@Ÿ12 II* Functorial Locs X* Ÿ12@Ÿ12 X*^ (Ÿ12@Ÿ12)^ @Ÿ12 @Ÿ12 @Ÿ12 @Ÿ12 I*^ I* I*^I*II*^ II* = Ÿ12@Ÿ12)^ (Ÿ II*^ II* Functorial Locs @Ÿ12 I* f@I*^ e0.4 I*^ II* f@I*^f@II*^ e8.0 e11.3 II*^ f@II*^ @Ÿ12 1Ÿ12 f = e11.0: Ÿ12 Ÿ12 e0.4.e11.0 = e11.3.e11.0 = e8.0 Functorial Locs @Ÿ12 I* I*^ I*^ II*^ II* II*^ @Ÿ12 0 = (I* II*)^ I*^ II*^ Functorial Locs Consequences for sheaves of functions (Xi) Xi Z (Xj) Xj (Xij) ¿ ≈ ? (Xji) Functorial Locs Grothendieck topology of finite covering families Xi Z (Xi) Xj ( Xi Z Xj) Xi Z Xj (Xj) concept modeling unity completeness discourse infinite recursion universal ramification ordered combinatorics concept concept concept concept concept modeling AnchorNote Pause Note Onset Duration Onset Pitch Loudness Duration – – – Ÿ STRG – concept modeling MakroNote • Ornaments • Schenker Analysis Satellites AnchorNote MakroNote Pause Onset – Note Duration – Onset Pitch – Ÿ Loudness Duration STRG – concept modeling • FM-Synthesis concept modeling • FM-Synthesis FM-Object Knot Support Modulator Amplitude Frequency Phase – – – FM-Object concept modeling F = form name one of five „space“ types Forms a name diagram √ in Mod@ an identifier monomorphism in Mod@ id: Functor(F) >Frame(√) > Functor(F) F:id.type(√) Frame(√) concept modeling Frame(√)-space for type: synonyme √ = „G“ Functor(G) synonyme(√) = Functor(G) simple √ = „“@B simple(√) = @B limit √ = name diagram Mod@ limit(√) = lim(n. diagram Mod@) colimit √ = name diagram Mod@ colimit(√) = colim(n. diagram Mod@) power √ = „G“ Functor(G) power(√) = WFunctor(G) renaming representation conjunction disjunction collection Denotators concept modeling D = denotator name address A A K K A@ Functor(F) „A-valued point“ > Form F Functor(F) D:A@F(K) Frame(√) concept modeling concept modeling ModE@ = Topos R E @ Mod Mod Forms Names F S Dia(Formsº, Mod@) S(F) = (typeF,idF, √F) Mono(Mod@) Types Sema(Forms, Mod@) = Types x Mono(Mod@) x Dia(Formsº, Mod@) concept modeling E = Topos R E Forms Names F S Dia(Formsº,E ) S(F) = (typeF,idF, √F) Mono(E ) Types Sema(Forms,E ) = Types x Mono(E ) x Dia(Formsº,E ) concept modeling Names F √G Forms typeF √F H typeG typeH √H G E -Denotators concept modeling D = denotator name R E „address“ A R A K K:A Topor(F) Topor(F) E Form F:id.type(√) id: Topor(F) D:A@F(K) Frame(√) > concept modeling Galois Theory Form Semiotic Defining equation Defining diagram fS(X) = 0 x1 x2 x3 Field S xn √F F2 F1 Fr Form Semiotic S