calculating with string diagrams
play

Calculating with string diagrams Ross Street Macquarie University - PowerPoint PPT Presentation

Calculating with string diagrams Ross Street Macquarie University Workshop on Diagrammatic Reasoning in Higher Education University of Newcastle Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 1 / 32 Reasons


  1. Calculating with string diagrams Ross Street Macquarie University Workshop on Diagrammatic Reasoning in Higher Education University of Newcastle Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 1 / 32

  2. Reasons for choice of this topic § A conviction that string diagrams can be understood better than algebraic equations by most students § My experience with postgraduate students and undergraduate vacation scholars using strings § As seen by the general public, knot theory for mathematics seems a bit like astronomy for physics § A belief that string diagrams are widely applicable and powerful in communicating and in discovery § That this is “advanced mathematics from an elementary viewpoint” (to quote Ronnie Brown’s twist on Felix Klein) Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 2 / 32

  3. Intentions § Moving from linear algebra, we will look at braided monoidal categories (bmc) and explain the string diagrams for which bmc provide the environment. § Familiar operations from vector calculus will be transported to bmc where the properties can be expressed in terms of equalities between string diagrams. § Geometrically appealing arguments will be used to prove the scarcity of multiplications on Euclidean space, a theorem of a type originally proved using higher powered methods. Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 3 / 32

  4. Arrows and categories § Already introduced in undergraduate teaching is the notation f : X Ñ A for a function taking each element x in the set X to an element f p x q of the set A . g f § In the situation X Ý Ñ A Ý Ñ K we can follow f by g and obtain a new function, called the composite of f and g , denoted by g ˝ f : X Ñ K . § There is an identity function 1 X : X Ñ X for every set X : 1 X p x q “ x . § If we now ignore the fact that X , A , K are sets (just call them vertices or objects) and that f , g are functions (just call them edges or morphisms) we are looking at a big directed graph. § If we admit the existence of a composition operation ˝ which is associative and has identities 1 X , we are looking at a category. Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 4 / 32

  5. Euclidean space § The set of real numbers is denoted by R . § A vector of length n is a list x “ p x 1 , . . . , x n q of real numbers. The set of these vectors is n -dimensional Euclidean space, denoted R n . § Algebra is about operations on sets. We can add vectors x and y entry by entry to give a new vector x ` y . We can scalar multiply a real number r by a vector x to obtain a vector rx . § For example, R 3 is ordinary 3-dimensional space. We have three particular unit vectors: e 1 “ p 1 , 0 , 0 q , e 2 “ p 0 , 1 , 0 q , e 3 “ p 0 , 0 , 1 q . Every vector x in R 3 is a unique linear combination x “ x 1 e 1 ` x 2 e 2 ` x 3 e 3 . Similarly in R n Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 5 / 32

  6. Linear algebra § A function f : R m Ñ R n is linear when it preserves linear combinations: f p x ` y q “ f p x q ` f p y q , f p rx q “ rf p x q . § Thus we have a category E : objects are Euclidean spaces and morphisms are linear functions. We write E p V , W q for the set of morphisms from object V to object W . § For this category E , we can add the morphisms in E p V , W q : define f ` g by p f ` g qp x q “ f p x q ` g p x q . Composition distributes over this addition. Such a category is called additive. § Notice that the only linear functions f : R Ñ R are those given by multiplying by a fixed real number. So E p R , R q can be identified with R . Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 6 / 32

  7. Multilinear algebra § Categorical algebra is about operations on categories. The category E has such an operation called tensor product: R m b R n – R mn . However, when thinking of the mn unit vectors of R mn as being in the tensor product they are denoted by e i b e j for 1 ď i ď m , 1 ď j ď n . Every element of R m b R n is a unique linear combination of these. § Bilinear functions U ˆ V Ñ W are in bijection with linear functions U b V Ñ W . § Note that R acts as unit for the tensor. § For linear functions f : R m Ñ R m 1 and g : R n Ñ R n 1 , we have a linear function f b g : R m b R n Ñ R m 1 b R n 1 defined by p f b g qp e i b e j q “ f p e i qb g p e j q . Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 7 / 32

  8. Monoidal categories and their string diagrams § A category V is monoidal when it is equipped with an operation called tensor product taking pairs of objects V , W to an object V b W and pairs of morphisms f : V Ñ V 1 , g : W Ñ W 1 to a morphism f b g : V b W Ñ V 1 b W 1 . There is also an object I acting as a unit for tensor. Composition and identity morphisms are respected in the expected way. An example is V “ E with I “ R . § A morphism such as f : U b V b W Ñ X b V is depicted as U V W f X V § Composition is performed vertically with splicing involved; tensor product is horizontal placement. Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 8 / 32

  9. D C d c ② ❏ ② ❏ ② ✞ ❏ C B ② ❏ ✞ ② ❏ ② ✞ ❏ ② ❏ D ✞ ② ❏ ❏ ② ✞ ② ✞ b ✞ ✞ B ✞ ✞ C ✞ ✞ ✞ a B A a b c d B b B Ý Ñ A , C b D Ý Ñ B , C Ý Ñ B b C , D Ý Ñ D b C . The value of the above diagram Γ is the composite 1 B b c b d � B b B b C b D b C v p Γ q “p B b C b D 1 b 1 b b b 1 � B b B b B b C a b 1 b 1 � A b B b C q . Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 9 / 32

  10. Here is a deformation of the previous Γ ; the value is the same using monoidal category axioms. C B c ✷ t D t ✷ t ✷ t t ✷ t t ✷ t B ✷ t t ✷ t ✷ a ✷ C ✷ ✷ ✷ ✷ d ✷ t t ✷ t ✷ t t ✷ t ✷ t t D ✷ t A t t b C B 1 B b c b 1 � B b B b C b D a b 1 b 1 � A b C b D v p Γ q “p B b C b D 1 b 1 b d � A b C b D b C 1 b b b 1 � A b B b C q . Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 10 / 32

  11. a b The geometry handles units well: if I Ý Ñ A b B and C Ý Ñ I , then the following three string diagrams all have the same value. C C C a a b ✴ ✴ ✎✎✎✎✎✎✎ ✴ ✎✎✎✎✎✎✎ ✴ b a b ✴ ✴ A ✴ B ✾ A ✴ B ✾ ✴ ✆✆✆✆✆ ✴ A ✾ B ✴ ✴ ✾ ✴ ✴ ✾ , , The straight lines can be curved while the nodes are really labelled points. There is no bending back of the curves allowed: the diagrams are progressive. These planar deformations are part of the geometry of monoidal categories. Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 11 / 32

  12. Progressive graph on Mollymook Beach Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 12 / 32

  13. Duals A morphism ε : A b B Ñ I is a counit for an adjunction A % B when there exists a morphism η : I Ñ B b A satisfying the two equations: A A B B η η B “ “ A ε ε A B We call B a right dual for A . Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 13 / 32

  14. Backtracking When there is no ambiguity, we denote counits by cups Y and units by caps X . So the duality condition becomes the more geometrically “obvious” operation of pulling the ends of the strings as below. A A B B B “ “ A A B The above are sometimes called the snake equations . The geometry of duality in monoidal categories allows backtracking in the plane. Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 14 / 32

  15. Dot product, vector product and the quaternions § For any x and y in R n , the dot product x ‚ y “ x 1 y 1 ` ¨ ¨ ¨ ` x n y n defines a bilinear function R n ˆ R n Ñ R and so a linear function ‚ : R n b R n Ñ R . § For any x and y in R 3 , the vector product x ^ y “ p x 2 y 3 ´ x 3 y 2 , x 3 y 1 ´ x 1 y 3 , x 1 y 2 ´ x 2 y 1 q defines a bilinear function R 3 ˆ R 3 Ñ R 3 and so a linear function ^ : R 3 b R 3 Ñ R 3 . § The quaternions is the non-commutative ring H “ R ˆ R 3 p– R 4 q with componentwise addition and associative multiplication defined by p r , x qp s , y q “ p rs ´ x ‚ y , ry ` sx ` x ^ y q Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 15 / 32

  16. Braiding Now suppose the monoidal category is braided . Then we have isomorphisms c X , Y : X b Y Ý Ñ Y b X which we depict by a left-over-right crossing of strings in three dimensions; the inverse is a right-over-left crossing. X Y Y X Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 16 / 32

  17. The braiding axioms reinforce the view that it behaves like a crossing. g f “ “ “ g f Z X Y Z X X Y Z X b Y Y b Z “ “ Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 17 / 32

  18. The following Reidemeister move or Yang-Baxter equation is a consequence. “ We will refer to these properties as the geometry of braiding. Ross Street Macquarie University Calculating with string diagrams 9 Nov 2018 18 / 32

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend