Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf ·...
Transcript of Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf ·...
![Page 1: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/1.jpg)
Operational Physics, Logic and Categories
Sean Tull
University of Oxford
Categories, Logic and PhysicsUniversity of Edinburgh, 14 April 2016
![Page 2: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/2.jpg)
The Plan
Operational Physics
Operational Logic
Operational Categories... in partial form... in total formEffectus Theory
Outlook
![Page 3: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/3.jpg)
I. Operational Physics
![Page 4: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/4.jpg)
Operational Physics
I Focus on the physical operations allowed by a given theory.
I Operational reconstructions of quantum theory due to Hardy,Chiribella-D’Ariano-Perinotti, . . .
I Categorical quantum mechanics: models quantum theory usingcertain symmetric monoidal categories.
I CDP’s operational-probabilistic theories: categorical + probabilistic.
![Page 5: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/5.jpg)
Operational Physics
I Focus on the physical operations allowed by a given theory.
I Operational reconstructions of quantum theory due to Hardy,Chiribella-D’Ariano-Perinotti, . . .
I Categorical quantum mechanics: models quantum theory usingcertain symmetric monoidal categories.
I CDP’s operational-probabilistic theories: categorical + probabilistic.
![Page 6: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/6.jpg)
Operational Physics
I Focus on the physical operations allowed by a given theory.
I Operational reconstructions of quantum theory due to Hardy,Chiribella-D’Ariano-Perinotti, . . .
I Categorical quantum mechanics: models quantum theory usingcertain symmetric monoidal categories.
I CDP’s operational-probabilistic theories: categorical + probabilistic.
![Page 7: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/7.jpg)
Operational Physics
I Focus on the physical operations allowed by a given theory.
I Operational reconstructions of quantum theory due to Hardy,Chiribella-D’Ariano-Perinotti, . . .
I Categorical quantum mechanics: models quantum theory usingcertain symmetric monoidal categories.
I CDP’s operational-probabilistic theories: categorical + probabilistic.
![Page 8: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/8.jpg)
Operational Physics
I Focus on the physical operations allowed by a given theory.
I Operational reconstructions of quantum theory due to Hardy,Chiribella-D’Ariano-Perinotti, . . .
I Categorical quantum mechanics: models quantum theory usingcertain symmetric monoidal categories.
I CDP’s operational-probabilistic theories: categorical + probabilistic.
![Page 9: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/9.jpg)
II. Operational Logic
![Page 10: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/10.jpg)
Operational Theories
An operational theory with control (OTC) Θ:
I Systems A,B,C , . . .
I Events f : A→ B , forming a symmetric monoidal category:
C
E F
A
g
fB
ρ
D
e
I Tests {fx : A→ Bx}x∈X , X = finite outcome setCall subsets {fy}y∈Y ⊆ {fx}x∈X partial tests.
I Coarse-graining:{fx : A→ B}x∈X ∪ {gy}y∈Y =⇒ {
Ŕx∈X fx} ∪ {gy}y∈Y
Partial ‘addition’ f > g : A→ B on events.
![Page 11: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/11.jpg)
Operational Theories
An operational theory with control (OTC) Θ:
I Systems A,B,C , . . .
I Events f : A→ B , forming a symmetric monoidal category:
C
E F
A
g
fB
ρ
D
e
I Tests {fx : A→ Bx}x∈X , X = finite outcome setCall subsets {fy}y∈Y ⊆ {fx}x∈X partial tests.
I Coarse-graining:{fx : A→ B}x∈X ∪ {gy}y∈Y =⇒ {
Ŕx∈X fx} ∪ {gy}y∈Y
Partial ‘addition’ f > g : A→ B on events.
![Page 12: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/12.jpg)
Operational Theories
An operational theory with control (OTC) Θ:
I Systems A,B,C , . . .
I Events f : A→ B , forming a symmetric monoidal category:
C
E F
A
g
fB
ρ
D
e
I Tests {fx : A→ Bx}x∈X , X = finite outcome setCall subsets {fy}y∈Y ⊆ {fx}x∈X partial tests.
I Coarse-graining:{fx : A→ B}x∈X ∪ {gy}y∈Y =⇒ {
Ŕx∈X fx} ∪ {gy}y∈Y
Partial ‘addition’ f > g : A→ B on events.
![Page 13: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/13.jpg)
Operational Theories
An operational theory with control (OTC) Θ:
I Systems A,B,C , . . .
I Events f : A→ B
, forming a symmetric monoidal category:
C
E F
A
g
fB
ρ
D
e
I Tests {fx : A→ Bx}x∈X , X = finite outcome setCall subsets {fy}y∈Y ⊆ {fx}x∈X partial tests.
I Coarse-graining:{fx : A→ B}x∈X ∪ {gy}y∈Y =⇒ {
Ŕx∈X fx} ∪ {gy}y∈Y
Partial ‘addition’ f > g : A→ B on events.
![Page 14: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/14.jpg)
Operational Theories
An operational theory with control (OTC) Θ:
I Systems A,B,C , . . .
I Events f : A→ B , forming a symmetric monoidal category:
C
E F
A
g
fB
ρ
D
e
I Tests {fx : A→ Bx}x∈X , X = finite outcome setCall subsets {fy}y∈Y ⊆ {fx}x∈X partial tests.
I Coarse-graining:{fx : A→ B}x∈X ∪ {gy}y∈Y =⇒ {
Ŕx∈X fx} ∪ {gy}y∈Y
Partial ‘addition’ f > g : A→ B on events.
![Page 15: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/15.jpg)
Operational Theories
An operational theory with control (OTC) Θ:
I Systems A,B,C , . . .
I Events f : A→ B , forming a symmetric monoidal category:
C
E F
A
g
fB
ρ
D
e
I Tests {fx : A→ Bx}x∈X , X = finite outcome set
Call subsets {fy}y∈Y ⊆ {fx}x∈X partial tests.
I Coarse-graining:{fx : A→ B}x∈X ∪ {gy}y∈Y =⇒ {
Ŕx∈X fx} ∪ {gy}y∈Y
Partial ‘addition’ f > g : A→ B on events.
![Page 16: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/16.jpg)
Operational Theories
An operational theory with control (OTC) Θ:
I Systems A,B,C , . . .
I Events f : A→ B , forming a symmetric monoidal category:
C
E F
A
g
fB
ρ
D
e
I Tests {fx : A→ Bx}x∈X , X = finite outcome setCall subsets {fy}y∈Y ⊆ {fx}x∈X partial tests.
I Coarse-graining:{fx : A→ B}x∈X ∪ {gy}y∈Y =⇒ {
Ŕx∈X fx} ∪ {gy}y∈Y
Partial ‘addition’ f > g : A→ B on events.
![Page 17: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/17.jpg)
Operational Theories
An operational theory with control (OTC) Θ:
I Systems A,B,C , . . .
I Events f : A→ B , forming a symmetric monoidal category:
C
E F
A
g
fB
ρ
D
e
I Tests {fx : A→ Bx}x∈X , X = finite outcome setCall subsets {fy}y∈Y ⊆ {fx}x∈X partial tests.
I Coarse-graining:{fx : A→ B}x∈X ∪ {gy}y∈Y =⇒ {
Ŕx∈X fx} ∪ {gy}y∈Y
Partial ‘addition’ f > g : A→ B on events.
![Page 18: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/18.jpg)
Assumptions of an OTC
Impossible events: zero arrows 0A,B : A→ B with f > 0 = f .
Control: given tests {fx : A→ Bx}x∈X and {g(x , y) : Bx → Cx ,y}y∈Yx forx ∈ X , the following is a test:{
A Bx Cx ,yfx g(x , y)
}x∈X ,y∈Yx
Causality: unique deterministic effect A : A→ I , i.e. { A} a test.Every event f has unique effect e with {f , e} a test.
Don’t assume scalars p : I → I are probabilitites p ∈ [0, 1].
Examples
Many! Classical: deterministic or probabilistic.Quantum: Hilbert spaces/C*-algebras and CP, sub-unital maps.
![Page 19: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/19.jpg)
Assumptions of an OTC
Impossible events: zero arrows 0A,B : A→ B with f > 0 = f .
Control: given tests {fx : A→ Bx}x∈X and {g(x , y) : Bx → Cx ,y}y∈Yx forx ∈ X , the following is a test:{
A Bx Cx ,yfx g(x , y)
}x∈X ,y∈Yx
Causality: unique deterministic effect A : A→ I , i.e. { A} a test.Every event f has unique effect e with {f , e} a test.
Don’t assume scalars p : I → I are probabilitites p ∈ [0, 1].
Examples
Many! Classical: deterministic or probabilistic.Quantum: Hilbert spaces/C*-algebras and CP, sub-unital maps.
![Page 20: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/20.jpg)
Assumptions of an OTC
Impossible events: zero arrows 0A,B : A→ B with f > 0 = f .
Control:
given tests {fx : A→ Bx}x∈X and {g(x , y) : Bx → Cx ,y}y∈Yx forx ∈ X , the following is a test:{
A Bx Cx ,yfx g(x , y)
}x∈X ,y∈Yx
Causality: unique deterministic effect A : A→ I , i.e. { A} a test.Every event f has unique effect e with {f , e} a test.
Don’t assume scalars p : I → I are probabilitites p ∈ [0, 1].
Examples
Many! Classical: deterministic or probabilistic.Quantum: Hilbert spaces/C*-algebras and CP, sub-unital maps.
![Page 21: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/21.jpg)
Assumptions of an OTC
Impossible events: zero arrows 0A,B : A→ B with f > 0 = f .
Control: given tests {fx : A→ Bx}x∈X and {g(x , y) : Bx → Cx ,y}y∈Yx forx ∈ X , the following is a test:{
A Bx Cx ,yfx g(x , y)
}x∈X ,y∈Yx
Causality: unique deterministic effect A : A→ I , i.e. { A} a test.Every event f has unique effect e with {f , e} a test.
Don’t assume scalars p : I → I are probabilitites p ∈ [0, 1].
Examples
Many! Classical: deterministic or probabilistic.Quantum: Hilbert spaces/C*-algebras and CP, sub-unital maps.
![Page 22: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/22.jpg)
Assumptions of an OTC
Impossible events: zero arrows 0A,B : A→ B with f > 0 = f .
Control: given tests {fx : A→ Bx}x∈X and {g(x , y) : Bx → Cx ,y}y∈Yx forx ∈ X , the following is a test:{
A Bx Cx ,yfx g(x , y)
}x∈X ,y∈Yx
Causality: unique deterministic effect A : A→ I , i.e. { A} a test.
Every event f has unique effect e with {f , e} a test.
Don’t assume scalars p : I → I are probabilitites p ∈ [0, 1].
Examples
Many! Classical: deterministic or probabilistic.Quantum: Hilbert spaces/C*-algebras and CP, sub-unital maps.
![Page 23: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/23.jpg)
Assumptions of an OTC
Impossible events: zero arrows 0A,B : A→ B with f > 0 = f .
Control: given tests {fx : A→ Bx}x∈X and {g(x , y) : Bx → Cx ,y}y∈Yx forx ∈ X , the following is a test:{
A Bx Cx ,yfx g(x , y)
}x∈X ,y∈Yx
Causality: unique deterministic effect A : A→ I , i.e. { A} a test.Every event f has unique effect e with {f , e} a test.
Don’t assume scalars p : I → I are probabilitites p ∈ [0, 1].
Examples
Many! Classical: deterministic or probabilistic.Quantum: Hilbert spaces/C*-algebras and CP, sub-unital maps.
![Page 24: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/24.jpg)
Assumptions of an OTC
Impossible events: zero arrows 0A,B : A→ B with f > 0 = f .
Control: given tests {fx : A→ Bx}x∈X and {g(x , y) : Bx → Cx ,y}y∈Yx forx ∈ X , the following is a test:{
A Bx Cx ,yfx g(x , y)
}x∈X ,y∈Yx
Causality: unique deterministic effect A : A→ I , i.e. { A} a test.Every event f has unique effect e with {f , e} a test.
Don’t assume scalars p : I → I are probabilitites p ∈ [0, 1].
Examples
Many! Classical: deterministic or probabilistic.Quantum: Hilbert spaces/C*-algebras and CP, sub-unital maps.
![Page 25: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/25.jpg)
Assumptions of an OTC
Impossible events: zero arrows 0A,B : A→ B with f > 0 = f .
Control: given tests {fx : A→ Bx}x∈X and {g(x , y) : Bx → Cx ,y}y∈Yx forx ∈ X , the following is a test:{
A Bx Cx ,yfx g(x , y)
}x∈X ,y∈Yx
Causality: unique deterministic effect A : A→ I , i.e. { A} a test.Every event f has unique effect e with {f , e} a test.
Don’t assume scalars p : I → I are probabilitites p ∈ [0, 1].
Examples
Many! Classical: deterministic or probabilistic.Quantum: Hilbert spaces/C*-algebras and CP, sub-unital maps.
![Page 26: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/26.jpg)
III. Operational Categories
![Page 27: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/27.jpg)
From Theories to Categories
Encode outcome sets X ‘in the objects’.
Θ has direct sums when ∀ {Bx}x∈X ∃ test {.x : B → Bx}x∈X s.t:
partial tests {fx : A→ Bx}x∈Xf : A→ B with .x ◦ f = fx
“B =⊕
x∈X Bx”
Direct sum completion Θ+ has EventΘ+ = ParTest(Θ):
I systems (Ax)x∈XI events M : (Ax)x∈X → (By )y∈Y are X -indexed partial tests{M(x , y) : Ax → By}y∈Y
Categorically, direct sums are finite coproducts:
Bi B1 + ...+ Bn =⊕n
k=1 Bk Bjκi .j
, .j◦κi =
{id i = j0 i 6= j
![Page 28: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/28.jpg)
From Theories to Categories
Encode outcome sets X ‘in the objects’.
Θ has direct sums when ∀ {Bx}x∈X ∃ test {.x : B → Bx}x∈X s.t:
partial tests {fx : A→ Bx}x∈Xf : A→ B with .x ◦ f = fx
“B =⊕
x∈X Bx”
Direct sum completion Θ+ has EventΘ+ = ParTest(Θ):
I systems (Ax)x∈XI events M : (Ax)x∈X → (By )y∈Y are X -indexed partial tests{M(x , y) : Ax → By}y∈Y
Categorically, direct sums are finite coproducts:
Bi B1 + ...+ Bn =⊕n
k=1 Bk Bjκi .j
, .j◦κi =
{id i = j0 i 6= j
![Page 29: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/29.jpg)
From Theories to Categories
Encode outcome sets X ‘in the objects’.
Θ has direct sums when ∀ {Bx}x∈X ∃ test {.x : B → Bx}x∈X s.t:
partial tests {fx : A→ Bx}x∈Xf : A→ B with .x ◦ f = fx
“B =⊕
x∈X Bx”
Direct sum completion Θ+ has EventΘ+ = ParTest(Θ):
I systems (Ax)x∈XI events M : (Ax)x∈X → (By )y∈Y are X -indexed partial tests{M(x , y) : Ax → By}y∈Y
Categorically, direct sums are finite coproducts:
Bi B1 + ...+ Bn =⊕n
k=1 Bk Bjκi .j
, .j◦κi =
{id i = j0 i 6= j
![Page 30: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/30.jpg)
From Theories to Categories
Encode outcome sets X ‘in the objects’.
Θ has direct sums when ∀ {Bx}x∈X ∃ test {.x : B → Bx}x∈X s.t:
partial tests {fx : A→ Bx}x∈Xf : A→ B with .x ◦ f = fx
“B =⊕
x∈X Bx”
Direct sum completion Θ+ has EventΘ+ = ParTest(Θ):
I systems (Ax)x∈XI events M : (Ax)x∈X → (By )y∈Y are X -indexed partial tests{M(x , y) : Ax → By}y∈Y
Categorically, direct sums are finite coproducts:
Bi B1 + ...+ Bn =⊕n
k=1 Bk Bjκi .j
, .j◦κi =
{id i = j0 i 6= j
![Page 31: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/31.jpg)
From Theories to Categories
Encode outcome sets X ‘in the objects’.
Θ has direct sums when ∀ {Bx}x∈X ∃ test {.x : B → Bx}x∈X s.t:
partial tests {fx : A→ Bx}x∈Xf : A→ B with .x ◦ f = fx
“B =⊕
x∈X Bx”
Direct sum completion Θ+ has EventΘ+ = ParTest(Θ):
I systems (Ax)x∈XI events M : (Ax)x∈X → (By )y∈Y are X -indexed partial tests{M(x , y) : Ax → By}y∈Y
Categorically, direct sums are finite coproducts:
Bi B1 + ...+ Bn =⊕n
k=1 Bk Bjκi .j
, .j◦κi =
{id i = j0 i 6= j
![Page 32: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/32.jpg)
From Theories to Categories
Encode outcome sets X ‘in the objects’.
Θ has direct sums when ∀ {Bx}x∈X ∃ test {.x : B → Bx}x∈X s.t:
partial tests {fx : A→ Bx}x∈Xf : A→ B with .x ◦ f = fx
“B =⊕
x∈X Bx”
Direct sum completion Θ+ has EventΘ+ = ParTest(Θ):
I systems (Ax)x∈XI events M : (Ax)x∈X → (By )y∈Y are X -indexed partial tests{M(x , y) : Ax → By}y∈Y
Categorically, direct sums are finite coproducts:
Bi B1 + ...+ Bn =⊕n
k=1 Bk Bjκi .j
, .j◦κi =
{id i = j0 i 6= j
![Page 33: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/33.jpg)
From Theories to Categories
Encode outcome sets X ‘in the objects’.
Θ has direct sums when ∀ {Bx}x∈X ∃ test {.x : B → Bx}x∈X s.t:
partial tests {fx : A→ Bx}x∈Xf : A→ B with .x ◦ f = fx
“B =⊕
x∈X Bx”
Direct sum completion Θ+ has EventΘ+ = ParTest(Θ):
I systems (Ax)x∈XI events M : (Ax)x∈X → (By )y∈Y are X -indexed partial tests{M(x , y) : Ax → By}y∈Y
Categorically, direct sums are finite coproducts:
Bi B1 + ...+ Bn =⊕n
k=1 Bk Bjκi .j
, .j◦κi =
{id i = j0 i 6= j
![Page 34: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/34.jpg)
From Theories to Categories
Encode outcome sets X ‘in the objects’.
Θ has direct sums when ∀ {Bx}x∈X ∃ test {.x : B → Bx}x∈X s.t:
partial tests {fx : A→ Bx}x∈Xf : A→ B with .x ◦ f = fx
“B =⊕
x∈X Bx”
Direct sum completion Θ+ has EventΘ+ = ParTest(Θ):
I systems (Ax)x∈XI events M : (Ax)x∈X → (By )y∈Y are X -indexed partial tests{M(x , y) : Ax → By}y∈Y
Categorically, direct sums are finite coproducts:
Bi B1 + ...+ Bn =⊕n
k=1 Bk Bjκi .j
, .j◦κi =
{id i = j0 i 6= j
![Page 35: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/35.jpg)
Operational Categories: Partial Form
Partial tests: {fi : A→ Bi}ni=1 ⇐⇒ f : A→ B1 + ...+ Bn s.t. .i ◦ f = fi .Tests: when f is total, i.e. ◦ f = .
Coarse-graining:Ŕn
i=1 fi = A B + . . .+ B Bf O
SMC (C,⊗) with finite coproducts, zero object and family A : A→ I s.t:
I .i : A1 + ...+ An → Ai jointly monic
I ∀f : A→ B ∃! total g : A→ B + I with f = .1 ◦ gI ⊗,+ distribute: A⊗ (B + C ) ' A⊗ B + A⊗ C
I A+B = [ A, B ], I = id, A⊗B = λ ◦ ( A ⊗ B)
![Page 36: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/36.jpg)
Operational Categories: Partial Form
Partial tests: {fi : A→ Bi}ni=1 ⇐⇒ f : A→ B1 + ...+ Bn s.t. .i ◦ f = fi .
Tests: when f is total, i.e. ◦ f = .
Coarse-graining:Ŕn
i=1 fi = A B + . . .+ B Bf O
SMC (C,⊗) with finite coproducts, zero object and family A : A→ I s.t:
I .i : A1 + ...+ An → Ai jointly monic
I ∀f : A→ B ∃! total g : A→ B + I with f = .1 ◦ gI ⊗,+ distribute: A⊗ (B + C ) ' A⊗ B + A⊗ C
I A+B = [ A, B ], I = id, A⊗B = λ ◦ ( A ⊗ B)
![Page 37: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/37.jpg)
Operational Categories: Partial Form
Partial tests: {fi : A→ Bi}ni=1 ⇐⇒ f : A→ B1 + ...+ Bn s.t. .i ◦ f = fi .Tests: when f is total, i.e. ◦ f = .
Coarse-graining:Ŕn
i=1 fi = A B + . . .+ B Bf O
SMC (C,⊗) with finite coproducts, zero object and family A : A→ I s.t:
I .i : A1 + ...+ An → Ai jointly monic
I ∀f : A→ B ∃! total g : A→ B + I with f = .1 ◦ gI ⊗,+ distribute: A⊗ (B + C ) ' A⊗ B + A⊗ C
I A+B = [ A, B ], I = id, A⊗B = λ ◦ ( A ⊗ B)
![Page 38: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/38.jpg)
Operational Categories: Partial Form
Partial tests: {fi : A→ Bi}ni=1 ⇐⇒ f : A→ B1 + ...+ Bn s.t. .i ◦ f = fi .Tests: when f is total, i.e. ◦ f = .
Coarse-graining:Ŕn
i=1 fi = A B + . . .+ B Bf O
SMC (C,⊗) with finite coproducts, zero object and family A : A→ I s.t:
I .i : A1 + ...+ An → Ai jointly monic
I ∀f : A→ B ∃! total g : A→ B + I with f = .1 ◦ gI ⊗,+ distribute: A⊗ (B + C ) ' A⊗ B + A⊗ C
I A+B = [ A, B ], I = id, A⊗B = λ ◦ ( A ⊗ B)
![Page 39: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/39.jpg)
Operational Categories: Partial Form
Partial tests: {fi : A→ Bi}ni=1 ⇐⇒ f : A→ B1 + ...+ Bn s.t. .i ◦ f = fi .Tests: when f is total, i.e. ◦ f = .
Coarse-graining:Ŕn
i=1 fi = A B + . . .+ B Bf O
Control: A B1 + . . .+ Bn C1 + . . .+ Cnf g1+...+gn
SMC (C,⊗) with finite coproducts, zero object and family A : A→ I s.t:
I .i : A1 + ...+ An → Ai jointly monic
I ∀f : A→ B ∃! total g : A→ B + I with f = .1 ◦ gI ⊗,+ distribute: A⊗ (B + C ) ' A⊗ B + A⊗ C
I A+B = [ A, B ], I = id, A⊗B = λ ◦ ( A ⊗ B)
![Page 40: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/40.jpg)
Operational Categories: Partial Form
Partial tests: {fi : A→ Bi}ni=1 ⇐⇒ f : A→ B1 + ...+ Bn s.t. .i ◦ f = fi .Tests: when f is total, i.e. ◦ f = .
Coarse-graining:Ŕn
i=1 fi = A B + . . .+ B Bf O
TheoremOTC with direct sums Θ ⇐⇒ operational category in partial form (C, ):
SMC (C,⊗) with finite coproducts, zero object and family A : A→ I s.t:
I .i : A1 + ...+ An → Ai jointly monic
I ∀f : A→ B ∃! total g : A→ B + I with f = .1 ◦ gI ⊗,+ distribute: A⊗ (B + C ) ' A⊗ B + A⊗ C
I A+B = [ A, B ], I = id, A⊗B = λ ◦ ( A ⊗ B)
![Page 41: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/41.jpg)
Operational Categories: Partial Form
Partial tests: {fi : A→ Bi}ni=1 ⇐⇒ f : A→ B1 + ...+ Bn s.t. .i ◦ f = fi .Tests: when f is total, i.e. ◦ f = .
Coarse-graining:Ŕn
i=1 fi = A B + . . .+ B Bf O
TheoremOTC with direct sums Θ ⇐⇒ operational category in partial form (C, ):SMC (C,⊗) with finite coproducts, zero object and family A : A→ I s.t:
I .i : A1 + ...+ An → Ai jointly monic
I ∀f : A→ B ∃! total g : A→ B + I with f = .1 ◦ gI ⊗,+ distribute: A⊗ (B + C ) ' A⊗ B + A⊗ C
I A+B = [ A, B ], I = id, A⊗B = λ ◦ ( A ⊗ B)
![Page 42: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/42.jpg)
Operational Categories: Partial Form
Partial tests: {fi : A→ Bi}ni=1 ⇐⇒ f : A→ B1 + ...+ Bn s.t. .i ◦ f = fi .Tests: when f is total, i.e. ◦ f = .
Coarse-graining:Ŕn
i=1 fi = A B + . . .+ B Bf O
TheoremOTC with direct sums Θ ⇐⇒ operational category in partial form (C, ):SMC (C,⊗) with finite coproducts, zero object and family A : A→ I s.t:
I .i : A1 + ...+ An → Ai jointly monic
I ∀f : A→ B ∃! total g : A→ B + I with f = .1 ◦ gI ⊗,+ distribute: A⊗ (B + C ) ' A⊗ B + A⊗ C
I A+B = [ A, B ], I = id, A⊗B = λ ◦ ( A ⊗ B)
![Page 43: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/43.jpg)
Operational Categories: Partial Form
Partial tests: {fi : A→ Bi}ni=1 ⇐⇒ f : A→ B1 + ...+ Bn s.t. .i ◦ f = fi .Tests: when f is total, i.e. ◦ f = .
Coarse-graining:Ŕn
i=1 fi = A B + . . .+ B Bf O
TheoremOTC with direct sums Θ ⇐⇒ operational category in partial form (C, ):SMC (C,⊗) with finite coproducts, zero object and family A : A→ I s.t:
I .i : A1 + ...+ An → Ai jointly monic
I ∀f : A→ B ∃! total g : A→ B + I with f = .1 ◦ g
I ⊗,+ distribute: A⊗ (B + C ) ' A⊗ B + A⊗ C
I A+B = [ A, B ], I = id, A⊗B = λ ◦ ( A ⊗ B)
![Page 44: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/44.jpg)
Operational Categories: Partial Form
Partial tests: {fi : A→ Bi}ni=1 ⇐⇒ f : A→ B1 + ...+ Bn s.t. .i ◦ f = fi .Tests: when f is total, i.e. ◦ f = .
Coarse-graining:Ŕn
i=1 fi = A B + . . .+ B Bf O
TheoremOTC with direct sums Θ ⇐⇒ operational category in partial form (C, ):SMC (C,⊗) with finite coproducts, zero object and family A : A→ I s.t:
I .i : A1 + ...+ An → Ai jointly monic
I ∀f : A→ B ∃! total g : A→ B + I with f = .1 ◦ gI ⊗,+ distribute: A⊗ (B + C ) ' A⊗ B + A⊗ C
I A+B = [ A, B ], I = id, A⊗B = λ ◦ ( A ⊗ B)
![Page 45: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/45.jpg)
Operational Categories: Total Form
Instead of ParTest(Θ), now consider B = Test(Θ).
I [.i , κ2] : (A1 + . . .+ An) + 1→ Ai + 1 jointly monic, where
.i ◦ κj =
{κ1 i = jκ2◦! i 6= j
I the following are pullbacks:
Examples
Classical: Set. Probabilistic: Kl(D) for distribution monad D.Quantum: CStaropcpu
![Page 46: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/46.jpg)
Operational Categories: Total Form
Instead of ParTest(Θ), now consider B = Test(Θ).(M : (Ax)x∈X → (By )y∈Y = collections of tests {M(x , y) : Ax → By}y∈Y )
I [.i , κ2] : (A1 + . . .+ An) + 1→ Ai + 1 jointly monic, where
.i ◦ κj =
{κ1 i = jκ2◦! i 6= j
I the following are pullbacks:
Examples
Classical: Set. Probabilistic: Kl(D) for distribution monad D.Quantum: CStaropcpu
![Page 47: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/47.jpg)
Operational Categories: Total Form
Instead of ParTest(Θ), now consider B = Test(Θ).An operational category B is an SMC with finite coproducts (+, 0) dis-tributed over by ⊗, s.t. I is a terminal object 1 and:
I [.i , κ2] : (A1 + . . .+ An) + 1→ Ai + 1 jointly monic, where
.i ◦ κj =
{κ1 i = jκ2◦! i 6= j
I the following are pullbacks:
Examples
Classical: Set. Probabilistic: Kl(D) for distribution monad D.Quantum: CStaropcpu
![Page 48: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/48.jpg)
Operational Categories: Total Form
Instead of ParTest(Θ), now consider B = Test(Θ).An operational category B is an SMC with finite coproducts (+, 0) dis-tributed over by ⊗, s.t. I is a terminal object 1 and:
I [.i , κ2] : (A1 + . . .+ An) + 1→ Ai + 1 jointly monic, where
.i ◦ κj =
{κ1 i = jκ2◦! i 6= j
I the following are pullbacks:
Examples
Classical: Set. Probabilistic: Kl(D) for distribution monad D.Quantum: CStaropcpu
![Page 49: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/49.jpg)
Operational Categories: Total Form
Instead of ParTest(Θ), now consider B = Test(Θ).An operational category B is an SMC with finite coproducts (+, 0) dis-tributed over by ⊗, s.t. I is a terminal object 1 and:
I [.i , κ2] : (A1 + . . .+ An) + 1→ Ai + 1 jointly monic, where
.i ◦ κj =
{κ1 i = jκ2◦! i 6= j
I the following are pullbacks:
A 1
A + 1 1 + 1
!
κ1 κ1
!+!
Examples
Classical: Set. Probabilistic: Kl(D) for distribution monad D.Quantum: CStaropcpu
![Page 50: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/50.jpg)
Operational Categories: Total Form
Instead of ParTest(Θ), now consider B = Test(Θ).An operational category B is an SMC with finite coproducts (+, 0) dis-tributed over by ⊗, s.t. I is a terminal object 1 and:
I [.i , κ2] : (A1 + . . .+ An) + 1→ Ai + 1 jointly monic, where
.i ◦ κj =
{κ1 i = jκ2◦! i 6= j
I the following are pullbacks:
(B)
(Ax)x∈X (I )
(Ax)x∈X + (I ) (I ) + (I )
!
{fx}x∈X∪{e}
∃ ! {fx}x∈X
κ1
!
κ1
! + !
Examples
Classical: Set. Probabilistic: Kl(D) for distribution monad D.Quantum: CStaropcpu
![Page 51: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/51.jpg)
Operational Categories: Total Form
Instead of ParTest(Θ), now consider B = Test(Θ).An operational category B is an SMC with finite coproducts (+, 0) dis-tributed over by ⊗, s.t. I is a terminal object 1 and:
I [.i , κ2] : (A1 + . . .+ An) + 1→ Ai + 1 jointly monic, where
.i ◦ κj =
{κ1 i = jκ2◦! i 6= j
I the following are pullbacks:
A 1
A + 1 1 + 1
!
κ1 κ1
!+!
Examples
Classical: Set. Probabilistic: Kl(D) for distribution monad D.Quantum: CStaropcpu
![Page 52: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/52.jpg)
Operational Categories: Total Form
Instead of ParTest(Θ), now consider B = Test(Θ).An operational category B is an SMC with finite coproducts (+, 0) dis-tributed over by ⊗, s.t. I is a terminal object 1 and:
I [.i , κ2] : (A1 + . . .+ An) + 1→ Ai + 1 jointly monic, where
.i ◦ κj =
{κ1 i = jκ2◦! i 6= j
I the following are pullbacks:
A 1
A + 1 1 + 1
!
κ1 κ1
!+!
Examples
Classical: Set. Probabilistic: Kl(D) for distribution monad D.Quantum: CStaropcpu
![Page 53: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/53.jpg)
Equivalence Between Total and Partial Forms
We’ve seen operational categories in partial form (C, ) = ParTest(Θ)and total form B = Test(Θ).B = Ctotal, subcategory of f with ◦ f = , in fact determines all of C:
partial tests {fx}x∈Xtests {fx}x∈X ∪ {e : A→ I}
f : A→ B in ParTest(Θ)
f : A→ B + 1 in Test(Θ)
f : A→B in Par(B)
So (C, ) ' Par(B), the Kleisli category of (−) + 1 on B.
ExamplesB Par(B)
Classical Set PFunProbabilistic Kl(D) Kl(Dsub)Quantum CStaropcpu CStaropcpsu
![Page 54: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/54.jpg)
Equivalence Between Total and Partial Forms
We’ve seen operational categories in partial form (C, ) = ParTest(Θ)and total form B = Test(Θ).
B = Ctotal, subcategory of f with ◦ f = , in fact determines all of C:
partial tests {fx}x∈Xtests {fx}x∈X ∪ {e : A→ I}
f : A→ B in ParTest(Θ)
f : A→ B + 1 in Test(Θ)
f : A→B in Par(B)
So (C, ) ' Par(B), the Kleisli category of (−) + 1 on B.
ExamplesB Par(B)
Classical Set PFunProbabilistic Kl(D) Kl(Dsub)Quantum CStaropcpu CStaropcpsu
![Page 55: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/55.jpg)
Equivalence Between Total and Partial Forms
We’ve seen operational categories in partial form (C, ) = ParTest(Θ)and total form B = Test(Θ).B = Ctotal, subcategory of f with ◦ f = , in fact determines all of C:
partial tests {fx}x∈Xtests {fx}x∈X ∪ {e : A→ I}
f : A→ B in ParTest(Θ)
f : A→ B + 1 in Test(Θ)
f : A→B in Par(B)
So (C, ) ' Par(B), the Kleisli category of (−) + 1 on B.
ExamplesB Par(B)
Classical Set PFunProbabilistic Kl(D) Kl(Dsub)Quantum CStaropcpu CStaropcpsu
![Page 56: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/56.jpg)
Equivalence Between Total and Partial Forms
We’ve seen operational categories in partial form (C, ) = ParTest(Θ)and total form B = Test(Θ).B = Ctotal, subcategory of f with ◦ f = , in fact determines all of C:
partial tests {fx}x∈X
tests {fx}x∈X ∪ {e : A→ I}f : A→ B in ParTest(Θ)
f : A→ B + 1 in Test(Θ)
f : A→B in Par(B)
So (C, ) ' Par(B), the Kleisli category of (−) + 1 on B.
ExamplesB Par(B)
Classical Set PFunProbabilistic Kl(D) Kl(Dsub)Quantum CStaropcpu CStaropcpsu
![Page 57: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/57.jpg)
Equivalence Between Total and Partial Forms
We’ve seen operational categories in partial form (C, ) = ParTest(Θ)and total form B = Test(Θ).B = Ctotal, subcategory of f with ◦ f = , in fact determines all of C:
partial tests {fx}x∈Xtests {fx}x∈X ∪ {e : A→ I}
f : A→ B in ParTest(Θ)
f : A→ B + 1 in Test(Θ)
f : A→B in Par(B)
So (C, ) ' Par(B), the Kleisli category of (−) + 1 on B.
ExamplesB Par(B)
Classical Set PFunProbabilistic Kl(D) Kl(Dsub)Quantum CStaropcpu CStaropcpsu
![Page 58: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/58.jpg)
Equivalence Between Total and Partial Forms
We’ve seen operational categories in partial form (C, ) = ParTest(Θ)and total form B = Test(Θ).B = Ctotal, subcategory of f with ◦ f = , in fact determines all of C:
partial tests {fx}x∈Xtests {fx}x∈X ∪ {e : A→ I}
f : A→ B in ParTest(Θ)
f : A→ B + 1 in Test(Θ)
f : A→B in Par(B)
So (C, ) ' Par(B), the Kleisli category of (−) + 1 on B.
ExamplesB Par(B)
Classical Set PFunProbabilistic Kl(D) Kl(Dsub)Quantum CStaropcpu CStaropcpsu
![Page 59: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/59.jpg)
Equivalence Between Total and Partial Forms
We’ve seen operational categories in partial form (C, ) = ParTest(Θ)and total form B = Test(Θ).B = Ctotal, subcategory of f with ◦ f = , in fact determines all of C:
partial tests {fx}x∈Xtests {fx}x∈X ∪ {e : A→ I}
f : A→ B in ParTest(Θ)
f : A→ B + 1 in Test(Θ)
f : A→B in Par(B)
So (C, ) ' Par(B), the Kleisli category of (−) + 1 on B.
ExamplesB Par(B)
Classical Set PFunProbabilistic Kl(D) Kl(Dsub)Quantum CStaropcpu CStaropcpsu
![Page 60: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/60.jpg)
Equivalence Between Total and Partial Forms
We’ve seen operational categories in partial form (C, ) = ParTest(Θ)and total form B = Test(Θ).B = Ctotal, subcategory of f with ◦ f = , in fact determines all of C:
partial tests {fx}x∈Xtests {fx}x∈X ∪ {e : A→ I}
f : A→ B in ParTest(Θ)
f : A→ B + 1 in Test(Θ)
f : A→B in Par(B)
So (C, ) ' Par(B), the Kleisli category of (−) + 1 on B.
ExamplesB Par(B)
Classical Set PFunProbabilistic Kl(D) Kl(Dsub)Quantum CStaropcpu CStaropcpsu
![Page 61: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/61.jpg)
Equivalence Between Total and Partial Forms
We’ve seen operational categories in partial form (C, ) = ParTest(Θ)and total form B = Test(Θ).B = Ctotal, subcategory of f with ◦ f = , in fact determines all of C:
partial tests {fx}x∈Xtests {fx}x∈X ∪ {e : A→ I}
f : A→ B in ParTest(Θ)
f : A→ B + 1 in Test(Θ)
f : A→B in Par(B)
So (C, ) ' Par(B), the Kleisli category of (−) + 1 on B.
ExamplesB Par(B)
Classical Set PFunProbabilistic Kl(D) Kl(Dsub)Quantum CStaropcpu CStaropcpsu
![Page 62: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/62.jpg)
Equivalence Between Total and Partial Forms
We’ve seen operational categories in partial form (C, ) = ParTest(Θ)and total form B = Test(Θ).B = Ctotal, subcategory of f with ◦ f = , in fact determines all of C:
partial tests {fx}x∈Xtests {fx}x∈X ∪ {e : A→ I}
f : A→ B in ParTest(Θ)
f : A→ B + 1 in Test(Θ)
f : A→B in Par(B)
So (C, ) ' Par(B), the Kleisli category of (−) + 1 on B.
ExamplesB Par(B)
Classical Set PFunProbabilistic Kl(D) Kl(Dsub)Quantum CStaropcpu CStaropcpsu
![Page 63: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/63.jpg)
Summary
TheoremThe following structures are equivalent:
I an operational theory with control Θ with direct sums;
I an operational category in partial form (C, );
I an operational category B;
under C = EventΘ, Θ = OT(C), B ' Ctotal, C ' Par(B).
OTC ⊥ OTCdsum ' OpCatPar ' OpCat
(−)+
U
Event(−)
OT(−)
(−)total
Par(−)
![Page 64: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/64.jpg)
Summary
TheoremThe following structures are equivalent:
I an operational theory with control Θ with direct sums;
I an operational category in partial form (C, );
I an operational category B;
under C = EventΘ, Θ = OT(C), B ' Ctotal, C ' Par(B).
OTC ⊥ OTCdsum ' OpCatPar ' OpCat
(−)+
U
Event(−)
OT(−)
(−)total
Par(−)
![Page 65: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/65.jpg)
Summary
TheoremThe following structures are equivalent:
I an operational theory with control Θ with direct sums;
I an operational category in partial form (C, );
I an operational category B;
under C = EventΘ, Θ = OT(C), B ' Ctotal, C ' Par(B).
OTC ⊥ OTCdsum ' OpCatPar ' OpCat
(−)+
U
Event(−)
OT(−)
(−)total
Par(−)
![Page 66: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/66.jpg)
Effectus Theory
Approach comes from effectus theory (Jacobs et al.)A (monoidal) effectus B is an operational category with extra pullbacks:
A 1
A + B 1 + 1
!
κ1 κ1
!+!
A + B 1 + B
A + 1 1 + 1
!+id
id+! !+!
!+!
I Positivity: {fx}x∈X , {fx}x∈X ∪ {gy}y∈Y tests =⇒ each gy = 0.
I Observations determine tests: { ◦ fx}x∈X a test =⇒ {fx}x∈X atest.
Each space of effects e : A→ I forms an effect algebra.
![Page 67: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/67.jpg)
Effectus Theory
Approach comes from effectus theory (Jacobs et al.)
A (monoidal) effectus B is an operational category with extra pullbacks:
A 1
A + B 1 + 1
!
κ1 κ1
!+!
A + B 1 + B
A + 1 1 + 1
!+id
id+! !+!
!+!
I Positivity: {fx}x∈X , {fx}x∈X ∪ {gy}y∈Y tests =⇒ each gy = 0.
I Observations determine tests: { ◦ fx}x∈X a test =⇒ {fx}x∈X atest.
Each space of effects e : A→ I forms an effect algebra.
![Page 68: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/68.jpg)
Effectus Theory
Approach comes from effectus theory (Jacobs et al.)A (monoidal) effectus B is an operational category with extra pullbacks:
A 1
A + B 1 + 1
!
κ1 κ1
!+!
A + B 1 + B
A + 1 1 + 1
!+id
id+! !+!
!+!
I Positivity: {fx}x∈X , {fx}x∈X ∪ {gy}y∈Y tests =⇒ each gy = 0.
I Observations determine tests: { ◦ fx}x∈X a test =⇒ {fx}x∈X atest.
Each space of effects e : A→ I forms an effect algebra.
![Page 69: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/69.jpg)
Effectus Theory
Approach comes from effectus theory (Jacobs et al.)A (monoidal) effectus B is an operational category with extra pullbacks:
A 1
A + B 1 + 1
!
κ1 κ1
!+!
A + B 1 + B
A + 1 1 + 1
!+id
id+! !+!
!+!
I Positivity: {fx}x∈X , {fx}x∈X ∪ {gy}y∈Y tests =⇒ each gy = 0.
I Observations determine tests: { ◦ fx}x∈X a test =⇒ {fx}x∈X atest.
Each space of effects e : A→ I forms an effect algebra.
![Page 70: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/70.jpg)
Effectus Theory
Approach comes from effectus theory (Jacobs et al.)A (monoidal) effectus B is an operational category with extra pullbacks:
A 1
A + B 1 + 1
!
κ1 κ1
!+!
A + B 1 + B
A + 1 1 + 1
!+id
id+! !+!
!+!
I Positivity: {fx}x∈X , {fx}x∈X ∪ {gy}y∈Y tests =⇒ each gy = 0.
I Observations determine tests: { ◦ fx}x∈X a test =⇒ {fx}x∈X atest.
Each space of effects e : A→ I forms an effect algebra.
![Page 71: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/71.jpg)
Effectus Theory
Approach comes from effectus theory (Jacobs et al.)A (monoidal) effectus B is an operational category with extra pullbacks:
A 1
A + B 1 + 1
!
κ1 κ1
!+!
A + B 1 + B
A + 1 1 + 1
!+id
id+! !+!
!+!
I Positivity: {fx}x∈X , {fx}x∈X ∪ {gy}y∈Y tests =⇒ each gy = 0.
I Observations determine tests: { ◦ fx}x∈X a test =⇒ {fx}x∈X atest.
Each space of effects e : A→ I forms an effect algebra.
![Page 72: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/72.jpg)
Outlook
Why Operational Categories?
I Categorical logic for operationalism. Reason about operationaltheories as operational categories, and vice versa.
I Operational interpretation of effectus theory: general probabilisticcomputation.
I Connections: categorical quantum mechanics ↔ general probabilistictheories.
Next: translate quantum reconstruction theorems into categorical form -axiomatise FDimCStaropcpu.
Thanks!
![Page 73: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/73.jpg)
Outlook
Why Operational Categories?
I Categorical logic for operationalism. Reason about operationaltheories as operational categories, and vice versa.
I Operational interpretation of effectus theory: general probabilisticcomputation.
I Connections: categorical quantum mechanics ↔ general probabilistictheories.
Next: translate quantum reconstruction theorems into categorical form -axiomatise FDimCStaropcpu.
Thanks!
![Page 74: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/74.jpg)
Outlook
Why Operational Categories?
I Categorical logic for operationalism. Reason about operationaltheories as operational categories, and vice versa.
I Operational interpretation of effectus theory: general probabilisticcomputation.
I Connections: categorical quantum mechanics ↔ general probabilistictheories.
Next: translate quantum reconstruction theorems into categorical form -axiomatise FDimCStaropcpu.
Thanks!
![Page 75: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/75.jpg)
Outlook
Why Operational Categories?
I Categorical logic for operationalism. Reason about operationaltheories as operational categories, and vice versa.
I Operational interpretation of effectus theory: general probabilisticcomputation.
I Connections: categorical quantum mechanics ↔ general probabilistictheories.
Next: translate quantum reconstruction theorems into categorical form -axiomatise FDimCStaropcpu.
Thanks!
![Page 76: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/76.jpg)
Outlook
Why Operational Categories?
I Categorical logic for operationalism. Reason about operationaltheories as operational categories, and vice versa.
I Operational interpretation of effectus theory: general probabilisticcomputation.
I Connections: categorical quantum mechanics ↔ general probabilistictheories.
Next: translate quantum reconstruction theorems into categorical form -axiomatise FDimCStaropcpu.
Thanks!
![Page 77: Operational Physics, Logic and Categoriesconferences.inf.ed.ac.uk/clapscotland/tull.pdf · Operational Physics, Logic and Categories Sean Tull University of Oxford sean.tull@cs.ox.ac.uk](https://reader034.fdocuments.in/reader034/viewer/2022042402/5f112a8b8489bd29f660f39b/html5/thumbnails/77.jpg)
Outlook
Why Operational Categories?
I Categorical logic for operationalism. Reason about operationaltheories as operational categories, and vice versa.
I Operational interpretation of effectus theory: general probabilisticcomputation.
I Connections: categorical quantum mechanics ↔ general probabilistictheories.
Next: translate quantum reconstruction theorems into categorical form -axiomatise FDimCStaropcpu.
Thanks!