Luke Rogers - Physics@Technion

Luke Rogers - Physics@Technion Luke Rogers - Physics@Technion

physics.technion.ac.il
from physics.technion.ac.il More from this publisher
09.05.2013 Views

Dirichlet forms and derivations on fractals Luke Rogers University of Connecticut Luke Rogers Dirichlet forms and derivations on fractals

Dirichlet forms and derivations on fractals<br />

<strong>Luke</strong> <strong>Rogers</strong><br />

University of Connecticut<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Goal: Quantize fractal analysis<br />

There are ways to do analysis on certain fractal sets.<br />

Typical examples include self-similar gaskets.<br />

Heart of this analysis is Dirichlet form/Laplacian/Heat<br />

semigroup.<br />

Different properties than Euclidean: e.g. anomalous<br />

diffusion, localized eigenfunctions.<br />

Would like to quantize this analysis and obtain a quantum<br />

theory for fractal substrate.<br />

This talk is about an approach to doing this by following<br />

some methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Goal: Quantize fractal analysis<br />

There are ways to do analysis on certain fractal sets.<br />

Typical examples include self-similar gaskets.<br />

Heart of this analysis is Dirichlet form/Laplacian/Heat<br />

semigroup.<br />

Different properties than Euclidean: e.g. anomalous<br />

diffusion, localized eigenfunctions.<br />

Would like to quantize this analysis and obtain a quantum<br />

theory for fractal substrate.<br />

This talk is about an approach to doing this by following<br />

some methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Goal: Quantize fractal analysis<br />

There are ways to do analysis on certain fractal sets.<br />

Typical examples include self-similar gaskets.<br />

Heart of this analysis is Dirichlet form/Laplacian/Heat<br />

semigroup.<br />

Different properties than Euclidean: e.g. anomalous<br />

diffusion, localized eigenfunctions.<br />

Would like to quantize this analysis and obtain a quantum<br />

theory for fractal substrate.<br />

This talk is about an approach to doing this by following<br />

some methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Goal: Quantize fractal analysis<br />

There are ways to do analysis on certain fractal sets.<br />

Typical examples include self-similar gaskets.<br />

Heart of this analysis is Dirichlet form/Laplacian/Heat<br />

semigroup.<br />

Different properties than Euclidean: e.g. anomalous<br />

diffusion, localized eigenfunctions.<br />

Would like to quantize this analysis and obtain a quantum<br />

theory for fractal substrate.<br />

This talk is about an approach to doing this by following<br />

some methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Goal: Quantize fractal analysis<br />

There are ways to do analysis on certain fractal sets.<br />

Typical examples include self-similar gaskets.<br />

Heart of this analysis is Dirichlet form/Laplacian/Heat<br />

semigroup.<br />

Different properties than Euclidean: e.g. anomalous<br />

diffusion, localized eigenfunctions.<br />

Would like to quantize this analysis and obtain a quantum<br />

theory for fractal substrate.<br />

This talk is about an approach to doing this by following<br />

some methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Analysis on fractals via scaling limits<br />

We will get our Dirichlet form, Laplacian, and heat<br />

semigroup as scaling limits from graphs.<br />

Recall: Can get Brownian motion from rescaled Simple<br />

Random Walk on integer lattice.<br />

Similarly: Can get Laplacian as scaling limit of Laplacian<br />

on lattice.<br />

And classical Dirichlet form <br />

R2 |∇u| 2 as scaling limit of<br />

Dirichlet forms on lattice<br />

Same idea works on some fractals, using graphs that<br />

converge to the fractal.<br />

For fractals, scaling is not the usual Euclidean scaling.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Analysis on fractals via scaling limits<br />

We will get our Dirichlet form, Laplacian, and heat<br />

semigroup as scaling limits from graphs.<br />

Recall: Can get Brownian motion from rescaled Simple<br />

Random Walk on integer lattice.<br />

Similarly: Can get Laplacian as scaling limit of Laplacian<br />

on lattice.<br />

And classical Dirichlet form <br />

R2 |∇u| 2 as scaling limit of<br />

Dirichlet forms on lattice<br />

Same idea works on some fractals, using graphs that<br />

converge to the fractal.<br />

For fractals, scaling is not the usual Euclidean scaling.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Analysis on fractals via scaling limits<br />

We will get our Dirichlet form, Laplacian, and heat<br />

semigroup as scaling limits from graphs.<br />

Recall: Can get Brownian motion from rescaled Simple<br />

Random Walk on integer lattice.<br />

Similarly: Can get Laplacian as scaling limit of Laplacian<br />

on lattice.<br />

And classical Dirichlet form <br />

R2 |∇u| 2 as scaling limit of<br />

Dirichlet forms on lattice<br />

Same idea works on some fractals, using graphs that<br />

converge to the fractal.<br />

For fractals, scaling is not the usual Euclidean scaling.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Analysis on fractals via scaling limits<br />

We will get our Dirichlet form, Laplacian, and heat<br />

semigroup as scaling limits from graphs.<br />

Recall: Can get Brownian motion from rescaled Simple<br />

Random Walk on integer lattice.<br />

Similarly: Can get Laplacian as scaling limit of Laplacian<br />

on lattice.<br />

And classical Dirichlet form <br />

R2 |∇u| 2 as scaling limit of<br />

Dirichlet forms on lattice<br />

Same idea works on some fractals, using graphs that<br />

converge to the fractal.<br />

For fractals, scaling is not the usual Euclidean scaling.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Analysis on fractals via scaling limits<br />

We will get our Dirichlet form, Laplacian, and heat<br />

semigroup as scaling limits from graphs.<br />

Recall: Can get Brownian motion from rescaled Simple<br />

Random Walk on integer lattice.<br />

Similarly: Can get Laplacian as scaling limit of Laplacian<br />

on lattice.<br />

And classical Dirichlet form <br />

R2 |∇u| 2 as scaling limit of<br />

Dirichlet forms on lattice<br />

Same idea works on some fractals, using graphs that<br />

converge to the fractal.<br />

For fractals, scaling is not the usual Euclidean scaling.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

Sierpinski Gasket<br />

Em(u) =<br />

2 u(y) − u(x)<br />

y∼0x<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

Zero level graph<br />

Dirichlet Form E0(u, u) =<br />

2 u(y) − u(x)<br />

y∼0x<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

First level graph<br />

Dirichlet Form E1(u, u) =<br />

2 u(y) − u(x)<br />

y∼1x<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

Second level graph<br />

Dirichlet Form E2(u, u) =<br />

2 u(y) − u(x)<br />

y∼2x<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

Sierpinski Gasket<br />

Dirichlet Form E(u, u) = lim<br />

m→∞<br />

<br />

5<br />

m 2 u(y) − u(x)<br />

3<br />

y∼mx<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

Zero level graph<br />

Laplacian ∆0u(x) =<br />

<br />

u(y) − u(x)<br />

y∼0x<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

First level graph<br />

Laplacian ∆1u(x) =<br />

<br />

u(y) − u(x)<br />

y∼1x<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

Second level graph<br />

Laplacian ∆2u(x) =<br />

<br />

u(y) − u(x)<br />

y∼2x<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

Sierpinski Gasket<br />

Laplacian ∆u(x) = 3<br />

2 lim<br />

m→∞<br />

5m <br />

y∼mx<br />

u(y) − u(x) <br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

Sierpinski Gasket<br />

<br />

Gauss-Green: E(u, v) = −<br />

(∆u)v dµ<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Scaling limits<br />

Sierpinski Gasket<br />

<br />

Gauss-Green: E(u, v) = −<br />

(∆u)v dµ +<br />

<br />

boundary<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals<br />

(du(x))v(x)


Harmonic and n-harmonic functions<br />

If we prescribe values at 3 boundary vertices of Sierpinski<br />

Gasket, then there is a unique function u with minimal<br />

value of E(u, u) and those boundary values.<br />

This minimizer is called harmonic. It has ∆u = 0.<br />

If prescribe values at all scale n vertices, there is a unique<br />

minimizer of E(u, u).<br />

It is called n-harmonic and has ∆u = 0 except at scale n<br />

vertices, where ∆u has point masses.<br />

Space of n-harmonic functions is a good basis when<br />

working with E as it is dense (when using E as inner<br />

product). It is also dense in C(X) and L 2 (X).<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Harmonic and n-harmonic functions<br />

If we prescribe values at 3 boundary vertices of Sierpinski<br />

Gasket, then there is a unique function u with minimal<br />

value of E(u, u) and those boundary values.<br />

This minimizer is called harmonic. It has ∆u = 0.<br />

If prescribe values at all scale n vertices, there is a unique<br />

minimizer of E(u, u).<br />

It is called n-harmonic and has ∆u = 0 except at scale n<br />

vertices, where ∆u has point masses.<br />

Space of n-harmonic functions is a good basis when<br />

working with E as it is dense (when using E as inner<br />

product). It is also dense in C(X) and L 2 (X).<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Harmonic and n-harmonic functions<br />

If we prescribe values at 3 boundary vertices of Sierpinski<br />

Gasket, then there is a unique function u with minimal<br />

value of E(u, u) and those boundary values.<br />

This minimizer is called harmonic. It has ∆u = 0.<br />

If prescribe values at all scale n vertices, there is a unique<br />

minimizer of E(u, u).<br />

It is called n-harmonic and has ∆u = 0 except at scale n<br />

vertices, where ∆u has point masses.<br />

Space of n-harmonic functions is a good basis when<br />

working with E as it is dense (when using E as inner<br />

product). It is also dense in C(X) and L 2 (X).<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Harmonic and n-harmonic functions<br />

If we prescribe values at 3 boundary vertices of Sierpinski<br />

Gasket, then there is a unique function u with minimal<br />

value of E(u, u) and those boundary values.<br />

This minimizer is called harmonic. It has ∆u = 0.<br />

If prescribe values at all scale n vertices, there is a unique<br />

minimizer of E(u, u).<br />

It is called n-harmonic and has ∆u = 0 except at scale n<br />

vertices, where ∆u has point masses.<br />

Space of n-harmonic functions is a good basis when<br />

working with E as it is dense (when using E as inner<br />

product). It is also dense in C(X) and L 2 (X).<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Fractal Analysis<br />

There is a good class of fractals for which approach on<br />

preceding slide is applicable (post-critically finite self-similar).<br />

When I say fractal analysis in this talk, I always have in mind<br />

that situation:<br />

A fractal set X.<br />

A Dirichlet form E with domain F (all functions in domain<br />

are continuous).<br />

Equivalently a Laplacian ∆, or a heat flow e −t∆ , or a<br />

Brownian motion.<br />

Note: It is not clear how to apply this analysis to infinitely<br />

ramified fractals like Sierpinski carpets.<br />

Recall that (eventual) goal is to quantize fractal analysis using<br />

methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Fractal Analysis<br />

There is a good class of fractals for which approach on<br />

preceding slide is applicable (post-critically finite self-similar).<br />

When I say fractal analysis in this talk, I always have in mind<br />

that situation:<br />

A fractal set X.<br />

A Dirichlet form E with domain F (all functions in domain<br />

are continuous).<br />

Equivalently a Laplacian ∆, or a heat flow e −t∆ , or a<br />

Brownian motion.<br />

Note: It is not clear how to apply this analysis to infinitely<br />

ramified fractals like Sierpinski carpets.<br />

Recall that (eventual) goal is to quantize fractal analysis using<br />

methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Fractal Analysis<br />

There is a good class of fractals for which approach on<br />

preceding slide is applicable (post-critically finite self-similar).<br />

When I say fractal analysis in this talk, I always have in mind<br />

that situation:<br />

A fractal set X.<br />

A Dirichlet form E with domain F (all functions in domain<br />

are continuous).<br />

Equivalently a Laplacian ∆, or a heat flow e −t∆ , or a<br />

Brownian motion.<br />

Note: It is not clear how to apply this analysis to infinitely<br />

ramified fractals like Sierpinski carpets.<br />

Recall that (eventual) goal is to quantize fractal analysis using<br />

methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Fractal Analysis<br />

There is a good class of fractals for which approach on<br />

preceding slide is applicable (post-critically finite self-similar).<br />

When I say fractal analysis in this talk, I always have in mind<br />

that situation:<br />

A fractal set X.<br />

A Dirichlet form E with domain F (all functions in domain<br />

are continuous).<br />

Equivalently a Laplacian ∆, or a heat flow e −t∆ , or a<br />

Brownian motion.<br />

Note: It is not clear how to apply this analysis to infinitely<br />

ramified fractals like Sierpinski carpets.<br />

Recall that (eventual) goal is to quantize fractal analysis using<br />

methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Fractal Analysis<br />

There is a good class of fractals for which approach on<br />

preceding slide is applicable (post-critically finite self-similar).<br />

When I say fractal analysis in this talk, I always have in mind<br />

that situation:<br />

A fractal set X.<br />

A Dirichlet form E with domain F (all functions in domain<br />

are continuous).<br />

Equivalently a Laplacian ∆, or a heat flow e −t∆ , or a<br />

Brownian motion.<br />

Note: It is not clear how to apply this analysis to infinitely<br />

ramified fractals like Sierpinski carpets.<br />

Recall that (eventual) goal is to quantize fractal analysis using<br />

methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Fractal Analysis<br />

There is a good class of fractals for which approach on<br />

preceding slide is applicable (post-critically finite self-similar).<br />

When I say fractal analysis in this talk, I always have in mind<br />

that situation:<br />

A fractal set X.<br />

A Dirichlet form E with domain F (all functions in domain<br />

are continuous).<br />

Equivalently a Laplacian ∆, or a heat flow e −t∆ , or a<br />

Brownian motion.<br />

Note: It is not clear how to apply this analysis to infinitely<br />

ramified fractals like Sierpinski carpets.<br />

Recall that (eventual) goal is to quantize fractal analysis using<br />

methods from non-commutative geometry.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Non-commutative geometry: motivational sketch<br />

Physical problem: In quantum world, co-ordinates do not<br />

commute.<br />

Solution: Replace co-ordinates by operators on Hilbert<br />

space.<br />

Problem: We also want fields.<br />

Problem: Interesting spaces are curved.<br />

Non-commutative geometry: Geometry and field theory on<br />

“spaces” where co-ordinates (i.e. functions) do not<br />

commute.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Non-commutative geometry: motivational sketch<br />

Physical problem: In quantum world, co-ordinates do not<br />

commute.<br />

Solution: Replace co-ordinates by operators on Hilbert<br />

space.<br />

Problem: We also want fields.<br />

Problem: Interesting spaces are curved.<br />

Non-commutative geometry: Geometry and field theory on<br />

“spaces” where co-ordinates (i.e. functions) do not<br />

commute.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Non-commutative geometry: motivational sketch<br />

Physical problem: In quantum world, co-ordinates do not<br />

commute.<br />

Solution: Replace co-ordinates by operators on Hilbert<br />

space.<br />

Problem: We also want fields.<br />

Problem: Interesting spaces are curved.<br />

Non-commutative geometry: Geometry and field theory on<br />

“spaces” where co-ordinates (i.e. functions) do not<br />

commute.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Non-commutative geometry: motivational sketch<br />

Physical problem: In quantum world, co-ordinates do not<br />

commute.<br />

Solution: Replace co-ordinates by operators on Hilbert<br />

space.<br />

Problem: We also want fields.<br />

Problem: Interesting spaces are curved.<br />

Non-commutative geometry: Geometry and field theory on<br />

“spaces” where co-ordinates (i.e. functions) do not<br />

commute.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Non-commutative geometry: motivational sketch<br />

Physical problem: In quantum world, co-ordinates do not<br />

commute.<br />

Solution: Replace co-ordinates by operators on Hilbert<br />

space.<br />

Problem: We also want fields.<br />

Problem: Interesting spaces are curved.<br />

Non-commutative geometry: Geometry and field theory on<br />

“spaces” where co-ordinates (i.e. functions) do not<br />

commute.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Geometry via algebra<br />

Non-commutative geometry: write geometric notions (e.g.<br />

curvature of manifold M) in terms of an algebra of<br />

functions on M. (e.g. algebra C(M) of cts functions on M).<br />

If the algebraic versions of these geometric ideas do not<br />

require commutativity of the algebra one then has<br />

“geometry” for non-commutative algebras of functions.<br />

Goal: Do something similar for fractal analysis: express<br />

analysis in terms of natural algebras of functions (e.g. cts<br />

functions or functions of finite energy).<br />

Next Step: replace the commutative algebras that come up<br />

in analysis on fractals with physically meaningful<br />

non-commutative ones.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Geometry via algebra<br />

Non-commutative geometry: write geometric notions (e.g.<br />

curvature of manifold M) in terms of an algebra of<br />

functions on M. (e.g. algebra C(M) of cts functions on M).<br />

If the algebraic versions of these geometric ideas do not<br />

require commutativity of the algebra one then has<br />

“geometry” for non-commutative algebras of functions.<br />

Goal: Do something similar for fractal analysis: express<br />

analysis in terms of natural algebras of functions (e.g. cts<br />

functions or functions of finite energy).<br />

Next Step: replace the commutative algebras that come up<br />

in analysis on fractals with physically meaningful<br />

non-commutative ones.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Geometry via algebra<br />

Non-commutative geometry: write geometric notions (e.g.<br />

curvature of manifold M) in terms of an algebra of<br />

functions on M. (e.g. algebra C(M) of cts functions on M).<br />

If the algebraic versions of these geometric ideas do not<br />

require commutativity of the algebra one then has<br />

“geometry” for non-commutative algebras of functions.<br />

Goal: Do something similar for fractal analysis: express<br />

analysis in terms of natural algebras of functions (e.g. cts<br />

functions or functions of finite energy).<br />

Next Step: replace the commutative algebras that come up<br />

in analysis on fractals with physically meaningful<br />

non-commutative ones.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Geometry via algebra<br />

Non-commutative geometry: write geometric notions (e.g.<br />

curvature of manifold M) in terms of an algebra of<br />

functions on M. (e.g. algebra C(M) of cts functions on M).<br />

If the algebraic versions of these geometric ideas do not<br />

require commutativity of the algebra one then has<br />

“geometry” for non-commutative algebras of functions.<br />

Goal: Do something similar for fractal analysis: express<br />

analysis in terms of natural algebras of functions (e.g. cts<br />

functions or functions of finite energy).<br />

Next Step: replace the commutative algebras that come up<br />

in analysis on fractals with physically meaningful<br />

non-commutative ones.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Geometry via algebra<br />

Non-commutative geometry: write geometric notions (e.g.<br />

curvature of manifold M) in terms of an algebra of<br />

functions on M. (e.g. algebra C(M) of cts functions on M).<br />

If the algebraic versions of these geometric ideas do not<br />

require commutativity of the algebra one then has<br />

“geometry” for non-commutative algebras of functions.<br />

Goal: Do something similar for fractal analysis: express<br />

analysis in terms of natural algebras of functions (e.g. cts<br />

functions or functions of finite energy).<br />

Next Step: replace the commutative algebras that come up<br />

in analysis on fractals with physically meaningful<br />

non-commutative ones.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Topology via algebra<br />

Gelfand: the property of being a compact Hausdorff space can<br />

be expressed in terms of algebra.<br />

SPACE ALGEBRA<br />

X compact Hausdorff A unital commutative C ∗<br />

Isomorphisms:<br />

X Hom(C(X), C) via x ↦→ point evaluation at x.<br />

A C Hom(A, C) via a ↦→ â with â(φ) = φ(a) where a ∈ A and<br />

φ ∈ Hom(A, C).<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Topology via algebra<br />

Gelfand: the property of being a compact Hausdorff space can<br />

be expressed in terms of algebra.<br />

SPACE ALGEBRA<br />

X compact Hausdorff<br />

A = C(X)<br />

✲<br />

✛<br />

X = Hom(A, C)<br />

A unital commutative C ∗<br />

Isomorphisms:<br />

X Hom(C(X), C) via x ↦→ point evaluation at x.<br />

A C Hom(A, C) via a ↦→ â with â(φ) = φ(a) where a ∈ A and<br />

φ ∈ Hom(A, C).<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Topology via algebra<br />

Gelfand: the property of being a compact Hausdorff space can<br />

be expressed in terms of algebra.<br />

SPACE ALGEBRA<br />

X compact Hausdorff<br />

A = C(X)<br />

✲<br />

✛<br />

X = Hom(A, C)<br />

A unital commutative C ∗<br />

Isomorphisms:<br />

X Hom(C(X), C) via x ↦→ point evaluation at x.<br />

A C Hom(A, C) via a ↦→ â with â(φ) = φ(a) where a ∈ A and<br />

φ ∈ Hom(A, C).<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Riemannian geometry via algebra (Connes)<br />

SPACE ALGEBRA<br />

X compact spin manifold H Hilbert space<br />

g Riemannian metric A algebra of operators on H<br />

D self-adjoint operator on H<br />

Some Properties<br />

Forward map:<br />

H is the spinor bundle<br />

A = C(X)<br />

D is Dirac operator (square root of Laplacian)<br />

(H, A, D) called a “spectral triple’<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Riemannian geometry via algebra (Connes)<br />

SPACE ALGEBRA<br />

X compact spin manifold<br />

g Riemannian metric<br />

Forward map:<br />

H is the spinor bundle<br />

✛<br />

✲ H Hilbert space<br />

A algebra of operators on H<br />

D self-adjoint operator on H<br />

Some Properties<br />

A = C(X) (acting by multiplication)<br />

D is Dirac operator (square root of Laplacian)<br />

(H, A, D) called a “spectral triple”<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Recovering distance from algebra<br />

Define differential da = [D, a] = Da − aD.<br />

1-Lipschitz function a has <br />

<br />

op<br />

[D, a] ≤ 1.<br />

<br />

Distance sup |a(x) − a(y)| : all a ∈ A, <br />

<br />

op<br />

[D, a] ≤ 1 .<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Recovering distance from algebra<br />

Define differential da = [D, a] = Da − aD.<br />

1-Lipschitz function a has <br />

<br />

op<br />

[D, a] ≤ 1.<br />

<br />

Distance sup |a(x) − a(y)| : all a ∈ A, <br />

<br />

op<br />

[D, a] ≤ 1 .<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Recovering distance from algebra<br />

Define differential da = [D, a] = Da − aD.<br />

1-Lipschitz function a has <br />

<br />

op<br />

[D, a] ≤ 1.<br />

<br />

Distance sup |a(x) − a(y)| : all a ∈ A, <br />

<br />

op<br />

[D, a] ≤ 1 .<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Recovering volume from algebra<br />

Compact operators play role of infinitesimals<br />

If T is a compact operator, λn(T ) is n th singular value<br />

Integration is given by application of a trace operator, the<br />

Dixmier trace, which is a version of<br />

1<br />

lim<br />

N→∞ log N<br />

N<br />

λn(T )<br />

0<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Recovering volume from algebra<br />

Compact operators play role of infinitesimals<br />

If T is a compact operator, λn(T ) is n th singular value<br />

Integration is given by application of a trace operator, the<br />

Dixmier trace, which is a version of<br />

1<br />

lim<br />

N→∞ log N<br />

N<br />

λn(T )<br />

0<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Recovering volume from algebra<br />

Compact operators play role of infinitesimals<br />

If T is a compact operator, λn(T ) is n th singular value<br />

Integration is given by application of a trace operator, the<br />

Dixmier trace, which is a version of<br />

1<br />

lim<br />

N→∞ log N<br />

N<br />

λn(T )<br />

0<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Spectral triple for fractal analysis<br />

Cipriani-Sauvageot (2007):<br />

Start with fractal X having a Dirichlet form E.<br />

Algebras of functions C(X) (continuous functions) and F<br />

(those with E(a, a) < ∞).<br />

Build Hilbert module H (1-forms).<br />

Build derivation ∂ : F → H.<br />

Has Leibniz rule ∂(ab) = a(∂b) + (∂a)b.<br />

Point is that E(a, a) = ∂a 2<br />

H .<br />

(compare to Euclidean case E(a) = |∇a| 2 .)<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Spectral triple for fractal analysis<br />

Cipriani-Sauvageot (2007):<br />

Start with fractal X having a Dirichlet form E.<br />

Algebras of functions C(X) (continuous functions) and F<br />

(those with E(a, a) < ∞).<br />

Build Hilbert module H (1-forms).<br />

Build derivation ∂ : F → H.<br />

Has Leibniz rule ∂(ab) = a(∂b) + (∂a)b.<br />

Point is that E(a, a) = ∂a 2<br />

H .<br />

(compare to Euclidean case E(a) = |∇a| 2 .)<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Spectral triple for fractal analysis<br />

Cipriani-Sauvageot (2007):<br />

Start with fractal X having a Dirichlet form E.<br />

Algebras of functions C(X) (continuous functions) and F<br />

(those with E(a, a) < ∞).<br />

Build Hilbert module H (1-forms).<br />

Build derivation ∂ : F → H.<br />

Has Leibniz rule ∂(ab) = a(∂b) + (∂a)b.<br />

Point is that E(a, a) = ∂a 2<br />

H .<br />

(compare to Euclidean case E(a) = |∇a| 2 .)<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Spectral triple for fractal analysis<br />

Cipriani-Sauvageot (2007):<br />

Start with fractal X having a Dirichlet form E.<br />

Algebras of functions C(X) (continuous functions) and F<br />

(those with E(a, a) < ∞).<br />

Build Hilbert module H (1-forms).<br />

Build derivation ∂ : F → H.<br />

Has Leibniz rule ∂(ab) = a(∂b) + (∂a)b.<br />

Point is that E(a, a) = ∂a 2<br />

H .<br />

(compare to Euclidean case E(a) = |∇a| 2 .)<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Spectral triple for fractal analysis<br />

Cipriani-Sauvageot (2007):<br />

Start with fractal X having a Dirichlet form E.<br />

Algebras of functions C(X) (continuous functions) and F<br />

(those with E(a, a) < ∞).<br />

Build Hilbert module H (1-forms).<br />

Build derivation ∂ : F → H.<br />

Has Leibniz rule ∂(ab) = a(∂b) + (∂a)b.<br />

Point is that E(a, a) = ∂a 2<br />

H .<br />

(compare to Euclidean case E(a) = |∇a| 2 .)<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Spectral triple for fractal analysis<br />

SPACE ALGEBRA<br />

X fractal<br />

H Hilbert space<br />

E Dirichlet form A algebra of operators on H<br />

J phase operator on H<br />

H is a quotient of F ⊗ F<br />

A = C(X) (acting by multiplication)<br />

J is sign of Dirac operator (so J = D|D| −1 )<br />

(H, A, J) called a “spectral triple’<br />

There is a derivation ∂ : F → H so E(a, a) = da 2<br />

H<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Spectral triple for fractal analysis<br />

SPACE ALGEBRA<br />

X fractal<br />

E Dirichlet form<br />

✛<br />

✲ H Hilbert space<br />

A algebra of operators on H<br />

J phase operator on H<br />

H is a certain quotient of F ⊗ F<br />

A = C(X) (acting by multiplication)<br />

J is phase of Dirac operator (so J = D|D| −1 )<br />

(H, A, J) called a “spectral triple’<br />

There is a derivation ∂ : F → H so E(a, a) = da 2<br />

H<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


H and the derivation<br />

The space H is supposed to be the 1-forms, but its definition is<br />

somewhat abstract.<br />

Construction:<br />

Begin with space F ⊗ F and let<br />

〈a ⊗ b, c ⊗ d〉H = 1<br />

<br />

<br />

E(a, cdb) + E(abd, c) − E(bd, ac) .<br />

2<br />

Quotient out by zero-norm subspace. Take completion to<br />

get H.<br />

Define<br />

a(b ⊗ c) = (ab) ⊗ c − a ⊗ (bc)<br />

(b ⊗ c)d = b ⊗ (cd)<br />

and extend to H by linearity, density and continuity.<br />

Let ∂ : F → H be map a ↦→ a ⊗ 1.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


H and the derivation<br />

The space H is supposed to be the 1-forms, but its definition is<br />

somewhat abstract.<br />

Construction:<br />

Begin with space F ⊗ F and let<br />

〈a ⊗ b, c ⊗ d〉H = 1<br />

<br />

<br />

E(a, cdb) + E(abd, c) − E(bd, ac) .<br />

2<br />

Quotient out by zero-norm subspace. Take completion to<br />

get H.<br />

Define<br />

a(b ⊗ c) = (ab) ⊗ c − a ⊗ (bc)<br />

(b ⊗ c)d = b ⊗ (cd)<br />

and extend to H by linearity, density and continuity.<br />

Let ∂ : F → H be map a ↦→ a ⊗ 1.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


H and the derivation<br />

The space H is supposed to be the 1-forms, but its definition is<br />

somewhat abstract.<br />

Construction:<br />

Begin with space F ⊗ F and let<br />

〈a ⊗ b, c ⊗ d〉H = 1<br />

<br />

<br />

E(a, cdb) + E(abd, c) − E(bd, ac) .<br />

2<br />

Quotient out by zero-norm subspace. Take completion to<br />

get H.<br />

Define<br />

a(b ⊗ c) = (ab) ⊗ c − a ⊗ (bc)<br />

(b ⊗ c)d = b ⊗ (cd)<br />

and extend to H by linearity, density and continuity.<br />

Let ∂ : F → H be map a ↦→ a ⊗ 1.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


H and the derivation<br />

The space H is supposed to be the 1-forms, but its definition is<br />

somewhat abstract.<br />

Construction:<br />

Begin with space F ⊗ F and let<br />

〈a ⊗ b, c ⊗ d〉H = 1<br />

<br />

<br />

E(a, cdb) + E(abd, c) − E(bd, ac) .<br />

2<br />

Quotient out by zero-norm subspace. Take completion to<br />

get H.<br />

Define<br />

a(b ⊗ c) = (ab) ⊗ c − a ⊗ (bc)<br />

(b ⊗ c)d = b ⊗ (cd)<br />

and extend to H by linearity, density and continuity.<br />

Let ∂ : F → H be map a ↦→ a ⊗ 1.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


H and the derivation<br />

The space H is supposed to be the 1-forms, but its definition is<br />

somewhat abstract.<br />

Construction:<br />

Begin with space F ⊗ F and let<br />

〈a ⊗ b, c ⊗ d〉H = 1<br />

<br />

<br />

E(a, cdb) + E(abd, c) − E(bd, ac) .<br />

2<br />

Quotient out by zero-norm subspace. Take completion to<br />

get H.<br />

Define<br />

a(b ⊗ c) = (ab) ⊗ c − a ⊗ (bc)<br />

(b ⊗ c)d = b ⊗ (cd)<br />

and extend to H by linearity, density and continuity.<br />

Let ∂ : F → H be map a ↦→ a ⊗ 1.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Understanding H with n-harmonic functions<br />

Take a scale n cell Xα.<br />

For prescribed values on boundary of this cell, take any<br />

n-harmonic function hα with these values.<br />

Then hα ⊗ 1α is an element of H. Think of as just hα on the<br />

cell and identically zero elsewhere.<br />

Scale n elements of H are sums of these functions; think<br />

of them as harmonic on each of the scale n cells, but with<br />

no matching conditions at the intersections of cells.<br />

Any element of H is a limit (in H norm) of such scale n<br />

functions.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Understanding H with n-harmonic functions<br />

Take a scale n cell Xα.<br />

For prescribed values on boundary of this cell, take any<br />

n-harmonic function hα with these values.<br />

Then hα ⊗ 1α is an element of H. Think of as just hα on the<br />

cell and identically zero elsewhere.<br />

Scale n elements of H are sums of these functions; think<br />

of them as harmonic on each of the scale n cells, but with<br />

no matching conditions at the intersections of cells.<br />

Any element of H is a limit (in H norm) of such scale n<br />

functions.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Understanding H with n-harmonic functions<br />

Take a scale n cell Xα.<br />

For prescribed values on boundary of this cell, take any<br />

n-harmonic function hα with these values.<br />

Then hα ⊗ 1α is an element of H. Think of as just hα on the<br />

cell and identically zero elsewhere.<br />

Scale n elements of H are sums of these functions; think<br />

of them as harmonic on each of the scale n cells, but with<br />

no matching conditions at the intersections of cells.<br />

Any element of H is a limit (in H norm) of such scale n<br />

functions.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Understanding H with n-harmonic functions<br />

Take a scale n cell Xα.<br />

For prescribed values on boundary of this cell, take any<br />

n-harmonic function hα with these values.<br />

Then hα ⊗ 1α is an element of H. Think of as just hα on the<br />

cell and identically zero elsewhere.<br />

Scale n elements of H are sums of these functions; think<br />

of them as harmonic on each of the scale n cells, but with<br />

no matching conditions at the intersections of cells.<br />

Any element of H is a limit (in H norm) of such scale n<br />

functions.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Understanding H with n-harmonic functions<br />

Take a scale n cell Xα.<br />

For prescribed values on boundary of this cell, take any<br />

n-harmonic function hα with these values.<br />

Then hα ⊗ 1α is an element of H. Think of as just hα on the<br />

cell and identically zero elsewhere.<br />

Scale n elements of H are sums of these functions; think<br />

of them as harmonic on each of the scale n cells, but with<br />

no matching conditions at the intersections of cells.<br />

Any element of H is a limit (in H norm) of such scale n<br />

functions.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Understanding ∂ with n-harmonic functions<br />

Any function a ∈ F (i.e. with E(a, a) < ∞) can be<br />

approximated in sense of E by n-harmonic functions.<br />

The map ∂ takes a to a ⊗ 1.<br />

So if a is n-harmonic then ∂a = a ⊗ 1 and everything from<br />

F can be approximated by these functions.<br />

H is limits of functions harmonic on scale n cells, with no<br />

matching conditions at vertices.<br />

Image of ∂ is limits of functions harmonic on scale n cells,<br />

with continuous matching at vertices.<br />

Using the Gauss-Green formula (locally) we find that the<br />

orthogonal space to image of ∂ is limits of functions<br />

harmonic on scale n cells, with normal derivatives<br />

summing to zero at vertices.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Understanding ∂ with n-harmonic functions<br />

Any function a ∈ F (i.e. with E(a, a) < ∞) can be<br />

approximated in sense of E by n-harmonic functions.<br />

The map ∂ takes a to a ⊗ 1.<br />

So if a is n-harmonic then ∂a = a ⊗ 1 and everything from<br />

F can be approximated by these functions.<br />

H is limits of functions harmonic on scale n cells, with no<br />

matching conditions at vertices.<br />

Image of ∂ is limits of functions harmonic on scale n cells,<br />

with continuous matching at vertices.<br />

Using the Gauss-Green formula (locally) we find that the<br />

orthogonal space to image of ∂ is limits of functions<br />

harmonic on scale n cells, with normal derivatives<br />

summing to zero at vertices.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Understanding ∂ with n-harmonic functions<br />

Any function a ∈ F (i.e. with E(a, a) < ∞) can be<br />

approximated in sense of E by n-harmonic functions.<br />

The map ∂ takes a to a ⊗ 1.<br />

So if a is n-harmonic then ∂a = a ⊗ 1 and everything from<br />

F can be approximated by these functions.<br />

H is limits of functions harmonic on scale n cells, with no<br />

matching conditions at vertices.<br />

Image of ∂ is limits of functions harmonic on scale n cells,<br />

with continuous matching at vertices.<br />

Using the Gauss-Green formula (locally) we find that the<br />

orthogonal space to image of ∂ is limits of functions<br />

harmonic on scale n cells, with normal derivatives<br />

summing to zero at vertices.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Understanding ∂ with n-harmonic functions<br />

Any function a ∈ F (i.e. with E(a, a) < ∞) can be<br />

approximated in sense of E by n-harmonic functions.<br />

The map ∂ takes a to a ⊗ 1.<br />

So if a is n-harmonic then ∂a = a ⊗ 1 and everything from<br />

F can be approximated by these functions.<br />

H is limits of functions harmonic on scale n cells, with no<br />

matching conditions at vertices.<br />

Image of ∂ is limits of functions harmonic on scale n cells,<br />

with continuous matching at vertices.<br />

Using the Gauss-Green formula (locally) we find that the<br />

orthogonal space to image of ∂ is limits of functions<br />

harmonic on scale n cells, with normal derivatives<br />

summing to zero at vertices.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Understanding ∂ with n-harmonic functions<br />

Any function a ∈ F (i.e. with E(a, a) < ∞) can be<br />

approximated in sense of E by n-harmonic functions.<br />

The map ∂ takes a to a ⊗ 1.<br />

So if a is n-harmonic then ∂a = a ⊗ 1 and everything from<br />

F can be approximated by these functions.<br />

H is limits of functions harmonic on scale n cells, with no<br />

matching conditions at vertices.<br />

Image of ∂ is limits of functions harmonic on scale n cells,<br />

with continuous matching at vertices.<br />

Using the Gauss-Green formula (locally) we find that the<br />

orthogonal space to image of ∂ is limits of functions<br />

harmonic on scale n cells, with normal derivatives<br />

summing to zero at vertices.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


DeRham cohomology and Hodge Theorem<br />

The result on the previous page says that we can recover<br />

the topology (the loops) of X from H.<br />

In mathematical terms, the map ∂ is the exterior derivative<br />

for a DeRham complex, and a version of the Hodge<br />

theorem is true.<br />

One consequence: Map ∂ is surjective iff there are no<br />

loops, i.e. X is a tree.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


DeRham cohomology and Hodge Theorem<br />

The result on the previous page says that we can recover<br />

the topology (the loops) of X from H.<br />

In mathematical terms, the map ∂ is the exterior derivative<br />

for a DeRham complex, and a version of the Hodge<br />

theorem is true.<br />

One consequence: Map ∂ is surjective iff there are no<br />

loops, i.e. X is a tree.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


DeRham cohomology and Hodge Theorem<br />

The result on the previous page says that we can recover<br />

the topology (the loops) of X from H.<br />

In mathematical terms, the map ∂ is the exterior derivative<br />

for a DeRham complex, and a version of the Hodge<br />

theorem is true.<br />

One consequence: Map ∂ is surjective iff there are no<br />

loops, i.e. X is a tree.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


The phase map J<br />

Map J is supposed to be the phase of the Dirac operator.<br />

Let P be projection in H onto image of ∂ and P ⊥ the<br />

orthogonal projection.<br />

Define J = P − P ⊥ .<br />

We can understand J very explicitly using preceding<br />

description of ∂.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


The phase map J<br />

Map J is supposed to be the phase of the Dirac operator.<br />

Let P be projection in H onto image of ∂ and P ⊥ the<br />

orthogonal projection.<br />

Define J = P − P ⊥ .<br />

We can understand J very explicitly using preceding<br />

description of ∂.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


The phase map J<br />

Map J is supposed to be the phase of the Dirac operator.<br />

Let P be projection in H onto image of ∂ and P ⊥ the<br />

orthogonal projection.<br />

Define J = P − P ⊥ .<br />

We can understand J very explicitly using preceding<br />

description of ∂.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Commutator [J, a]<br />

Initially consider commutator [J, a] of J with multiplication<br />

by a ∈ C(X).<br />

We can permit discontinuities of a at vertices of cells.<br />

If a is constant on scale n cells then [J, a]h = 0 for any h<br />

that is orthogonal to n-harmonic functions<br />

Can approximate any [J, a] in operator norm by operators<br />

with finite dimensional image<br />

Hence [J, a] is a compact operator.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Commutator [J, a]<br />

Initially consider commutator [J, a] of J with multiplication<br />

by a ∈ C(X).<br />

We can permit discontinuities of a at vertices of cells.<br />

If a is constant on scale n cells then [J, a]h = 0 for any h<br />

that is orthogonal to n-harmonic functions<br />

Can approximate any [J, a] in operator norm by operators<br />

with finite dimensional image<br />

Hence [J, a] is a compact operator.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Commutator [J, a]<br />

Initially consider commutator [J, a] of J with multiplication<br />

by a ∈ C(X).<br />

We can permit discontinuities of a at vertices of cells.<br />

If a is constant on scale n cells then [J, a]h = 0 for any h<br />

that is orthogonal to n-harmonic functions<br />

Can approximate any [J, a] in operator norm by operators<br />

with finite dimensional image<br />

Hence [J, a] is a compact operator.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Commutator [J, a]<br />

Initially consider commutator [J, a] of J with multiplication<br />

by a ∈ C(X).<br />

We can permit discontinuities of a at vertices of cells.<br />

If a is constant on scale n cells then [J, a]h = 0 for any h<br />

that is orthogonal to n-harmonic functions<br />

Can approximate any [J, a] in operator norm by operators<br />

with finite dimensional image<br />

Hence [J, a] is a compact operator.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Commutator [J, a]<br />

Initially consider commutator [J, a] of J with multiplication<br />

by a ∈ C(X).<br />

We can permit discontinuities of a at vertices of cells.<br />

If a is constant on scale n cells then [J, a]h = 0 for any h<br />

that is orthogonal to n-harmonic functions<br />

Can approximate any [J, a] in operator norm by operators<br />

with finite dimensional image<br />

Hence [J, a] is a compact operator.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Summability of [J, a]<br />

Compact operator T = [J, a] is an “infinitesimal”.<br />

Let λn(T ) be the singular values (note λn → 0)<br />

Study zeta function n λ s n .<br />

Can control λn using oscillation of a on m-cells, where<br />

n ≍ e mdH (Hausdorff dimension).<br />

Crude bounds show zeta function converges for ℜ(s) > d S<br />

(spectral dimension).<br />

This region of convergence was earlier proved by<br />

Cipriani-Sauvageot via heat kernel methods<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Summability of [J, a]<br />

Compact operator T = [J, a] is an “infinitesimal”.<br />

Let λn(T ) be the singular values (note λn → 0)<br />

Study zeta function n λ s n .<br />

Can control λn using oscillation of a on m-cells, where<br />

n ≍ e mdH (Hausdorff dimension).<br />

Crude bounds show zeta function converges for ℜ(s) > d S<br />

(spectral dimension).<br />

This region of convergence was earlier proved by<br />

Cipriani-Sauvageot via heat kernel methods<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Summability of [J, a]<br />

Compact operator T = [J, a] is an “infinitesimal”.<br />

Let λn(T ) be the singular values (note λn → 0)<br />

Study zeta function n λ s n .<br />

Can control λn using oscillation of a on m-cells, where<br />

n ≍ e mdH (Hausdorff dimension).<br />

Crude bounds show zeta function converges for ℜ(s) > d S<br />

(spectral dimension).<br />

This region of convergence was earlier proved by<br />

Cipriani-Sauvageot via heat kernel methods<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Summability of [J, a]<br />

Compact operator T = [J, a] is an “infinitesimal”.<br />

Let λn(T ) be the singular values (note λn → 0)<br />

Study zeta function n λ s n .<br />

Can control λn using oscillation of a on m-cells, where<br />

n ≍ e mdH (Hausdorff dimension).<br />

Crude bounds show zeta function converges for ℜ(s) > d S<br />

(spectral dimension).<br />

This region of convergence was earlier proved by<br />

Cipriani-Sauvageot via heat kernel methods<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Summability of [J, a]<br />

Compact operator T = [J, a] is an “infinitesimal”.<br />

Let λn(T ) be the singular values (note λn → 0)<br />

Study zeta function n λ s n .<br />

Can control λn using oscillation of a on m-cells, where<br />

n ≍ e mdH (Hausdorff dimension).<br />

Crude bounds show zeta function converges for ℜ(s) > d S<br />

(spectral dimension).<br />

This region of convergence was earlier proved by<br />

Cipriani-Sauvageot via heat kernel methods<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Summability of [J, a]<br />

Compact operator T = [J, a] is an “infinitesimal”.<br />

Let λn(T ) be the singular values (note λn → 0)<br />

Study zeta function n λ s n .<br />

Can control λn using oscillation of a on m-cells, where<br />

n ≍ e mdH (Hausdorff dimension).<br />

Crude bounds show zeta function converges for ℜ(s) > d S<br />

(spectral dimension).<br />

This region of convergence was earlier proved by<br />

Cipriani-Sauvageot via heat kernel methods<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Summability below Spectral dimension<br />

For a more restricted class of fractals the zeta function for<br />

[J, a] converges below the spectral dimension if a is<br />

n-harmonic.<br />

Naotaka Kajino has pointed out that in the case of the<br />

Sierpinski Gasket the spectral dimension for the gasket in<br />

harmonic coordinates may be more natural than the<br />

spectral dimension in the usual coordinates; we have not<br />

yet checked this.<br />

Our proof is suggestive of the possibility that the “correct”<br />

condition for convergence of the zeta function at s may be<br />

that a is in a Besov space with exponent depending on s.<br />

Perhaps an analogue of a result of Connes-Sullivan for<br />

quasicircles.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Summability below Spectral dimension<br />

For a more restricted class of fractals the zeta function for<br />

[J, a] converges below the spectral dimension if a is<br />

n-harmonic.<br />

Naotaka Kajino has pointed out that in the case of the<br />

Sierpinski Gasket the spectral dimension for the gasket in<br />

harmonic coordinates may be more natural than the<br />

spectral dimension in the usual coordinates; we have not<br />

yet checked this.<br />

Our proof is suggestive of the possibility that the “correct”<br />

condition for convergence of the zeta function at s may be<br />

that a is in a Besov space with exponent depending on s.<br />

Perhaps an analogue of a result of Connes-Sullivan for<br />

quasicircles.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals


Summability below Spectral dimension<br />

For a more restricted class of fractals the zeta function for<br />

[J, a] converges below the spectral dimension if a is<br />

n-harmonic.<br />

Naotaka Kajino has pointed out that in the case of the<br />

Sierpinski Gasket the spectral dimension for the gasket in<br />

harmonic coordinates may be more natural than the<br />

spectral dimension in the usual coordinates; we have not<br />

yet checked this.<br />

Our proof is suggestive of the possibility that the “correct”<br />

condition for convergence of the zeta function at s may be<br />

that a is in a Besov space with exponent depending on s.<br />

Perhaps an analogue of a result of Connes-Sullivan for<br />

quasicircles.<br />

<strong>Luke</strong> <strong>Rogers</strong> Dirichlet forms and derivations on fractals

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!