• Aucun résultat trouvé

Continuous space-time transformations

N/A
N/A
Protected

Academic year: 2022

Partager "Continuous space-time transformations"

Copied!
43
0
0

Texte intégral

(1)

Continuous space-time transformations

Clément de Seguins Pazzis, Peter Šemrl

KU Leuven 2016

(2)

The setting

V =R4equipped with the Lorentz quadratic form q : (x,y,z,t) 7→t2x2y2z2. Space-time events a and b are coherent iff

q(ba) =0, they are adjacent iff coherent and a6=b.

(3)

Light cones

Basic light cone:

C(0) :={m∈V : q(m) =0}.

Light cone with vertex aV :

C(a) ={m∈V : q(ma) =0}=a+C(0).

Critical note: the maximal coherent sets are lines in light cones.

(4)

Light cones

Basic light cone:

C(0) :={m∈V : q(m) =0}.

Light cone with vertex aV :

C(a) ={m∈V : q(ma) =0}=a+C(0).

Critical note: the maximal coherent sets are lines in light cones.

(5)

Alexandrov’s problem

Q: What are the bijectionsφ:R4→R4that preserve coherency in both directions?

A: Only the standard maps (or Poincaré similarities):

m7→λu(m) +a

with a∈R4,λ∈R\ {0}, u :R4→R4a linear q-isometry.

Note: this generalizes to an n-dimensional Alexandrov space, i.e.

real quadratic space(V,q), with q regular (i.e. non-degenerate) of signature(1,n−1), with n≥4.

(6)

Alexandrov’s problem

Q: What are the bijectionsφ:R4→R4that preserve coherency in both directions?

A: Only the standard maps (or Poincaré similarities):

m7→λu(m) +a

with a∈R4,λ∈R\ {0}, u :R4→R4a linear q-isometry.

Note: this generalizes to an n-dimensional Alexandrov space, i.e.

real quadratic space(V,q), with q regular (i.e. non-degenerate) of signature(1,n−1), with n≥4.

(7)

Alexandrov’s problem

Q: What are the bijectionsφ:R4→R4that preserve coherency in both directions?

A: Only the standard maps (or Poincaré similarities):

m7→λu(m) +a

with a∈R4,λ∈R\ {0}, u :R4→R4a linear q-isometry.

Note: this generalizes to an n-dimensional Alexandrov space, i.e.

real quadratic space(V,q), with q regular (i.e. non-degenerate) of signature(1,n−1), with n≥4.

(8)

Matrix formulation

A model of 4-dimensional Alexandrov space:

H2=

a c c b

|(a,b)∈R2, c∈C

, with

q :M ∈ H27→det M.

A and B adjacent iff rk(AB) =1.

Standard maps:

M7→ǫPMP+A, A∈ H2, P∈GL2(C), ǫ∈ {1,−1}

M7→ǫPMTP+A, A∈ H2, P∈GL2(C), ǫ∈ {1,−1}.

(9)

Matrix formulation

A model of 4-dimensional Alexandrov space:

H2=

a c c b

|(a,b)∈R2, c∈C

, with

q :M ∈ H27→det M.

A and B adjacent iff rk(AB) =1.

Standard maps:

M7→ǫPMP+A, A∈ H2, P∈GL2(C), ǫ∈ {1,−1}

M7→ǫPMTP+A, A∈ H2, P∈GL2(C), ǫ∈ {1,−1}.

(10)

Problem

What if we relax the assumptions?

Drop bijectivity?

Assume the preservation of coherency in one direction only?

(11)

Problem

What if we relax the assumptions?

Drop bijectivity?

Assume the preservation of coherency in one direction only?

(12)

Problem

What if we relax the assumptions?

Drop bijectivity?

Assume the preservation of coherency in one direction only?

(13)

A theorem of Huang and Šemrl

Letφ:H2→ H2preserve adjacency in one direction only. Then:

eitherφis standard;

orφmapsH2into a line (degenerate adjacency preserver).

Example of a degenerate preserver:

φ:A7→

tr(A) 0

0 0

.

The trace can be replaced with an injective map . . . Note: generalization to mapsHn→ Hn.

W.-l. Huang, P. Šemrl,

Adjacency preserving maps on hermitian matrices, Canad. J. Math. 60 (2008), 1050–1066.

(14)

A theorem of Huang and Šemrl

Letφ:H2→ H2preserve adjacency in one direction only. Then:

eitherφis standard;

orφmapsH2into a line (degenerate adjacency preserver).

Example of a degenerate preserver:

φ:A7→

tr(A) 0

0 0

. The trace can be replaced with an injective map . . . Note: generalization to mapsHn→ Hn.

W.-l. Huang, P. Šemrl,

Adjacency preserving maps on hermitian matrices, Canad. J. Math. 60 (2008), 1050–1066.

(15)

A theorem of Huang and Šemrl

Letφ:H2→ H2preserve adjacency in one direction only. Then:

eitherφis standard;

orφmapsH2into a line (degenerate adjacency preserver).

Example of a degenerate preserver:

φ:A7→

tr(A) 0

0 0

. The trace can be replaced with an injective map . . . Note: generalization to mapsHn→ Hn.

W.-l. Huang, P. Šemrl,

Adjacency preserving maps on hermitian matrices, Canad. J. Math. 60 (2008), 1050–1066.

(16)

Problem: what if we only assume that coherency is preserved (in one direction only)?

Very hard problem . . .

We add continuity.

Reformulation: describe the continuous mapsφ:VV s.t.

∀(a,b)V2, q(ab) =0⇒q φ(a)−φ(b)

=0.

(17)

Problem: what if we only assume that coherency is preserved (in one direction only)?

Very hard problem . . . We add continuity.

Reformulation: describe the continuous mapsφ:VV s.t.

∀(a,b)V2, q(ab) =0⇒q φ(a)−φ(b)

=0.

(18)

Problem: what if we only assume that coherency is preserved (in one direction only)?

Very hard problem . . . We add continuity.

Reformulation: describe the continuous mapsφ:VV s.t.

∀(a,b)V2, q(ab) =0⇒q φ(a)−φ(b)

=0.

(19)

Problem: what if we only assume that coherency is preserved (in one direction only)?

Very hard problem . . . We add continuity.

Reformulation: describe the continuous mapsφ:VV s.t.

∀(a,b)V2, q(ab) =0⇒q φ(a)−φ(b)

=0.

(20)

Main result

Theorem (dSP, Šemrl (2015))

Ifφis a continuous coherency preserver then:

eitherφis standard;

orφ(V)⊂ C(a)for some aV (degenerate preserver).

C. de Seguins Pazzis, P. Šemrl,

Continuous space-time transformations, arXiv: http://arxiv.org/abs/1502.01149

(21)

Surprise: Not all degenerate continuous coherency preservers map into a line!

Choose a sequence(Un)n∈Nof open subsets of V s.t.

∀(m,n)∈N2, m6=n⇒ ∀(a,b)Un×Um, q(ab)6=0.

Choose:

aV ;

For each n∈N, choose xn ∈ C(0)withkxnk=1;

α:V →Rcontinuous with support in S

n∈N

Un; Take

φ:m7→

(a+α(m)xn if mUn

a otherwise

(22)

Surprise: Not all degenerate continuous coherency preservers map into a line!

Choose a sequence(Un)n∈Nof open subsets of V s.t.

∀(m,n)∈N2, m6=n⇒ ∀(a,b)Un×Um, q(ab)6=0.

Choose:

aV ;

For each n∈N, choose xn ∈ C(0)withkxnk=1;

α:V →Rcontinuous with support in S

n∈N

Un; Take

φ:m7→

(a+α(m)xn if mUn

a otherwise

(23)

Example of a sequence ( U

n

)

n∈N

With the standard Euclidian normk − konR4, Un:=Bo

(0R3,n),1 4

.

For(x,s)Unand(y,t)Upwith n6=p (and(s,t) ∈R2, (x,y)∈(R3)2),

(s−t)2> 1

4 >kx −yk2 whence

q (x,s)−(y,t)

>0.

(24)

Theorem (dSP, Šemrl (2015))

Any degenerate continuous coherency preserver is of the previous type.

(25)

Letφ:VV be a non-degenerate continuous coherency preserver.

Basic strategy: prove thatφpreserves adjacency;

Basic technique: analyze the action on light cones.

(26)

Letφ:VV be a non-degenerate continuous coherency preserver.

Basic strategy: prove thatφpreserves adjacency;

Basic technique: analyze the action on light cones.

(27)

Letφ:VV be a non-degenerate continuous coherency preserver.

Basic strategy: prove thatφpreserves adjacency;

Basic technique: analyze the action on light cones.

(28)

Step 1

φnever constant on a coherent line.

Consequence:

φmaps any coherent line into a unique coherent line.

(29)

Step 1

φnever constant on a coherent line.

Consequence:

φmaps any coherent line into a unique coherent line.

(30)

Step 2

An equivalence relation ofC(0)\ {0}:

ab

def ∃λ∈R : b=λa Set

Q:= (C(0)\ {0})/∼

Notes:

Qis a projective quadric, homeomorphic to the 2-sphere;

for each aV , we have a natural bijection fromQto the set of all lines inC(a).

(31)

For all aV ,

φ C(a)

⊂ C φ(a) yields

ϕa :Q → Q.

Then, one proves that

aV 7→ϕa ∈ C(Q,Q) is continuous.

(32)

Step 3

Ifϕanon constant and∃b∈ C(a)\ {a}s.t. φ(a) =φ(b), then φ−1{φ(a)}has non-empty interior.

Same conclusion ifϕanonconstant and non-injective.

Definition

State a called generic ifφ1{φ(a)}has empty interior.

Note: there are generic points.

(33)

Step 3

Ifϕanon constant and∃b∈ C(a)\ {a}s.t. φ(a) =φ(b), then φ−1{φ(a)}has non-empty interior.

Same conclusion ifϕanonconstant and non-injective.

Definition

State a called generic ifφ1{φ(a)}has empty interior.

Note: there are generic points.

(34)

Step 3

Ifϕanon constant and∃b∈ C(a)\ {a}s.t. φ(a) =φ(b), then φ−1{φ(a)}has non-empty interior.

Same conclusion ifϕanonconstant and non-injective.

Definition

State a called generic ifφ1{φ(a)}has empty interior.

Note: there are generic points.

(35)

Reformulation of Step 3

If a generic andϕanonconstant then:

ϕainjective;

For each b adjacent to a,φ(b)adjacent toφ(a).

(36)

Step 4

There exists cV generic s.t.ϕcnonconstant.

Consequences:

ϕc injective (Step 3);

ϕc a homeomorphism ofQ(invariance of domain theorem + compactness and connectedness ofQ);

ϕbnonconstant for all b∈ Q(homotopy theory of spheres).

(37)

Step 4

There exists cV generic s.t.ϕcnonconstant.

Consequences:

ϕc injective (Step 3);

ϕc a homeomorphism ofQ(invariance of domain theorem + compactness and connectedness ofQ);

ϕbnonconstant for all b∈ Q(homotopy theory of spheres).

(38)

Step 4

There exists cV generic s.t.ϕcnonconstant.

Consequences:

ϕc injective (Step 3);

ϕc a homeomorphism ofQ(invariance of domain theorem + compactness and connectedness ofQ);

ϕbnonconstant for all b∈ Q(homotopy theory of spheres).

(39)

Step 4

There exists cV generic s.t.ϕcnonconstant.

Consequences:

ϕc injective (Step 3);

ϕc a homeomorphism ofQ(invariance of domain theorem + compactness and connectedness ofQ);

ϕbnonconstant for all b∈ Q(homotopy theory of spheres).

(40)

Conclusion of the proof

For all adjacent a,b with a generic,φ(a)andφ(b)adjacent (see Step 3).

Then, every point is generic and one concludes.

(41)

Conclusion of the proof

For all adjacent a,b with a generic,φ(a)andφ(b)adjacent (see Step 3).

Then, every point is generic and one concludes.

(42)

Possible further research

Extend the result to continuous coherency preserversHn→ Hn.

Consider a more general regular real quadratic space(V,q).

(43)

Possible further research

Extend the result to continuous coherency preserversHn→ Hn.

Consider a more general regular real quadratic space(V,q).

Références

Documents relatifs

The aim of this special issue is to give a broad perspective of the state-of-the-art on the subject and to provide the community with an up-to-date account of recent advances in

This paper aims at providing the very first step towards bridging the existing gap between LPV control and identification via the introduction of a direct CT identification

For each subgroup B of A such that A/B has Q-rank 1, the number of ho- momorphisms of A/B onto Z is 2, and we choose one of the two as u B. So, using the maximality of B, the near

An illustrating corollary is the fact that the group of piece- wise projective self-transformations of the circle has no infinite subgroup with Kazhdan’s Property T; this corollary

Keywords: optimal control, reinforcement learning, policy search, sensitivity analysis, para- metric optimization, gradient estimate, likelihood ratio method, pathwise derivation..

Having surveyed the field of continuous time computation theory, we see that it provides insights into many diverse areas such as verification, control the- ory, VLSI design,

Section 3 is devoted to the proof of this result: firstly we consider an approximation via stochastic master equations driven by Pois- son processes (diffusion approximation);

It concerns a research program [BJP, BJLP1, BJLP2, BJLP3] devoted to the characterization of the absolutely continuous spectrum of a self-adjoint operator H in terms of the