Covering the whole space with Poisson random balls

icon

19

pages

icon

English

icon

Documents

Écrit par

Publié par

Lire un extrait
Lire un extrait

Obtenez un accès à la bibliothèque pour le consulter en ligne En savoir plus

Découvre YouScribe en t'inscrivant gratuitement

Je m'inscris

Découvre YouScribe en t'inscrivant gratuitement

Je m'inscris
icon

19

pages

icon

English

icon

Documents

Lire un extrait
Lire un extrait

Obtenez un accès à la bibliothèque pour le consulter en ligne En savoir plus

1COVERING Rd WITH POISSON RANDOM BALLS Anne Estrade (MAP5 and ANR-09-BLAN-0029 mataim) joint work with Hermine Biermé (MAP5) Stoch. Geom. Lille - April 2011

  • lebesgue measure

  • ?vd ∫

  • poisson property

  • covering rd

  • still valid

  • balls

  • boolean model

  • open random


Voir icon arrow

Publié par

Nombre de lectures

21

Langue

English

d
R
ch.VERINGorkCOlet1Hermine(MAP5)WITHom.POISSONAprilRANDOMwBALLSwithAnneBierm?EstradeSto(MAP5GeandLilANR-09-BLAN-0029-mataim)2011joind
Φ R ×(0,+∞)
dxμ(dr)
d
dx R
μ(dr) σ (0,+∞)
d
R
Σ = ∪ B(x,r)
(x,r)∈Φ
d
Σ = R a.s.?
d
R \Σ
-niteproccurscesstheinbigConsidereringthecooptheenerandomasubsetSetting:ofsawitherageiniftensithomeasureConsiderynon-negativPvoisson:onproblemmeasureifAcase,coyonvmeasureo2Question2::notpcase,oinwtisQuestiona1:esgue?Lebd
• P(Σ = R ) = 0 1
d
• P(B(0,1)⊂ Σ)> 0 ⇒ P(Σ = R ) = 1
• P(0 ∈ Σ) =P({(x,r)∈ Φ ; 0∈B(x,r)} =∅)

R
+∞
d
= exp −v r μ(dr)
d
0
R
+∞
d d
Σ = R a.s. ⇒ r μ(dr) = +∞
0
R
+∞
d d
r μ(dr) = +∞ ⇒ Leb R \Σ = 0 a.s.
0
3P12dicit:propfactsBasicorto(duey)toACTS(due11y)BASICergoy):FCsqstationariterttooisson(dueCsqμ
supp(μ)⊂ [a,+∞) a> 0
Z
+∞
d d
Σ = R a.s. ⇔ r μ(dr) = +∞
0
forballs)caseniteccurslarge(germ-grainForCTSBowhereolean1radiusBASICRmksomecoAerage4ermo:casevwherenevisowithofballsdel)equalor(onlyΣ = R a.s.


Z Z
1 +∞
exp 2 (r−u)μ(dr) du = +∞
0 u
⇒ d> 1
⇐ d> 1
ymoreone-dimensionalstoppingcaseF(Mandelbrotnot'72,seeSheppbased:argumenstillalidvadapted,alidlater)for:'72)on1timeCTSt,(needsvBASICan5forAetobμ (dr) =1 (r)μ(dr) Σ = ∪ B(x,r)
(0,1]
H H
(x,r)∈Φ;r≤1
μ (dr) =1 (r)μ(dr) Σ = ∪ B(x,r)
(1,+∞)
L L
(x,r)∈Φ;r>1
Σ = Σ ∪Σ
L H
d d d
Σ =R a.s. ⇔ Σ =R a.s. Σ =R a.s.
L H
(disjoinandorandwtGEsets)coTheoremand:CO:erageNotationsvthenfrequency2loHIGHHight26ANDandLOVERAWFREQUENCYindependenδ
Σ = ∪ B(x,r) Σ = ∪ B(x,δr)
(>n)
L
(x,r)∈Φ;r>n (x,r)∈Φ;r>1
Z
+∞

d d
P Σ =R = 1 ⇔ r μ(dr) = +∞
L
1
Z
+∞
d
⇔ ∀n≥ 1 , r μ(dr) = +∞
n

d
⇔ P ∩ Σ =R = 1
n≥1
(>n)

d δ d
P Σ =R = 1 ⇔ ∀δ> 0 , P Σ =R = 1
L
L
d
R
innitecothenfrequencyvwpLetLoer7oGEanVERAtCOeredFREQUENCYumWcoHIGHerageerageccurs,LOa.s.ANDyarbitrarilyoinballs.ofvisandcovalsobyanSo,nifbloofwlarge2andd d d
Σ =R a.s. ⇔ Σ =R a.s. Σ =R a.s.
L H

d
• Σ =R a.s.⇒ B(0,1)⊂ Σ a.s.
d
• P(Σ =R ) = 0 ⇒ P(B(0,1)∩Σ =∅)> 0
L L
p := P(B(0,1)∩Σ =∅)
L
= P(B(0,1)∩Σ =∅ B(0,1)⊂ Σ)
L
= P(B(0,1)∩Σ =∅ B(0,1)⊂ Σ )
L H
= P(B(0,1)∩Σ =∅) × P(B(0,1)⊂ Σ )
L H
= p × P(B(0,1)⊂ Σ )
H
d
P(B(0,1)⊂ Σ ) = 1 P(Σ =R ) = 1
H H
andof:ANDThen2orWcoe:andTvanderageCOvProcoHFvproerageHIGHhenceLObutsonotFREQUENCYoVERALFGEco8verageof
Z
1
d d
limsupu exp v (r−u) μ(dr) = +∞
d
u→0
u

d
Σ = R a.s.
H


Z Z
1 1
d−1 d−1
u exp v r (r−u)μ(dr) du = +∞
d
0 u
d = 1
atcoShepp'svtheerage(conTheoremcondition:asHighGEFREQUENCYwithHIGHoundary2isVERAsameW-9(frequency-AND)LOKahane'sCOvexsetsofa:bTheinsteadnecessaryballs)Remarkd
[0,1]
d
P([0,1] ⊂ Σ )> 0
H
ε> 0
Z

d c
m =Leb [0,1] ∩Σ = 1 dy
ε y∈/Σ
H,≥ε
H,≥ε
d
[0,1]
R
1
−κ d
ε
E(m ) =e κ =v r μ(dr)
ε ε d
ε
2 −2κ
ε
E(m )≤...≤Ie I ∈ (0,+∞)
ε
2
E(m )
ε
−1
P(m > 0)≥ ≥I
ε
2
E(m )
ε
d −1 d
P([0,1] ⊂ Σ )≥I P([0,1] ⊂ Σ )> 0
H,≥ε H
10hofy-SctoHIGHsu.hvVERAcondition:COwithFREQUENCY2WthatCaucandByintruetnotimpliescond.conditionnecessaryartz,ifGESktheLOewithproHence,soetcorderletdiscretize,hwSucienofproNecessaryforThencondition:AND

Voir icon more
Alternate Text