tutorial on additive levy processes´ - mathdavar/ps-pdf-files/seattle06tutorial1.pdfintroduction...
TRANSCRIPT
![Page 1: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/1.jpg)
Tutorial on Additive Levy ProcessesLecture #1
Davar Khoshnevisan
Department of MathematicsUniversity of Utah
http://www.math.utah.edu/˜davar
International Conference on Stochastic Analysisand Its ApplicationsAugust 7–11, 2006
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 1 / 20
![Page 2: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/2.jpg)
Introduction
Some Terminology
An “(N ,d) random field” X has N parameters and takes values inRd (Adler, 1981);i.e., X (t) ∈ Rd for all t := (t1 , . . . , tN) ∈ RN .
Could have RN+ in place of RN , etc.
X1, . . . ,XN independent Brownian motions in Rd .“Additive Brownian motion”:
X (t) := X1(t1) + · · ·+ XN(tN) for all t = (t1 , . . . , tN) ∈ RN+.
Likewise, can have “additive stable,” “additive Levy,” etc.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 2 / 20
![Page 3: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/3.jpg)
Introduction
Some Terminology
An “(N ,d) random field” X has N parameters and takes values inRd (Adler, 1981);i.e., X (t) ∈ Rd for all t := (t1 , . . . , tN) ∈ RN .
Could have RN+ in place of RN , etc.
X1, . . . ,XN independent Brownian motions in Rd .“Additive Brownian motion”:
X (t) := X1(t1) + · · ·+ XN(tN) for all t = (t1 , . . . , tN) ∈ RN+.
Likewise, can have “additive stable,” “additive Levy,” etc.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 2 / 20
![Page 4: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/4.jpg)
Introduction
Some Terminology
An “(N ,d) random field” X has N parameters and takes values inRd (Adler, 1981);i.e., X (t) ∈ Rd for all t := (t1 , . . . , tN) ∈ RN .
Could have RN+ in place of RN , etc.
X1, . . . ,XN independent Brownian motions in Rd .“Additive Brownian motion”:
X (t) := X1(t1) + · · ·+ XN(tN) for all t = (t1 , . . . , tN) ∈ RN+.
Likewise, can have “additive stable,” “additive Levy,” etc.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 2 / 20
![Page 5: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/5.jpg)
Introduction
Some Terminology
An “(N ,d) random field” X has N parameters and takes values inRd (Adler, 1981);i.e., X (t) ∈ Rd for all t := (t1 , . . . , tN) ∈ RN .
Could have RN+ in place of RN , etc.
X1, . . . ,XN independent Brownian motions in Rd .“Additive Brownian motion”:
X (t) := X1(t1) + · · ·+ XN(tN) for all t = (t1 , . . . , tN) ∈ RN+.
Likewise, can have “additive stable,” “additive Levy,” etc.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 2 / 20
![Page 6: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/6.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
W := white noise on RN+.
I.e., Gaussian, and
If A∩B = ∅ then W (A) and W (B) are indept.
EW (A) = 0 and VarW (A) = meas(A).
Lemma
If A1,A2, . . . are nonrandom and disjoint then a.s.,
W
(∞[
n=1
An
)=
∞∑n=1
W (An),
as long as meas(An) <∞ for all n.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 3 / 20
![Page 7: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/7.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
W := white noise on RN+. I.e., Gaussian, and
If A∩B = ∅ then W (A) and W (B) are indept.
EW (A) = 0 and VarW (A) = meas(A).
Lemma
If A1,A2, . . . are nonrandom and disjoint then a.s.,
W
(∞[
n=1
An
)=
∞∑n=1
W (An),
as long as meas(An) <∞ for all n.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 3 / 20
![Page 8: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/8.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
W := white noise on RN+. I.e., Gaussian, and
If A∩B = ∅ then W (A) and W (B) are indept.
EW (A) = 0 and VarW (A) = meas(A).
Lemma
If A1,A2, . . . are nonrandom and disjoint then a.s.,
W
(∞[
n=1
An
)=
∞∑n=1
W (An),
as long as meas(An) <∞ for all n.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 3 / 20
![Page 9: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/9.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
W := white noise on RN+. I.e., Gaussian, and
If A∩B = ∅ then W (A) and W (B) are indept.
EW (A) = 0 and VarW (A) = meas(A).
Lemma
If A1,A2, . . . are nonrandom and disjoint then a.s.,
W
(∞[
n=1
An
)=
∞∑n=1
W (An),
as long as meas(An) <∞ for all n.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 3 / 20
![Page 10: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/10.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
W := white noise on RN+. I.e., Gaussian, and
If A∩B = ∅ then W (A) and W (B) are indept.
EW (A) = 0 and VarW (A) = meas(A).
Lemma
If A1,A2, . . . are nonrandom and disjoint then a.s.,
W
(∞[
n=1
An
)=
∞∑n=1
W (An),
as long as meas(An) <∞ for all n.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 3 / 20
![Page 11: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/11.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
(N ,1) “Brownian sheet” B := the dF of W ; i.e.,
B(t) := W ([0 , t1]× · · · × [0 , tN ]) for all t = (t1 , . . . , tN) ∈ RN+.
Could also index B by RN , etc.
B is the noise in many SPDEs.
In many cases, “SPDEs look like B.” [Girsanov]
(N ,d) Brownian sheet B(t) := (B1(t) , . . . ,Bd(t)), whereB1, . . . ,Bd are indept. (N ,1) Brownian sheets.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 4 / 20
![Page 12: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/12.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
(N ,1) “Brownian sheet” B := the dF of W ; i.e.,
B(t) := W ([0 , t1]× · · · × [0 , tN ]) for all t = (t1 , . . . , tN) ∈ RN+.
Could also index B by RN , etc.
B is the noise in many SPDEs.
In many cases, “SPDEs look like B.” [Girsanov]
(N ,d) Brownian sheet B(t) := (B1(t) , . . . ,Bd(t)), whereB1, . . . ,Bd are indept. (N ,1) Brownian sheets.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 4 / 20
![Page 13: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/13.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
(N ,1) “Brownian sheet” B := the dF of W ; i.e.,
B(t) := W ([0 , t1]× · · · × [0 , tN ]) for all t = (t1 , . . . , tN) ∈ RN+.
Could also index B by RN , etc.
B is the noise in many SPDEs.
In many cases, “SPDEs look like B.” [Girsanov]
(N ,d) Brownian sheet B(t) := (B1(t) , . . . ,Bd(t)), whereB1, . . . ,Bd are indept. (N ,1) Brownian sheets.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 4 / 20
![Page 14: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/14.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
(N ,1) “Brownian sheet” B := the dF of W ; i.e.,
B(t) := W ([0 , t1]× · · · × [0 , tN ]) for all t = (t1 , . . . , tN) ∈ RN+.
Could also index B by RN , etc.
B is the noise in many SPDEs.
In many cases, “SPDEs look like B.” [Girsanov]
(N ,d) Brownian sheet B(t) := (B1(t) , . . . ,Bd(t)), whereB1, . . . ,Bd are indept. (N ,1) Brownian sheets.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 4 / 20
![Page 15: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/15.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
(N ,1) “Brownian sheet” B := the dF of W ; i.e.,
B(t) := W ([0 , t1]× · · · × [0 , tN ]) for all t = (t1 , . . . , tN) ∈ RN+.
Could also index B by RN , etc.
B is the noise in many SPDEs.
In many cases, “SPDEs look like B.” [Girsanov]
(N ,d) Brownian sheet B(t) := (B1(t) , . . . ,Bd(t)), whereB1, . . . ,Bd are indept. (N ,1) Brownian sheets.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 4 / 20
![Page 16: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/16.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
(2 ,1) Brownian sheet
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 5 / 20
![Page 17: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/17.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
Let B(t1 , t2) denote 2-parameter Brownian sheet in Rd .
B(t1 + ε , t2 + δ) = W (�) + W (�) + W (�) + W (�)
t
t1 t1
δ+
ε
2
+
t2
all four independentX (ε) := W (�) = BMY (δ) := W (�) = BMZ (ε , δ) := W (�) = BS
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 6 / 20
![Page 18: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/18.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
Let B(t1 , t2) denote 2-parameter Brownian sheet in Rd .
B(t1 + ε , t2 + δ) = W (�) + W (�) + W (�) + W (�)
t
t1 t1
δ+
ε
2
+
t2
all four independentX (ε) := W (�) = BMY (δ) := W (�) = BMZ (ε , δ) := W (�) = BS
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 6 / 20
![Page 19: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/19.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
Let B(t1 , t2) denote 2-parameter Brownian sheet in Rd .
B(t1 + ε , t2 + δ) = W (�) + W (�) + W (�) + W (�)
t
t1 t1
δ+
ε
2
+
t2
all four independent
X (ε) := W (�) = BMY (δ) := W (�) = BMZ (ε , δ) := W (�) = BS
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 6 / 20
![Page 20: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/20.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
Let B(t1 , t2) denote 2-parameter Brownian sheet in Rd .
B(t1 + ε , t2 + δ) = W (�) + W (�) + W (�) + W (�)
t
t1 t1
δ+
ε
2
+
t2
all four independentX (ε) := W (�) = BM
Y (δ) := W (�) = BMZ (ε , δ) := W (�) = BS
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 6 / 20
![Page 21: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/21.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
Let B(t1 , t2) denote 2-parameter Brownian sheet in Rd .
B(t1 + ε , t2 + δ) = W (�) + W (�) + W (�) + W (�)
t
t1 t1
δ+
ε
2
+
t2
all four independentX (ε) := W (�) = BMY (δ) := W (�) = BM
Z (ε , δ) := W (�) = BS
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 6 / 20
![Page 22: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/22.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
Let B(t1 , t2) denote 2-parameter Brownian sheet in Rd .
B(t1 + ε , t2 + δ) = W (�) + W (�) + W (�) + W (�)
t
t1 t1
δ+
ε
2
+
t2
all four independentX (ε) := W (�) = BMY (δ) := W (�) = BMZ (ε , δ) := W (�) = BS
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 6 / 20
![Page 23: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/23.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
B(1 + ε ,1 + δ)−W (�)︷ ︸︸ ︷
B(1 ,1) =
W (�)+W (�)=ABM︷ ︸︸ ︷X (ε) + Y (δ) +
W (�)=BS︷ ︸︸ ︷Z (ε , δ)
= ABM + indept BS
Var (X (ε) + Y (δ)) = ε + δ
VarZ (ε , δ) = εδ � ε + δ for small ε, δ
B(1 + ε ,1 + δ)−B(1 ,1) ≈ ABM for small ε, δ.
References: Orey and Pruitt (1973), Kendall (1980), Ehm (1981),Dalang and Walsh (1992; 1993; 1996), Kh. (1995; 1999; 2003),Dalang and Mountford (1996; 1997; 2001), Kh. and Shi (1999), Kh.Xiao (2005), Kh. Xiao, and Wu (2006) . . . .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 7 / 20
![Page 24: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/24.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
B(1 + ε ,1 + δ)−W (�)︷ ︸︸ ︷
B(1 ,1) =
W (�)+W (�)=ABM︷ ︸︸ ︷X (ε) + Y (δ) +
W (�)=BS︷ ︸︸ ︷Z (ε , δ)
= ABM + indept BS
Var (X (ε) + Y (δ)) = ε + δ
VarZ (ε , δ) = εδ � ε + δ for small ε, δ
B(1 + ε ,1 + δ)−B(1 ,1) ≈ ABM for small ε, δ.
References: Orey and Pruitt (1973), Kendall (1980), Ehm (1981),Dalang and Walsh (1992; 1993; 1996), Kh. (1995; 1999; 2003),Dalang and Mountford (1996; 1997; 2001), Kh. and Shi (1999), Kh.Xiao (2005), Kh. Xiao, and Wu (2006) . . . .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 7 / 20
![Page 25: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/25.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
B(1 + ε ,1 + δ)−W (�)︷ ︸︸ ︷
B(1 ,1) =
W (�)+W (�)=ABM︷ ︸︸ ︷X (ε) + Y (δ) +
W (�)=BS︷ ︸︸ ︷Z (ε , δ)
= ABM + indept BS
Var (X (ε) + Y (δ)) = ε + δ
VarZ (ε , δ) = εδ � ε + δ for small ε, δ
B(1 + ε ,1 + δ)−B(1 ,1) ≈ ABM for small ε, δ.
References: Orey and Pruitt (1973), Kendall (1980), Ehm (1981),Dalang and Walsh (1992; 1993; 1996), Kh. (1995; 1999; 2003),Dalang and Mountford (1996; 1997; 2001), Kh. and Shi (1999), Kh.Xiao (2005), Kh. Xiao, and Wu (2006) . . . .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 7 / 20
![Page 26: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/26.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
B(1 + ε ,1 + δ)−W (�)︷ ︸︸ ︷
B(1 ,1) =
W (�)+W (�)=ABM︷ ︸︸ ︷X (ε) + Y (δ) +
W (�)=BS︷ ︸︸ ︷Z (ε , δ)
= ABM + indept BS
Var (X (ε) + Y (δ)) = ε + δ
VarZ (ε , δ) = εδ
� ε + δ for small ε, δ
B(1 + ε ,1 + δ)−B(1 ,1) ≈ ABM for small ε, δ.
References: Orey and Pruitt (1973), Kendall (1980), Ehm (1981),Dalang and Walsh (1992; 1993; 1996), Kh. (1995; 1999; 2003),Dalang and Mountford (1996; 1997; 2001), Kh. and Shi (1999), Kh.Xiao (2005), Kh. Xiao, and Wu (2006) . . . .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 7 / 20
![Page 27: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/27.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
B(1 + ε ,1 + δ)−W (�)︷ ︸︸ ︷
B(1 ,1) =
W (�)+W (�)=ABM︷ ︸︸ ︷X (ε) + Y (δ) +
W (�)=BS︷ ︸︸ ︷Z (ε , δ)
= ABM + indept BS
Var (X (ε) + Y (δ)) = ε + δ
VarZ (ε , δ) = εδ � ε + δ for small ε, δ
B(1 + ε ,1 + δ)−B(1 ,1) ≈ ABM for small ε, δ.
References: Orey and Pruitt (1973), Kendall (1980), Ehm (1981),Dalang and Walsh (1992; 1993; 1996), Kh. (1995; 1999; 2003),Dalang and Mountford (1996; 1997; 2001), Kh. and Shi (1999), Kh.Xiao (2005), Kh. Xiao, and Wu (2006) . . . .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 7 / 20
![Page 28: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/28.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
B(1 + ε ,1 + δ)−W (�)︷ ︸︸ ︷
B(1 ,1) =
W (�)+W (�)=ABM︷ ︸︸ ︷X (ε) + Y (δ) +
W (�)=BS︷ ︸︸ ︷Z (ε , δ)
= ABM + indept BS
Var (X (ε) + Y (δ)) = ε + δ
VarZ (ε , δ) = εδ � ε + δ for small ε, δ
B(1 + ε ,1 + δ)−B(1 ,1) ≈ ABM for small ε, δ.
References: Orey and Pruitt (1973), Kendall (1980), Ehm (1981),Dalang and Walsh (1992; 1993; 1996), Kh. (1995; 1999; 2003),Dalang and Mountford (1996; 1997; 2001), Kh. and Shi (1999), Kh.Xiao (2005), Kh. Xiao, and Wu (2006) . . . .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 7 / 20
![Page 29: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/29.jpg)
Local Dynamics of the Brownian Sheet
ABM and the Local Dynamics of Brownian Sheet
B(1 + ε ,1 + δ)−W (�)︷ ︸︸ ︷
B(1 ,1) =
W (�)+W (�)=ABM︷ ︸︸ ︷X (ε) + Y (δ) +
W (�)=BS︷ ︸︸ ︷Z (ε , δ)
= ABM + indept BS
Var (X (ε) + Y (δ)) = ε + δ
VarZ (ε , δ) = εδ � ε + δ for small ε, δ
B(1 + ε ,1 + δ)−B(1 ,1) ≈ ABM for small ε, δ.
References: Orey and Pruitt (1973), Kendall (1980), Ehm (1981),Dalang and Walsh (1992; 1993; 1996), Kh. (1995; 1999; 2003),Dalang and Mountford (1996; 1997; 2001), Kh. and Shi (1999), Kh.Xiao (2005), Kh. Xiao, and Wu (2006) . . . .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 7 / 20
![Page 30: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/30.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem
Let B = (2 ,1) BS; choose and fix s, t > 0.
Theorem (Kendall, 1980)
The connected component of {(u ,v) : B(u ,v) = B(s , t)} that contains(s , t) is a.s. {(s , t)}.
“A.e. point in a.e. level-set is totally disconnected from the rest.”
WLOG s = t = 1.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 8 / 20
![Page 31: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/31.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem
Let B = (2 ,1) BS; choose and fix s, t > 0.
Theorem (Kendall, 1980)
The connected component of {(u ,v) : B(u ,v) = B(s , t)} that contains(s , t) is a.s. {(s , t)}.
“A.e. point in a.e. level-set is totally disconnected from the rest.”
WLOG s = t = 1.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 8 / 20
![Page 32: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/32.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem
Let B = (2 ,1) BS; choose and fix s, t > 0.
Theorem (Kendall, 1980)
The connected component of {(u ,v) : B(u ,v) = B(s , t)} that contains(s , t) is a.s. {(s , t)}.
“A.e. point in a.e. level-set is totally disconnected from the rest.”
WLOG s = t = 1.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 8 / 20
![Page 33: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/33.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem
Let B = (2 ,1) BS; choose and fix s, t > 0.
Theorem (Kendall, 1980)
The connected component of {(u ,v) : B(u ,v) = B(s , t)} that contains(s , t) is a.s. {(s , t)}.
“A.e. point in a.e. level-set is totally disconnected from the rest.”
WLOG s = t = 1.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 8 / 20
![Page 34: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/34.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C(r) := {(x ,y) ∈ R2 : |x − 1| ∨ |y − 1| ≤ r}.
J(r) := {B(1 ,1) > sup∂C(r) B}.Goal: lim infr↓0 P(J(r)) > 0.
0–1 law (Orey and Pruitt, 1973): J(r) infinitely often r ↓ 0 a.s.
I will sketch the proof of a slightly weaker variant. [In fact it isequivalent.]
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 9 / 20
![Page 35: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/35.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C(r) := {(x ,y) ∈ R2 : |x − 1| ∨ |y − 1| ≤ r}.J(r) := {B(1 ,1) > sup∂C(r) B}.
Goal: lim infr↓0 P(J(r)) > 0.
0–1 law (Orey and Pruitt, 1973): J(r) infinitely often r ↓ 0 a.s.
I will sketch the proof of a slightly weaker variant. [In fact it isequivalent.]
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 9 / 20
![Page 36: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/36.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C(r) := {(x ,y) ∈ R2 : |x − 1| ∨ |y − 1| ≤ r}.J(r) := {B(1 ,1) > sup∂C(r) B}.Goal: lim infr↓0 P(J(r)) > 0.
0–1 law (Orey and Pruitt, 1973): J(r) infinitely often r ↓ 0 a.s.
I will sketch the proof of a slightly weaker variant. [In fact it isequivalent.]
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 9 / 20
![Page 37: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/37.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C(r) := {(x ,y) ∈ R2 : |x − 1| ∨ |y − 1| ≤ r}.J(r) := {B(1 ,1) > sup∂C(r) B}.Goal: lim infr↓0 P(J(r)) > 0.
0–1 law (Orey and Pruitt, 1973): J(r) infinitely often r ↓ 0 a.s.
I will sketch the proof of a slightly weaker variant. [In fact it isequivalent.]
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 9 / 20
![Page 38: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/38.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C(r) := {(x ,y) ∈ R2 : |x − 1| ∨ |y − 1| ≤ r}.J(r) := {B(1 ,1) > sup∂C(r) B}.Goal: lim infr↓0 P(J(r)) > 0.
0–1 law (Orey and Pruitt, 1973): J(r) infinitely often r ↓ 0 a.s.
I will sketch the proof of a slightly weaker variant. [In fact it isequivalent.]
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 9 / 20
![Page 39: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/39.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C ′(r) := ({1 + r}× [1 , r ])∪ ([1 , r ]×{1 + r}).
(1,1)
J ′(r) := {B(1 ,1) > supC ′(r) B}.Goal: limr↓0 P(J ′(r)) > 0.
Local dynamics + scaling ⇒
P(J ′(r))→ P
{X (1) + Y (1) > sup
(u,v)∈C ′(1)
(X (u) + Y (v))
}> 0.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 10 / 20
![Page 40: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/40.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C ′(r) := ({1 + r}× [1 , r ])∪ ([1 , r ]×{1 + r}).
(1,1)
J ′(r) := {B(1 ,1) > supC ′(r) B}.Goal: limr↓0 P(J ′(r)) > 0.
Local dynamics + scaling ⇒
P(J ′(r))→ P
{X (1) + Y (1) > sup
(u,v)∈C ′(1)
(X (u) + Y (v))
}> 0.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 10 / 20
![Page 41: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/41.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C ′(r) := ({1 + r}× [1 , r ])∪ ([1 , r ]×{1 + r}).
(1,1)
J ′(r) := {B(1 ,1) > supC ′(r) B}.Goal: limr↓0 P(J ′(r)) > 0.
Local dynamics + scaling ⇒
P(J ′(r))→ P
{X (1) + Y (1) > sup
(u,v)∈C ′(1)
(X (u) + Y (v))
}> 0.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 10 / 20
![Page 42: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/42.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C ′(r) := ({1 + r}× [1 , r ])∪ ([1 , r ]×{1 + r}).
(1,1)
J ′(r) := {B(1 ,1) > supC ′(r) B}.
Goal: limr↓0 P(J ′(r)) > 0.
Local dynamics + scaling ⇒
P(J ′(r))→ P
{X (1) + Y (1) > sup
(u,v)∈C ′(1)
(X (u) + Y (v))
}> 0.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 10 / 20
![Page 43: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/43.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C ′(r) := ({1 + r}× [1 , r ])∪ ([1 , r ]×{1 + r}).
(1,1)
J ′(r) := {B(1 ,1) > supC ′(r) B}.Goal: limr↓0 P(J ′(r)) > 0.
Local dynamics + scaling ⇒
P(J ′(r))→ P
{X (1) + Y (1) > sup
(u,v)∈C ′(1)
(X (u) + Y (v))
}> 0.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 10 / 20
![Page 44: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/44.jpg)
A Theorem of W. Kendall
First Application: Contours of Brownian SheetKendall’s Theorem, Continued
C ′(r) := ({1 + r}× [1 , r ])∪ ([1 , r ]×{1 + r}).
(1,1)
J ′(r) := {B(1 ,1) > supC ′(r) B}.Goal: limr↓0 P(J ′(r)) > 0.
Local dynamics + scaling ⇒
P(J ′(r))→ P
{X (1) + Y (1) > sup
(u,v)∈C ′(1)
(X (u) + Y (v))
}> 0.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 10 / 20
![Page 45: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/45.jpg)
A Theorem of W. Kendall
The Zero-Set of B
Kendall’s theorem holds for “most” points in the zero-set too.Can you see it?
(Kh., Revesz, and Shi, 2005)
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 11 / 20
![Page 46: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/46.jpg)
A Theorem of W. Kendall
The Zero-Set of B
Kendall’s theorem holds for “most” points in the zero-set too.Can you see it?
(Kh., Revesz, and Shi, 2005)
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 11 / 20
![Page 47: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/47.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem
X and Y = independent BMs in Rd . Then:“Two BM paths can cross only in dim ≤ 3.”
Theorem (Dvoretzky, Erdos, and Kakutani, 1950)
TFAE:
1 X ((0 ,∞))∩Y ((0 ,∞)) 6= ∅;2 d ≤ 3.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 12 / 20
![Page 48: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/48.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem
X and Y = independent BMs in Rd . Then:“Two BM paths can cross only in dim ≤ 3.”
Theorem (Dvoretzky, Erdos, and Kakutani, 1950)
TFAE:
1 X ((0 ,∞))∩Y ((0 ,∞)) 6= ∅;2 d ≤ 3.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 12 / 20
![Page 49: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/49.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem
X and Y = independent BMs in Rd . Then:“Two BM paths can cross only in dim ≤ 3.”
Theorem (Dvoretzky, Erdos, and Kakutani, 1950)
TFAE:1 X ((0 ,∞))∩Y ((0 ,∞)) 6= ∅;
2 d ≤ 3.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 12 / 20
![Page 50: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/50.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem
X and Y = independent BMs in Rd . Then:“Two BM paths can cross only in dim ≤ 3.”
Theorem (Dvoretzky, Erdos, and Kakutani, 1950)
TFAE:1 X ((0 ,∞))∩Y ((0 ,∞)) 6= ∅;2 d ≤ 3.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 12 / 20
![Page 51: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/51.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Equivalently, TFAE:
X ([1 ,2])∩Y ([1 ,2]) 6= ∅ ⇔ d ≤ 3. (1)
There are now several proofs; most involve potential theory and/orPDEs. I will describe a relatively recent, very elementary, proof that isbased on ABM (Kh., 2003).Key idea: X ([1 ,2])∩Y ([1 ,2]) 6= ∅⇔
A(s ,t)=ABM︷ ︸︸ ︷X (s)−Y (t) = 0 for some (s , t) ∈ [1 ,2]2,
⇔0 ∈ A
([1 ,2]2
).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 13 / 20
![Page 52: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/52.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Equivalently, TFAE:
X ([1 ,2])∩Y ([1 ,2]) 6= ∅ ⇔ d ≤ 3. (1)
There are now several proofs; most involve potential theory and/orPDEs. I will describe a relatively recent, very elementary, proof that isbased on ABM (Kh., 2003).
Key idea: X ([1 ,2])∩Y ([1 ,2]) 6= ∅⇔
A(s ,t)=ABM︷ ︸︸ ︷X (s)−Y (t) = 0 for some (s , t) ∈ [1 ,2]2,
⇔0 ∈ A
([1 ,2]2
).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 13 / 20
![Page 53: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/53.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Equivalently, TFAE:
X ([1 ,2])∩Y ([1 ,2]) 6= ∅ ⇔ d ≤ 3. (1)
There are now several proofs; most involve potential theory and/orPDEs. I will describe a relatively recent, very elementary, proof that isbased on ABM (Kh., 2003).Key idea: X ([1 ,2])∩Y ([1 ,2]) 6= ∅⇔
A(s ,t)=ABM︷ ︸︸ ︷X (s)−Y (t) = 0 for some (s , t) ∈ [1 ,2]2,
⇔0 ∈ A
([1 ,2]2
).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 13 / 20
![Page 54: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/54.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Equivalently, TFAE:
X ([1 ,2])∩Y ([1 ,2]) 6= ∅ ⇔ d ≤ 3. (1)
There are now several proofs; most involve potential theory and/orPDEs. I will describe a relatively recent, very elementary, proof that isbased on ABM (Kh., 2003).Key idea: X ([1 ,2])∩Y ([1 ,2]) 6= ∅⇔
A(s ,t)=ABM︷ ︸︸ ︷X (s)−Y (t) = 0 for some (s , t) ∈ [1 ,2]2,
⇔0 ∈ A
([1 ,2]2
).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 13 / 20
![Page 55: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/55.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
When d ≤ 3 one can directly construct a “local time.” This is ana.s.-nontrivial random measure on the setA−1{0} = {(s , t) ∈ [1 ,2]2 : X (s) = Y (t)}. Therefore,0 ∈ A([1 ,2]2) a.s.
We prove that if d ≥ 5 then A(s , t) = X (s)−Y (t) 6= 0 for alls, t ∈ [1 ,2]. This proof can be pushed through when d = 4 butneeds more care; see the original paper (Kh., Expos. Math.,2003).
Exercise
Prove that A([1 ,2]2) = A(1 ,1) + A([0 ,1]2), where A is a copy of A,independent of A(1 ,1), and a + S = {a + s : s ∈ S} for all a ∈ Rd andS ⊂ Rd .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 14 / 20
![Page 56: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/56.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
When d ≤ 3 one can directly construct a “local time.” This is ana.s.-nontrivial random measure on the setA−1{0} = {(s , t) ∈ [1 ,2]2 : X (s) = Y (t)}. Therefore,0 ∈ A([1 ,2]2) a.s.
We prove that if d ≥ 5 then A(s , t) = X (s)−Y (t) 6= 0 for alls, t ∈ [1 ,2]. This proof can be pushed through when d = 4 butneeds more care; see the original paper (Kh., Expos. Math.,2003).
Exercise
Prove that A([1 ,2]2) = A(1 ,1) + A([0 ,1]2), where A is a copy of A,independent of A(1 ,1), and a + S = {a + s : s ∈ S} for all a ∈ Rd andS ⊂ Rd .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 14 / 20
![Page 57: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/57.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
When d ≤ 3 one can directly construct a “local time.” This is ana.s.-nontrivial random measure on the setA−1{0} = {(s , t) ∈ [1 ,2]2 : X (s) = Y (t)}. Therefore,0 ∈ A([1 ,2]2) a.s.
We prove that if d ≥ 5 then A(s , t) = X (s)−Y (t) 6= 0 for alls, t ∈ [1 ,2]. This proof can be pushed through when d = 4 butneeds more care; see the original paper (Kh., Expos. Math.,2003).
Exercise
Prove that A([1 ,2]2) = A(1 ,1) + A([0 ,1]2), where A is a copy of A,independent of A(1 ,1), and a + S = {a + s : s ∈ S} for all a ∈ Rd andS ⊂ Rd .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 14 / 20
![Page 58: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/58.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
When d ≤ 3 one can directly construct a “local time.” This is ana.s.-nontrivial random measure on the setA−1{0} = {(s , t) ∈ [1 ,2]2 : X (s) = Y (t)}. Therefore,0 ∈ A([1 ,2]2) a.s.
We prove that if d ≥ 5 then A(s , t) = X (s)−Y (t) 6= 0 for alls, t ∈ [1 ,2]. This proof can be pushed through when d = 4 butneeds more care; see the original paper (Kh., Expos. Math.,2003).
Exercise
Prove that A([1 ,2]2) = A(1 ,1) + A([0 ,1]2), where A is a copy of A,independent of A(1 ,1), and a + S = {a + s : s ∈ S} for all a ∈ Rd andS ⊂ Rd .
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 14 / 20
![Page 59: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/59.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .By the Exercise,
P{
0 ∈ A([1 ,2]2
)}
=Z
RdP{
x ∈ A([0 ,1]2
)}P {A(1 ,1) ∈ −dx}︸ ︷︷ ︸
=ϕ(x)dx , ϕ>0
�Z
RdP{
x ∈ A([0 ,1]2
)}dx
= E{
m(
A([0 ,1]2
))}.
(Levy, 1940; Kahane, 1983) “a � b” means “a > 0⇔ b > 0.”
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 15 / 20
![Page 60: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/60.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .By the Exercise,
P{
0 ∈ A([1 ,2]2
)}=
ZRd
P{
x ∈ A([0 ,1]2
)}P {A(1 ,1) ∈ −dx}︸ ︷︷ ︸
=ϕ(x)dx , ϕ>0
�Z
RdP{
x ∈ A([0 ,1]2
)}dx
= E{
m(
A([0 ,1]2
))}.
(Levy, 1940; Kahane, 1983) “a � b” means “a > 0⇔ b > 0.”
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 15 / 20
![Page 61: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/61.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .By the Exercise,
P{
0 ∈ A([1 ,2]2
)}=
ZRd
P{
x ∈ A([0 ,1]2
)}P {A(1 ,1) ∈ −dx}︸ ︷︷ ︸
=ϕ(x)dx , ϕ>0
�Z
RdP{
x ∈ A([0 ,1]2
)}dx
= E{
m(
A([0 ,1]2
))}.
(Levy, 1940; Kahane, 1983) “a � b” means “a > 0⇔ b > 0.”
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 15 / 20
![Page 62: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/62.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .By the Exercise,
P{
0 ∈ A([1 ,2]2
)}=
ZRd
P{
x ∈ A([0 ,1]2
)}P {A(1 ,1) ∈ −dx}︸ ︷︷ ︸
=ϕ(x)dx , ϕ>0
�Z
RdP{
x ∈ A([0 ,1]2
)}dx
= E{
m(
A([0 ,1]2
))}.
(Levy, 1940; Kahane, 1983) “a � b” means “a > 0⇔ b > 0.”
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 15 / 20
![Page 63: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/63.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .By the Exercise,
P{
0 ∈ A([1 ,2]2
)}=
ZRd
P{
x ∈ A([0 ,1]2
)}P {A(1 ,1) ∈ −dx}︸ ︷︷ ︸
=ϕ(x)dx , ϕ>0
�Z
RdP{
x ∈ A([0 ,1]2
)}dx
= E{
m(
A([0 ,1]2
))}.
(Levy, 1940; Kahane, 1983) “a � b” means “a > 0⇔ b > 0.”
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 15 / 20
![Page 64: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/64.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .
Goal: Em(A([0,1]2)) = 0.
By Brownian scaling,
A([0 ,2]2)︸ ︷︷ ︸∼X(2s)−Y (2t)
D=√
2A([0 ,1]2)
⇔ m(A([0 ,2]2)D= 2d/2m(A[0 ,1]2)
⇔ Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 16 / 20
![Page 65: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/65.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .Goal: Em(A([0,1]2)) = 0.
By Brownian scaling,
A([0 ,2]2)︸ ︷︷ ︸∼X(2s)−Y (2t)
D=√
2A([0 ,1]2)
⇔ m(A([0 ,2]2)D= 2d/2m(A[0 ,1]2)
⇔ Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 16 / 20
![Page 66: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/66.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .Goal: Em(A([0,1]2)) = 0.
By Brownian scaling,
A([0 ,2]2)︸ ︷︷ ︸
∼X(2s)−Y (2t)
D=√
2A([0 ,1]2)
⇔ m(A([0 ,2]2)D= 2d/2m(A[0 ,1]2)
⇔ Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 16 / 20
![Page 67: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/67.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .Goal: Em(A([0,1]2)) = 0.
By Brownian scaling,
A([0 ,2]2)︸ ︷︷ ︸∼X(2s)−Y (2t)
D=√
2A([0 ,1]2)
⇔ m(A([0 ,2]2)D= 2d/2m(A[0 ,1]2)
⇔ Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 16 / 20
![Page 68: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/68.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .Goal: Em(A([0,1]2)) = 0.
By Brownian scaling,
A([0 ,2]2)︸ ︷︷ ︸∼X(2s)−Y (2t)
D=√
2A([0 ,1]2)
⇔ m(A([0 ,2]2)D= 2d/2m(A[0 ,1]2)
⇔ Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 16 / 20
![Page 69: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/69.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .Goal: Em(A([0,1]2)) = 0.
By Brownian scaling,
A([0 ,2]2)︸ ︷︷ ︸∼X(2s)−Y (2t)
D=√
2A([0 ,1]2)
⇔ m(A([0 ,2]2)D= 2d/2m(A[0 ,1]2)
⇔ Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 16 / 20
![Page 70: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/70.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .Goal: Em(A([0,1]2)) = 0.
By Brownian scaling,
A([0 ,2]2)︸ ︷︷ ︸∼X(2s)−Y (2t)
D=√
2A([0 ,1]2)
⇔ m(A([0 ,2]2)D= 2d/2m(A[0 ,1]2)
⇔ Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 16 / 20
![Page 71: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/71.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .
We know: Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).We want: Em(A([0 ,1]2) = 0.Note that
A([0 ,2]2
)= A
([0 ,1]2
)∪A ([0 ,1]× [1 ,2])
∪A ([1 ,2]× [0 ,1])∪A([1 ,2]2
).
Therefore,
m(
A([0 ,2]2
))≤ m
(A([0 ,1]2
))+ m (A ([0 ,1]× [1 ,2]))
+m (A ([1 ,2]× [0 ,1])) + m(
A([1 ,2]2
)).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 17 / 20
![Page 72: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/72.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .We know: Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).
We want: Em(A([0 ,1]2) = 0.Note that
A([0 ,2]2
)= A
([0 ,1]2
)∪A ([0 ,1]× [1 ,2])
∪A ([1 ,2]× [0 ,1])∪A([1 ,2]2
).
Therefore,
m(
A([0 ,2]2
))≤ m
(A([0 ,1]2
))+ m (A ([0 ,1]× [1 ,2]))
+m (A ([1 ,2]× [0 ,1])) + m(
A([1 ,2]2
)).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 17 / 20
![Page 73: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/73.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .We know: Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).We want: Em(A([0 ,1]2) = 0.
Note that
A([0 ,2]2
)= A
([0 ,1]2
)∪A ([0 ,1]× [1 ,2])
∪A ([1 ,2]× [0 ,1])∪A([1 ,2]2
).
Therefore,
m(
A([0 ,2]2
))≤ m
(A([0 ,1]2
))+ m (A ([0 ,1]× [1 ,2]))
+m (A ([1 ,2]× [0 ,1])) + m(
A([1 ,2]2
)).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 17 / 20
![Page 74: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/74.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .We know: Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).We want: Em(A([0 ,1]2) = 0.Note that
A([0 ,2]2
)= A
([0 ,1]2
)∪A ([0 ,1]× [1 ,2])
∪A ([1 ,2]× [0 ,1])∪A([1 ,2]2
).
Therefore,
m(
A([0 ,2]2
))≤ m
(A([0 ,1]2
))+ m (A ([0 ,1]× [1 ,2]))
+m (A ([1 ,2]× [0 ,1])) + m(
A([1 ,2]2
)).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 17 / 20
![Page 75: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/75.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Let A(s , t) := X (s)−Y (t), d ≥ 5; m := Leb. meas. on Rd .We know: Em(A([0 ,2]2) = 2d/2Em(A([0 ,1]2).We want: Em(A([0 ,1]2) = 0.Note that
A([0 ,2]2
)= A
([0 ,1]2
)∪A ([0 ,1]× [1 ,2])
∪A ([1 ,2]× [0 ,1])∪A([1 ,2]2
).
Therefore,
m(
A([0 ,2]2
))≤ m
(A([0 ,1]2
))+ m (A ([0 ,1]× [1 ,2]))
+m (A ([1 ,2]× [0 ,1])) + m(
A([1 ,2]2
)).
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 17 / 20
![Page 76: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/76.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
We know:
m(
A([0 ,2]2
))≤ m
(A([0 ,1]2
))+ m (A ([0 ,1]× [1 ,2]))
+m (A ([1 ,2]× [0 ,1])) + m(
A([1 ,2]2
)).
By the Exercise,
⇒ m(
A([1 ,2]2
))D= m
(A([0 ,1]2
)).
Same with the other terms in the first display.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 18 / 20
![Page 77: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/77.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
We know:
m(
A([0 ,2]2
))≤ m
(A([0 ,1]2
))+ m (A ([0 ,1]× [1 ,2]))
+m (A ([1 ,2]× [0 ,1])) + m(
A([1 ,2]2
)).
By the Exercise,
⇒ m(
A([1 ,2]2
))D= m
(A([0 ,1]2
)).
Same with the other terms in the first display.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 18 / 20
![Page 78: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/78.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Thus,Em
(A([0 ,2]2
))≤ 4Em
(A([0 ,1]2
))
⇒2d/2Em
(A([0 ,1]2
))≤ 4Em
(A([0 ,1]2
)).
d ≥ 5⇒ Em(A([0 ,1]2)) = 0, as desired.
Exercise
Prove that Em(A([0 ,1]2) <∞.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 19 / 20
![Page 79: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/79.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Thus,Em
(A([0 ,2]2
))≤ 4Em
(A([0 ,1]2
))⇒
2d/2Em(
A([0 ,1]2
))
≤ 4Em(
A([0 ,1]2
)).
d ≥ 5⇒ Em(A([0 ,1]2)) = 0, as desired.
Exercise
Prove that Em(A([0 ,1]2) <∞.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 19 / 20
![Page 80: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/80.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Thus,Em
(A([0 ,2]2
))≤ 4Em
(A([0 ,1]2
))⇒
2d/2Em(
A([0 ,1]2
))≤ 4Em
(A([0 ,1]2
)).
d ≥ 5⇒ Em(A([0 ,1]2)) = 0, as desired.
Exercise
Prove that Em(A([0 ,1]2) <∞.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 19 / 20
![Page 81: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/81.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Thus,Em
(A([0 ,2]2
))≤ 4Em
(A([0 ,1]2
))⇒
2d/2Em(
A([0 ,1]2
))≤ 4Em
(A([0 ,1]2
)).
d ≥ 5⇒ Em(A([0 ,1]2)) = 0, as desired.
Exercise
Prove that Em(A([0 ,1]2) <∞.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 19 / 20
![Page 82: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/82.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
Second Application: Intersections of Brownian MotionsThe Dvoretzky–Erdos–Kakutani Theorem, Continued
Thus,Em
(A([0 ,2]2
))≤ 4Em
(A([0 ,1]2
))⇒
2d/2Em(
A([0 ,1]2
))≤ 4Em
(A([0 ,1]2
)).
d ≥ 5⇒ Em(A([0 ,1]2)) = 0, as desired.
Exercise
Prove that Em(A([0 ,1]2) <∞.
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 19 / 20
![Page 83: Tutorial on Additive Levy Processes´ - Mathdavar/ps-pdf-files/Seattle06Tutorial1.pdfIntroduction Some Terminology An “(N ,d) random field” X has N parameters and takes values](https://reader034.vdocuments.site/reader034/viewer/2022042709/5f52167fdb0b1605760770b3/html5/thumbnails/83.jpg)
A Theorem of A. Dvoretzky, P. Erdos, and S. Kakutani
In Memory of Ron Pyke (1931–2005)
D. Khoshnevisan (Salt Lake City, Utah) ICSAA, Seattle ’06 20 / 20