A point process associated with local maxima of Brownian motion (Q707601): Difference between revisions
From MaRDI portal
Created a new Item |
Added link to MaRDI item. |
||
links / mardi / name | links / mardi / name | ||
Revision as of 10:57, 30 January 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | A point process associated with local maxima of Brownian motion |
scientific article |
Statements
A point process associated with local maxima of Brownian motion (English)
0 references
8 October 2010
0 references
Let \(B_t\), \(t\in R\), be standard Brownian motion. The author studies the set \(M\) of times \(t\) where \(B_t\) is a local maximum. Put \(U_t= \text{inf}\{s> 0 : B_{t-s}> B_t\}\), \(V_t= \text{inf}\{s> 0: B_{t+s}> B_t\}\). Thus \(t\in M\) iff \(U_t> 0\), \(V_t> 0\). Let \(M_{ab}= \{t\in R: B_t= \max\{B_y: t-a\leq y\leq t+ b\}= (t\in R: U_t> a, V_t> b\}\), \(a> 0\), \(b> 0\). The set \(M_{ab}\) is shown to consist a.s. of isolated points, separated by at least \(\min(a, b)\). Put \(\rho(s, t)=\) smallest \(\tau\in [s,t]\) with \(B_\tau= \max\{B_u: s\leq u\leq t\}\) and \(G(s,t)= B(\rho(s,t))- B_s\), \(D(s,t)= B(\rho(s,t)- B_t\), \(N_{ab}= \text{card\,} A\cap M_{ab}\). The joint distribution of \(\rho(s, t)\), \(G(s,t)\) and \(D(s,t)\) is characterized in terms of three independent r.v.'s uniform on \([0,\pi/2]\) and exponential. The correlation function \(EN_{ab} (dt_1)\cdots N_{ab}(dt_n)\) is derived in Theorem 5. Let \(T_i\), \(i\in Z\), denote the successive points of \(M_{ab}\) with \(T_0\leq 0< T_1\). The \(T_i- T_{i-1}\) are shown to be i.i.d. and the densities of \(T_i- T_{i-1}\) and \((T_{i-1},T_i)\) are derived ( Theorem 8 ). A second proof of this theorem is found by studying \(R_a= \{t\in R_+: B_t= \max\{B_s: [t- a]_+\leq s\leq t\}\}= \{t\in R_+: A^a_t= B_t\}\), where \(A^a_t\) is \(\max\{B_s: [t-a]_+\leq s\leq t]\). This proof applies the equivalence of \(t\in M_{ab}\) and \(R_a\cap [t,t+ b]= \{t\}\). Then it is shown that the \(T_i- T_{i-1}\) are i.i.d. with distribution expressed in terms of a subordinator \(\sigma_a\). This subordinator is the right-continuous inverse of the process \(\int^t I\{A^a_s- B_s\}\,dA^a_s\). The subordinator and its Lévy measure are studied in section 7.
0 references
Brownian motion
0 references
local maxima
0 references
point process
0 references
stable distribution
0 references