Wiese Kay Jörg, Majumdar Satya N, Rosso Alberto
CNRS-Laboratoire de Physique Théorique de l'Ecole Normale Supérieure, Paris, France.
Phys Rev E Stat Nonlin Soft Matter Phys. 2011 Jun;83(6 Pt 1):061141. doi: 10.1103/PhysRevE.83.061141. Epub 2011 Jun 24.
Fractional Brownian motion is a Gaussian process x(t) with zero mean and two-time correlations (x(t(1))x(t(2)))=D(t(1)(2H)+t(2)(2H)-|t(1)-t(2)|(2H)), where H, with 0<H<1, is called the Hurst exponent. For H=1/2, x(t) is a Brownian motion, while for H≠1/2, x(t) is a non-Markovian process. Here we study x(t) in presence of an absorbing boundary at the origin and focus on the probability density P(+)(x,t) for the process to arrive at x at time t, starting near the origin at time 0, given that it has never crossed the origin. It has a scaling form P(+)(x,t)t(-H)R(+)(x/t(H)). Our objective is to compute the scaling function R(+)(y), which up to now was only known for the Markov case H=1/2. We develop a systematic perturbation theory around this limit, setting H=1/2+ε, to calculate the scaling function R(+)(y) to first order in ε. We find that R(+)(y) behaves as R(+)(y)y(ϕ) as y→0 (near the absorbing boundary), while R(+)(y)~y(γ)exp(-y(2)/2) as y→∞, with ϕ=1-4ε+O(ε(2)) and γ=1-2ε+O(ε(2)). Our ε-expansion result confirms the scaling relation ϕ=(1-H)/H proposed in Zoia, Rosso, and Majumdar [Phys. Rev. Lett. 102, 120602 (2009)]. We verify our findings via numerical simulations for H=2/3. The tools developed here are versatile, powerful, and adaptable to different situations.
分数布朗运动是一个均值为零的高斯过程(x(t)),其二阶关联函数为((x(t_1)x(t_2)) = D(t_1^{2H} + t_2^{2H} - |t_1 - t_2|^{2H})),其中(H)满足(0 < H < 1),被称为赫斯特指数。当(H = 1/2)时,(x(t))是布朗运动;而当(H \neq 1/2)时,(x(t))是一个非马尔可夫过程。在此,我们研究在原点处存在吸收边界情况下的(x(t)),并关注该过程在时间(t)到达(x)处的概率密度(P_+(x,t)),即给定其在时间(0)从原点附近出发且从未穿过原点的条件下。它具有标度形式(P_+(x,t) \sim t^{-H}R_+(x/t^H))。我们的目标是计算标度函数(R_+(y)),到目前为止仅在马尔可夫情形(H = 1/2)时已知其结果。我们围绕此极限发展了一种系统的微扰理论,令(H = \frac{1}{2} + \varepsilon),以计算标度函数(R_+(y))到(\varepsilon)的一阶项。我们发现,当(y \to 0)(在吸收边界附近)时,(R_+(y))的行为类似于(R_+(y) \sim y^{\phi});而当(y \to \infty)时,(R_+(y) \sim y^{\gamma}\exp(-y^2/2)),其中(\phi = 1 - 4\varepsilon + O(\varepsilon^2))且(\gamma = 1 - 2\varepsilon + O(\varepsilon^2))。我们关于(\varepsilon)展开的结果证实了佐亚、罗索和马宗达在[《物理评论快报》102, 120602 (2009)]中提出的标度关系(\phi = (1 - H)/H)。我们通过对(H = 2/3)进行数值模拟验证了我们的发现。这里开发的工具具有通用性、强大性且适用于不同情况。