Patterns Embedded In Time Series By Means Of RMT Huijie Yang (杨会杰) Biz School University of Shanghai for Science and Technology Jungong Rd#516,Shanghai, 200092 E-mail: hjyang@ustc.edu.cn hjyang@usst.edu.cn Date: June 3, 2011 Add.: Biz School USST
Random Matrix Theory 中国科学技术大学 顾雁
Statistics of Nuclear Levels/spectra Neutron scattering NNLS distribution
2. It is too complicated to be dealt with in detail. Hence,it can be dealt with in a simple way, The Hamiltonian of a complex quantum system
2. Some concepts 本征值分布 本征值间隔分布
谱刚度 量子混沌 Classical system: Chaos, regular Quantum system: Brody, Poisson
Localization and extension state Disorder induced localization
Tight-binding Hamiltonian 1) regular lattice: extension 2) Disorder lattice: localized state
3) What about aperiodic lattice? MULTIFRACTAL Power-law
Ideas Map time series to 1-dimensional disorder lattice by series element to site energy Structure of the lattice determines the characteristics of The waves on the lattice Characteristics of the wave on the lattice Can be used to Measure the structures of the series
time series where The site energies in 1-D lattice
Hopping integral Unfolding procedure:
1. RMT for Fractional Brownian Motion (FBM) series Some Results 1. RMT for Fractional Brownian Motion (FBM) series shuffling
Hurst exponents for the NNLS series H and H’ are uncorrelated: A new kind of fractal
q in the Brody distribution q-1 : Wigner distribution Non-distinguishable
Chaotic series 1 Logistic. Fractals for NNLS series
Conclusions Series analysis ---Anderson localizations For FBM series, new fractals NNLS distribution are non-distinguishable For Chaos series, NNLS distribution are consistent with Lyapunov exponent 4. Advantages: Physical picture; Use condensed mater theories
Thanks for your helpful comments Contributors Lily Zhao from USST Guimei Zhu from NUS Jie Ren from NUS Thanks for your helpful comments