-
Fast Algorithm for computing a matrix transform used to detect trends in noisy data
Authors:
D. J. Kestner,
G. R. Ierley,
A. B. Kostinski
Abstract:
A recently discovered universal rank-based matrix method to extract trends from noisy time series is described in [1] but the formula for the output matrix elements, implemented there as an open-access supplement MATLAB computer code, is ${\cal O}(N^4)$, with $N$ the matrix dimension. This can become prohibitively large for time series with hundreds of sample points or more. Based on recurrence re…
▽ More
A recently discovered universal rank-based matrix method to extract trends from noisy time series is described in [1] but the formula for the output matrix elements, implemented there as an open-access supplement MATLAB computer code, is ${\cal O}(N^4)$, with $N$ the matrix dimension. This can become prohibitively large for time series with hundreds of sample points or more. Based on recurrence relations, here we derive a much faster ${\cal O}(N^2)$ algorithm and provide code implementations in MATLAB and in open-source JULIA. In some cases one has the output matrix and needs to solve an inverse problem to obtain the input matrix. A fast algorithm and code for this companion problem, also based on the above recurrence relations, are given. Finally, in the narrower, but common, domains of (i) trend detection and (ii) parameter estimation of a linear trend, users require, not the individual matrix elements, but simply their accumulated mean value. For this latter case we provide a yet faster ${\cal O}(N)$ heuristic approximation that relies on a series of rank one matrices. These algorithms are illustrated on a time series of high energy cosmic rays with $N > 4 \times 10^4$. [1] Universal Rank-Order Transform to Extract Signals from Noisy Data, Glenn Ierley and Alex Kostinski, Phys. Rev. X 9 031039 (2019).
△ Less
Submitted 27 January, 2020;
originally announced January 2020.
-
Nucleation at the contact line observed on nano-textured surfaces
Authors:
C. W. Gurganus,
J. C. Charnawskas,
A. B. Kostinski,
R. A. Shaw
Abstract:
It has been conjectured that roughness plays a role in surface nucleation, the tendency for freezing to begin preferentially at the liquid-gas interface. Using high speed imaging, we sought evidence for freezing at the contact line on catalyst substrates with imposed characteristic length scales (texture). Length scales consistent with the critical nucleus size and with $δ\sim τ/ σ$, where $τ$ is…
▽ More
It has been conjectured that roughness plays a role in surface nucleation, the tendency for freezing to begin preferentially at the liquid-gas interface. Using high speed imaging, we sought evidence for freezing at the contact line on catalyst substrates with imposed characteristic length scales (texture). Length scales consistent with the critical nucleus size and with $δ\sim τ/ σ$, where $τ$ is a relevant line tension and $σ$ is the surface tension, range from nanometers to micron. It is found that nano-scale texture causes a shift in the nucleation of ice in supercooled water to the three-phase contact line, while micro-scale texture does not.
△ Less
Submitted 28 October, 2014;
originally announced October 2014.
-
Record-breaking statistics for random walks in the presence of measurement error and noise
Authors:
Yaniv Edery,
Alexander B. Kostinski,
Satya N. Majumdar,
Brian Berkowitz
Abstract:
We address the question of distance record-setting by a random walker in the presence of measurement error, $δ$, and additive noise, $γ$ and show that the mean number of (upper) records up to $n$ steps still grows universally as $< R_n> \sim n^{1/2}$ for large $n$ for all jump distributions, including Lévy flights, and for all $δ$ and $γ$. In contrast to the universal growth exponent of 1/2, the p…
▽ More
We address the question of distance record-setting by a random walker in the presence of measurement error, $δ$, and additive noise, $γ$ and show that the mean number of (upper) records up to $n$ steps still grows universally as $< R_n> \sim n^{1/2}$ for large $n$ for all jump distributions, including Lévy flights, and for all $δ$ and $γ$. In contrast to the universal growth exponent of 1/2, the pace of record setting, measured by the pre-factor of $n^{1/2}$, depends on $δ$ and $γ$. In the absence of noise ($γ=0$), the pre-factor $S(δ)$ is evaluated explicitly for arbitrary jump distributions and it decreases monotonically with increasing $δ$ whereas, in case of perfect measurement $(δ=0)$, the corresponding pre-factor $T(γ)$ increases with $γ$. Our analytical results are supported by extensive numerical simulations and qualitatively similar results are found in two and three dimensions.
△ Less
Submitted 4 February, 2013;
originally announced February 2013.
