Testing time symmetry in time series using data compression dictionaries.
read more
Citations
Towards parameter-free data mining
Complex network approaches to nonlinear time series analysis
Time series irreversibility: a visibility graph approach
Time series irreversibility: a visibility graph approach
Compression-based data mining of sequential data
References
Elements of information theory
Compression of individual sequences via variable-rate coding
Testing for nonlinearity in time series: the method of surrogate data
Testing for nonlinearity in time series: The method of surrogate data
An Introduction to Symbolic Dynamics and Coding
Related Papers (5)
Multiscale Analysis of Heart Rate Dynamics: Entropy and Time Irreversibility Measures
Frequently Asked Questions (17)
Q2. What is the standard deviation of z for the shortest data sets?
For the larger data sets the standard deviation of z is near unity and distribution of pk is uniform, but for the shortest data sets, N=250, the standard deviation of z is less than 1, i.e., there is somewhat of a central ten-dency in the pk.
Q3. What is the reversible representation of the symbolic sequences?
Despite the topological equivalence, the presence or absence of probabilistic reversibility in the symbolic sequences becomes reversed by the change in presentation.
Q4. What is the main experimental parameter for the spark ignition engine?
For the spark-ignition engine, the input air-fuel ratio was maintained in stoichiometric conditions, but the proportion of exhaust gas recirculation (EGR) was altered for various runs and was the principal experimental parameter.
Q5. What is the implicit-state version of the graph?
The implicit-state version is a sofic shift with an associated graph and labeling: a distinct binary symbol is emitted depending on which edge is taken on the transition.
Q6. What is the effective experimental parameter for the diesel data?
For the Diesel data, the fraction of residual gas remaining from one combustion cycle to the next was estimated with changes in experimental parameters and is the effective experimental parameter.
Q7. What is the length of the string?
The dictionary is initialized with A length-one strings, each comprising each unique symbol in the alphabet of size A. Absent a priori bounds on the maximum size of the integers, the length, in bits, of the compressed stream is proportional to n log2 n with n the number of phrases.
Q8. What is the language of theoretical symbolic dynamics?
In the language of theoretical symbolic dynamics [19], the explicit-state version is a presentation of a “vertex shift,” as a symbol is emitted corresponding to each new vertex of the transition graph which is visited, and hence explicitly a shift of finite type (with memory 1) on a three-symbol alphabet.
Q9. What is the likely explanation for the low-dimensional chaos?
Physically what is most likely is that this dynamics is dominated by sufficiently high-dimensional turbulent fluctuations that globally averaged quantities such as the one considered here are effectively indistinguishable from linear processes by some kind of central limit theorem effect.
Q10. What is the way to compress a dictionary?
The algorithm appends only one symbol at a time to each dictionary entry to form new dictionary entries, thus the phrases it finds are not sufficiently long to have excellent compression.
Q11. What is the standard deviation of z under the null?
As expected, there is no indication of time asymmetry in u or z, and the standard deviation of z under the null is close to unity.
Q12. What is the difficult requirement to test for time symmetry?
The following statistic and test, though, was powerful in detecting irreversibility, the relatively easy task, as well as having a good calibration of the null hypothesis under various diverse instantiations of reversible dynamics, which is the more difficult requirement.
Q13. What is the shortest segment in the input sequence?
The Lempel-Ziv [11] dictionary compression algorithm sequentially parses the input symbol sequence from left to right, at each step finding the longest segment in the remaining input which already exists in a dictionary of codewords [21].
Q14. What is the reversibility test for squaring?
As squaring is a nonmonotonic transformation, these data would reject the null with this sort of surrogate data method, but here the reversibility test correctly recognizes the data as being in the null class.
Q15. What is the way to break up a data set?
In that case, the data set ought to be broken up into more, shorter, interleaved training and test sets, accumulated and repeated.
Q16. What is the effect of changing alphabet size?
Figure 6 shows the effect of changing alphabets: with significant irreversibility, increasing alphabet size improved detecting it, but if irreversibility were minimal, the alphabet size was unimportant.
Q17. What is the entropy rate of the two representations?
The Shannon entropy rates fhSsM1d <0.5623 bits/ symbol,hSsM2d<0.7602 bit/ symbolg of the two representations are identical, as there is the same amount of uncertainty about the next state and the same invariant density.