CHANGELOG.md

main

v3.2

• periodicorbits uses new internal datastructure to avoid duplicates in the output. New keyword argument abstol is introduced and keyword argument roundtol is deprecated.
• fixedpoints can now use order keyword argument to compute fixed points of n-th order iterations of DeterministicIteratedMap.
• New algorithm davidchacklai for detection of UPOs using stabilizing transformations.

v3.1

• fixedpoints can now be called without a Jacobian (giving nothing as third argument, also the default value). In this case a Jacobian is estimated automatically via ForwardDiff.jl.

v3.0

Major release part of the DynamicalSystems.jl v3. The DynamicalSystems.jl v3 changelog summarizes the highlights. Here we will list all changes to this specific package.

Breaking

• All diffeq keywords to all functions have been removed as per update to DynamicalSystems v3.0. Arguments to DifferentialEquations.jl solvers are now given in system construction when making a CoupledODEs (previously ContinuousDynamicalSystem).
• The dependency list of ChaosTools.jl has been reduced by half; many things that used to be exported from here are no longer (see refactoring).
• Low-level call signatures have been adjusted to fit DynamicalSystems.jl v3.0

Refactoring/removal

• All functionality related to attractors and basins of attraction (e.g., AttractorMapper, basins_fractions) has been moved to a new package Attractors.jl. Notice that functions such as fixedpoints and periodicorbits stay in ChaosTools.jl because they aren't only about attractors; they also find unstable fixed points and/or periodic orbits.
• All functionality related to Poincare maps has been moved to DynamicalSystemsBase.jl
• The state space samplers have been moved to StateSpaceSets.jl
• All fractal dimension related functionality has been moved to a new package FractalDimensions.jl. This includes the functionality for finding linear regions and fitting them.

Rare events

• New dedicated folder structure and functionality targeting rare events in ChaosTools.jl. If it becomes extensive, it can be split off to a new package.
• Source code for exit_entry_times has been completely overhauled and is now much much clearer.
• Algorithm for exit_entry_times for continuous systems has been re-written from scratch, and is now much more accurate, and much faster. Two ways are provided for finding crossings: linear intersections and high order interpolation via optimization.
• mean_return_times is now just a wrapper function.
• New function first_return_times for efficiently computing only the first time to return to sets.

Other enhancements

• orbitdiagram is a completely generic function that works for any kind of DynamicalSystem. produce_orbitdiagram is deprecated as it is now practically useless.
• ChaosTools.jl now has its own documentation as per DynamicalSystems.jl v3.0.
• More examples have been placed that also better highlight how to parallelize.
• Increased the default amount of c in testchaos01.
• Tests have been overhauled and never use predefined systems (which is a really bad practice when it comes to testing)

2.9

• Improved the AttractorsViaFeaturizing algorithm by improving the method for finding the optimal radius used in the clustering. This consisted in (i) maximizing the average silhouette values, instead of minimum (slight improvement), (ii) min-max rescaling the features for the clustering (big improvement); (iii) adding an alternative method ,called elbow method, that is faster but worse at clustering.
• Changed attractor_mapping_tests.jl to deal better with the Featurizing method.

2.8

• Brand new AttractorMapper infrastructure. It is a generic framework for mapping initial conditions to attractors and hence calculating basins of attraction and related quantities. Existing originally disparate functionality has been brought together under this framework.
• The old basins_of_attraction function has been completely deprecated in favor of using the version basins_of_attraction(mapper::AttractorMapper, grid), which utilizes the new AttractorMapper interface and is more intuitive and generalizable.
• New method for mapping initial conditions to known attractors using proximity.
• Old functions related to clustering via featurizing have been removed in favor of just using AttractorsViaFeaturizing.
• BREAKING Function basin_fractions_clustering is removed.
• basin_fractions is deprecated in favor of basins_fractions.

2.7

• New function basin_fractions_clustering that estimates fractions of basins of attraction via a random sampling and clustering technique.
• New function statespace_sampler that conveniently creates functions that sample state space regions.

2.6

• New yin function that applies the YIN algorithm to detect a signal's fundamental frequency. Also added it as a possible method in estimate_period.

2.5.3

• Updated lyapunovspectrum_convergence for continuous systems to be similar to lyapunovspectrum, with the more performant _buffered_qr and allowing passing a DynamicalSystem instead of the integrator directly.
• Added lyapunovspectrum_convergence for discrete systems.

2.5

• New fixedpoints function that finds fixed points, and their stability, for either continuous or discrete dynamical systems.
• Automatic Δt estimation for basins_of_attraction.

2.4

• linear_region now detects a saturation regime also at the start of the curve y(x). Furthermore a keyword sat now decides what the saturation regime is.

2.3

• Correlation sum functions now have a show_progress = false keyword. It shows a progress bar!
• linear_region and estimate_boxsizes have warning = true keyword.
• The estimate_boxsizes function now has keywords autoexpand = true, we = w, ze = z.

2.2

• Functions estimate_r0_buenoorovio, estimate_r0_theiler now return r0, ε0 with ε0 the minimum inter-point distance. This increases performance of other methods by reducing duplicate computations.
• Fixed bug in estimate_r0_buenoorovio that didn't consider min pairwise distance = 0.
• Exported the already implemented estimate_r0_theiler.
• For accuracy improvement, the mean of maximal lengths along each dimension is used in estimate_boxsizes. Before it was the maximum of maxima.

2.1

• Various improvements to the boxed correlation sum method. Now it also uses an automatic prism dimension.

2.0

• The keyword dt that was used to denote a chunk of time in many functions that a DynamicalSystem has been changed to Δt due to conflicts with DifferentialEquations.jl solver options. This change is breaking and cannot be warned or deprecated. Functions affected: lyapunov, lyapunovspectrum, gali, expansionentropy, orbitdiagram
• All deprecations have been removed, switch to previous stable version to see any.

1.34

• New option in basins_of_attraction that allows refining already found basins.

1.33

• lyapunovspectrum now has a progress meter option.

1.32

• New functions for examining the fractal nature of basin boundaries: basins_fractal_dimension, basin_entropy, basins_fractal_test.

1.31

• New, general function basins_of_attraction that replaces the existing basins_2D and basins_general and can produce arbitrary-dimensional basins of attraction.

1.30

• New function local_growth_rates
• New function match_attractors!

1.29

• Performance improvements for basins of attraction
• New function uncertainty_exponent
• New function basin_fractions
• New function tipping_probabilities

1.28

• New functions basins_2D and basins_general that efficiently compute basins of attraction on a plane!

1.27

• new function correlationsum_fixedmass implements a fixed mass algorithm for the correlationsum given by Grassberger in 1988.

1.26

• new function dyca() to perform Dynamical Component Analysis

1.25

• new function poincaremap for iterating over the Poincare map step by step.
• mean_return_times has been improved for continuous systems and now also allows diffeq... keyword propagation. The keyword m is also deprecated in favor of dmin.

1.24

• Theiler window is now possible in boxed_correlationsum.

1.23

• Various improvements on the functionality of estamte_boxsizes and linear_region.

1.22

• Orbit diagrams now use the previous state at each new parameter, providing faster convergence to attractor for smaller Ttr. The previous option is still available by passing explicitly u0 = get_state(ds).

1.21

• Using keyword α is deprecated in favor of q in all entropy-related discussions (q is more common in the literature).
• Added boxed_correlationsum and boxed_correlationdim that distribute the data into boxes before calculating the correlationsum.
• Added estimate_r0_buenoorovio to find the optimal boxsize for the former two functions.

1.20

• Keyword u0 is now valid for lyapunov.

1.19

• A lot of functions have been deprecated in favor of the new syntax that uses Entropies.jl: non0hist, binhist, genentropy.
• information_dim, capacity_dim, boxcounting_dim are deprecated.
• Permutation entropy has been re-written from scratch to use the Entropies.jl version. This drops the (completely unnecessary) argument interval, however the old method is available as ChaosTools.permentropy_old. It will be removed completely in version 2.0.
• correlationsum now features the keyword q to calculate the q-order correlationsum.
• Add fractal dimension estimation method by Molteno et al molteno_dim.
• lyapunovs is deprecated in favor of lyapunovspectrum.

1.18

• poincaresos function now also works with input Dataset (and does linear interpolation between points sandwiching the hyperplane)

1.17

• transit_time_statistics deprecated in favor of exit_entry_times.
• Added mean_return_times function for discrete systems.
• Added mean_return_times function for continuous systems.

1.16

• takens_best_estimate now returns three arguments, the estimate plus the upper and lower 95%-confidence limits.

1.15

• New function transit_time_statistics that allows computing return times and transit times to subsets of the state space. (Currently for discrete systems only)
• Moved support to Julia 1.5+.

1.14

• orbitdiagram now allows only collecting states within user-provided limits.

1.13

• Takens' best estimate method for estimating the correlation dimension is available as takens_best_estimate.

1.12

• binhist method that returns data histogram and bin edges
• further optimization of correlationsum for vector εs.

1.11

• Theiler's correction is now possible in estimating the correlation sum (as a result, norm is now a keyword).

1.10

• Added method for estimating correlation dimension (and as a pre-requisite, also the correlation sum) based on the method of Grassberger-Proccacia.
• Added "kernel density nearest neighbor" estimator for probabilistic description of a dataset.

1.9

• The expansion entropy from Hunt and Ott (defining chaos 2015) is now included as expansionentropy! Thanks and welcome to our new contributor @yuxiliu1995 !

1.8

• new function testchaos01 which implements the so called "0-1" test for chaos, that can test if a numeric timeseries is chaotic or not.

1.7

• New function estimate_period that attempts to estimate the period of a signal using the following methods:
  • The autocorrelation function (when it comes close to 1 again)

1.6

• Implementation of Wernecke, H., Sándor, B. & Gros, C. How to test for partially predictable chaos. Scientific Reports 7, (2017). Thanks and welcome to our new contributor @efosong ! Implemented with the function predictability.

1.5

• Updated everything to new SciMLBase 5.0 and the new default integrator (SimpleATsit5)
• Simplified low-level call signature for poincaresos.
• Small documentation improvements throughout.

1.4

• estimate_boxsizes now slightly expands borders if upper and lower do not have 2 orders of magnitude difference.

1.3

• increased performance of non0hist
• Documentation improvements

1.2.1

• Actually fix the issue with poincaresos and states starting on the plane
• The default of direction had to change to -1. We realized that having 1 was wrong and unintuitive with how the PSOS data is returned and plotted and thus we consider this a bugfix.

1.2

• allow u0 as keyword in poincaresos

1.1

• Fixed a bug in poincaresos when the initial condition was on the plane.
• Allowed poinracesos and produce_orbitdiagram to also configure the keywords of the root finding.

1.0 ≡ 0.13

First major release.

0.13

• Renamed all low-level methods that were exposed as part of the API to not start with an underscore anymore.
• Updated everything (REQUIRES etc) to julia 1.0.
• Added convergence return function for lyapunov exponents.

0.12

BREAKING

• Dropped support for all julia < 0.7
• Method for estimating reconstruction dimension now estimate temporal neighbor number to account for reconstruct.
• Reworked a bit how orbitdiagrams and co. behave. Now if given multiple states it is assumed that each state is for a different parameter. Please read the documentation strings of the functions.

New Features

• Brand new algorithm that computes poincare sections. Now uses interpolation of DiffEq and root finding of Roots.jl. This gives at least an order of magnitude speedup in produce_orbitdiagram and makes the source code massively more clear!!!
• It is now possible to choose which variables to save in both discrete and continuous orbit diagrams.
• Added method to compute mutual information, from A. Kraskov et al., [Phys. Rev. E 69, pp 066138 (2004)]
• Added method in finding delay time that uses mutual information. At the moment this method is vastly inferior to all others both in speed and in actual results.
• lyapunovspectrum is 1 to 2 orders of magnitude faster.

0.11

• Changed gali call signature to be the same as lyapunovspectrum.
• Bugfixed 1D lyapunov computation
• Bugfixed set_deviations! for continuous systems
• Updated for DynamicalSystemsBase 0.10 (using get_state etc.)
• Significantly reduced code repetition in the source.

v0.9

• Upgraded poincaresos to work for any arbitrary hyperplane. The call signature had to change to make it work.

v0.8.2

• Docstring typos fixed
• Removed aliases from genentropy .

v0.8

• Methods for estimating Reconstruction parameters moved here.
• Added Cao's method for estimating dimension.
• Added methods to find delay time:
  • First minimum
  • First zero
  • Exp. decay
  • Mutual information (NOT exported, needs to be tested and documented)
• Generalized numericallyapunov to allow choosing Theiler window (and adopted it to new neighborhood function).
• Keyword method in numericallyapunov changed to

v0.7

Breaking

• The code for estimating reconstruction parameters was moved to DynamicalSystemsBase.

Non-Breaking

• Theiler window is now configurable in numericallyapunov.
• Updated source and documentation for new Reconstruction.
• Type-stability tests and improvements.

v0.6

This version updates all internals to be on par with DynamicalSystemsBase 0.6

All internal implementations of methods that use a DynamicalSystem have been reworked almost from the ground up.

This lead to (only minor) changes to some of the high level interfaces. Please see the documentation strings of each function before using it.

v0.5.1

Non-breaking

• Bugfix: gali was invasive in the diff_eq_kwargs

v0.5.0

Breaking

• Updated everything to be on par with the changes of DynamicalSystemsBase v0.5

Non-breaking

• Minor bug fixes resulting from typos.
• New method for permutation entropy: permentropy !

v0.4.3

• Added Broomhead King function
• Estimate_delay and other functions about reconstructions are now in this repo.

v0.4.1

• Many bugfixes relating evolution of continuous systems and correct callback propagation.

v0.4.0

Breaking

• Updated everything to be on par with Base 0.3.1
• Added more tests
• Propagate callbacks etc in orbit diagrams

v0.3.0

BREAKING

• Overhauled the way estimate_boxsizes works.

v0.2.0

• added method for Poincare surface of section
• added orbit diagram for maps
• added production of orbit diagram for flows through Poincare surface of sections

v0.1.0

Initial release, see: https://juliadynamics.github.io/DynamicalSystems.jl/v1.0.0/ for the features up to this point.