--- title: "A Quick Tour through State Space Reconstruction" subtitle: "Tutorial 1 of The Short Course 'An Introduction to Empirical Dynamics Modelling', ICTP SAIRF" author: "Brenno Cabella, Paulo Inácio Prado, Renato Coutinho, Marina Rillo, Rafael Lopes, Roberto Kraenkel" date: "ICTP SAIFR, School on Physics Applications in Biology, January 2018" output: rmdformats::readthedown: self_contained: true thumbnails: true lightbox: true gallery: false highlight: tango toc_depth: 4 --- ```{r setup, echo=FALSE, warning=FALSE, message=F} library(knitr) library(rEDM) library(dplyr) library(plotly) opts_chunk$set(fig.align = 'center', fig.show = 'hold', fig.height = 5, fig.width = 5, warning = FALSE, message = FALSE, error = FALSE, echo=TRUE) options(formatR.arrow = TRUE,width = 90)###, cache=TRUE) ``` This tutorial presents the general idea of attractor reconstruction with synthetic data, and proposes two exercises with simulated data. For these activities, you will need the most recent version of [R](https://cran.r-project.org/) and the rEDM package installed in your work computer. To start, open R and load the package by typing the following code in the R console: ```{r load rEDM, echo=TRUE} library(rEDM) ``` Before you proceed please read through the essential concepts by reading the introduction and the first section (*"Empirical Dynamic Modelling"*) of rEDM's tutorial that you can find [here](https://cran.r-project.org/web/packages/rEDM/vignettes/rEDM-tutorial.html) or in your local R installation with the command ```{r call vignette, eval=FALSE} vignette("rEDM-tutorial", package="rEDM") ``` # A simple case: harmonic oscillator Let's first get to grips with R language, and figure out how to reconstruct a state space with only one time series and its lags. For the first example, we use a very well-known system and its representation on state space, the harmonic oscillator: $\frac{dx}{dt}=y$ $\frac{dy}{dt}=-x$ which has the solution $x(t)=A \sin(t), \ y(t)=A \cos(t)$, so we can draw the state space of this system by laying down in two Cartesian axis its two state variables, $x(t)$ and $y(t)$. The attractors of this system are circles, that is, all possible trajectories of a solution form a circle in the state space: ```{r generates data for the harmonic oscillator, echo=TRUE} ## Generate the data at tau = 0.2 time intervals time1<-seq(0, 500, by = 0.2) x1<-sin(time1) y1<-cos(time1) ## Plot the state space plot(x1, y1, type = 'l', main = "State-space of harmonic oscillator", cex=0.2, xlab="X(t)", ylab="Y(t)") ``` And here are the time series for the two state variables: ```{r time series of each variable} ## Time series of the 1st 250 observations ind <- 1:250 plot(time1[ind], x1[ind], type = 'l', main = "Time series of the variables of the system", col = "red", xlab = "Time", ylab = "Amplitude", ylim=c(-1,1.25)) lines(time1[ind], y1[ind], col = "blue") legend("topright", c("X","Y"), col=c("red", "blue"), lty=1, bty="n") ``` ## A reconstruction of the state space Now suppose that all the information you have about this system is a time series of one of the state variables, say $X(t)$. Because $Y(t)$ is missing, we are not able to construct the state space as we see above, but Takens theorem tells us that we can do a reconstruction of the state space that recovers the information from the al state space. This **state space reconstruction** creates an **embedding** of a **shadow manifold** of the attractor. So we can define the shadow manifold as a valid topological reconstruction of the attractor. To do a state space reconstruction we create new variables by lagging the time series by multiples of a time interval $\tau$. In this case we will use the interval between observations ($\tau=0.2$ time units). The R commands below create a data table object (which is called a *data frame* in R) with the original time series and its lagged version. The handy function `make_block` written by the authors of the rEDM package does this job. Use the command below to download the function code and load it in your R workspace (this function will be included in future releases of the rEDM package): ```{r load make_block, eval=TRUE, echo=TRUE} source("https://raw.githubusercontent.com/mathbio/edmTutorials/master/utilities/make_block.R") ``` Now you can create a data frame with the original and lagged series of data using the `make_block` function: ```{r plot one time series alone} ## Data frame with observations with tau = 0.2 with one lag df1 <- make_block(data.frame(x=x1), t=time1, max_lag = 2) ``` The most important arguments of this function are `max_lag` and `tau`; check the help page of this function (`help(make_tutorial)`) for more details. A quick look at the first and last lines of the data frame can help you understand the lagged variable structure created by `make_block`: ```{r check lagged data frame} ## Take a look in the lagged data to understand how the function works head(df1) tail(df1) ``` By plotting the time series as a function of their lagged versions we get the shadow manifold, which in this bi-dimensional case is easy to see: ```{r shadow manifold} plot(x_1 ~ x, data=df1, type="l", xlab = "X", ylab = "X(t + tau)" ) ``` As expected, the shadow is similar to the attractor. In fact, this similarity is well defined in topological terms. A key property of a valid shadow manifold is that it is a one-to-one map of the attractor. However, a shadow manifold will only be valid in this sense if it is embedded in the correct number of dimensions. There is an optimal number of dimensions to build a valid state space reconstruction, which we call the **embedding dimension** of a manifold. In the simple case of the harmonic oscillator we obviously need two dimensions. The Takens' Theorem gives us an upper limit of the number of dimensions - it guarantees that the number of dimensions is up to 2D + 1, where D is the dimensionality of the attractor. Finding the optimal embedding dimension is part of a vast research topic called *Embedology* (Sauer *et al.* 1991), and the package `rEDM` provides some methods to do this. We will learn more about these methods in the following tutorials of this course. # Noisy data How robust is the reconstruction of the attractor to random noise? There are two main types of noise or error in data, which we call **process errors** and **observational errors**. The first one reflects random changes of the state variables, like environmental fluctuations that affect the dynamics. The second type of error is related to measurement error in the data, for instance when we miss individuals of some species when counting them, or have limited precision (inherent to all measurement). Observational errors do not affect the dynamics, only the portrait we make of it. In this section, we check how robust the attractor is to measurement error by adding simulated noise to the original time series. Use the commands below to simulate the time series with independent measurement errors of constant variance. To each observation, we add a value drawn from a Gaussian distribution with zero mean and standard deviation equal 1/12 of the standard deviation of the time series itself. ```{r plot one 2nd time series alone} ## a new vector of observed values + Gaussian iid error x2 <- x1 + rnorm(n = length(x1), mean = 0, sd = sd(x1)/12) ## Time series plot(x1[ind] ~ time1[ind], type="l", xlab = "Time", ylab = "X", col="grey", lwd = 5, ylim=range(x2)) lines(x2[ind] ~ time1[ind], type="l", col="red", lwd=2) ``` We proceed by creating a new data frame with the lagged variables including the noise, and then plot the new shadow manifold: ```{r 2D plot noisy series} ## Data frame with lagged variables df2 <- make_block( data.frame(x = x2)) ## Shadow manifold with embedding = 2 plot(x_1 ~ x, data=df2, type="l", ylab = "X(t + tau)", xlab = "X" , col="grey", lwd=0.5) ``` In this case, the shadow manifold still looks like the original shape of the attractor, despite the noise caused by measurement errors. ## Increasing the noise Measurement errors make attractors more complex by spreading points (and thus trajectories) over the state space. Hence, the trajectories may cross in the shadow manifold, which breaks down the one-to-one mapping to the original attractor. This effect gets clearly larger as we increase the measurement error. To investigate the effect of larger measurement errors in the reconstruction of the attractor, we will double the measurement error of the time series and again plot the shadow manifold: ```{r more noise} x3 <- x1 + rnorm(n = length(x1), mean = 0, sd = sd(x1)/6) df3 <- make_block(data.frame(x=x3), max_lag = 5) plot(x_1 ~ x, data=df3, type="l", col="grey", lwd=0.5, ylab = "X(t + tau)", xlab = "X" ) ``` The attractor reconstruction looks worse now, but we can overcome this problem by adding more dimensions to the state space reconstruction. By doing this we unfold the shadow manifold, by unfolding crossing trajectories. To add more dimensions, we simply add more axes, that correspond to the time series lagged by the further time steps ($t + \tau, \ t + 2\tau, \ t + 3\tau \dots$). You can check with the interactive plot below how an additional dimension helps unfold the shadow manifold. [^1] In the following days, we will learn some methods to find the optimal dimension of the state space reconstruction. For now, the take-home message is: the more **complex** the attractor, or the larger the **error** (observational or not), the more dimensions you will need to make a good reconstruction. ```{r more noise 3D plot, echo=FALSE, fig.align="center"} plot_ly(df3, x = ~x, y=~x_1, z=~x_2, type="scatter3d", mode="lines", opacity=0.25) %>% layout(scene=list(camera = list(eye = list(x = 0, y = 0, z = -2)), xaxis = list(title = 'X'), yaxis = list(title = 'X (t + tau)'), zaxis = list(title = 'X (t + 2tau)'))) ``` # Exercises ## What is the effect of sampling interval? The commands below simulate a new time series of the bi-dimensional harmonic oscillator without measurement error but with a much smaller interval between observations (0.01 time units instead of 0.2). ```{r simulates oscillator series with smaller sampling } time4<-seq(0, 500, by = 0.01) x4<-sin(time4) y4<-cos(time4) ``` This change of the sampling frequency affects the shape of the reconstruction: ```{r smaller sampling interval manifold} df4 <- make_block(data.frame(x=x4), t=time4) plot(x_1 ~ x, data=df4, type="l", xlab = "X", ylab = "X(t + tau)" ) ``` How do you explain this? Which practical issues does this fact bring and how to solve them? ### Hints **1.** The command below adds a line $X(t) = X(t + \tau)$ to the state space reconstruction, which can help you understand the problem regarding small values of $\tau$: ```{r abline, eval=FALSE} abline(0,1, col="red") ``` **2.** Check this site: (http://www.physics.emory.edu/faculty/weeks//research/tseries4.html) ## Attractor reconstruction In this exercise you have to reconstruct the state space of a time series, and check if a three-dimensional plot improves the reconstruction. To load the data in R run the command: ```{r problem time series} prob2 <- read.csv("https://raw.githubusercontent.com/mathbio/edmTutorials/master/takens/problem-time-series.csv") ``` To take a look at the time series plot run: ```{r plot time series problem 2} plot( X ~ time, data = prob2, type="l") ``` ### Hints **1.** Recall from the previous problem that the sampling interval changes the shape of the shadow manifold. The sampling frequency can be controlled when using models, but with data this is usually not possible. With data, you can only vary $\tau$, which is by how many time steps each variable is lagged. For details see the help page of the `make_block` function. **2.** To create a simple 3D plot in R you can use the `scatterplot3d` package. The example below plots it for one of the noisy time series we examined above: ```{r 3D plot with scatterplot3d} ## Loads the package library(scatterplot3d) ## 3D plot scatterplot3d(x = df2$x, y = df2$x_1, z = df2$x_2, type = "l", color="grey", lwd = 0.5, xlab = "x(t)", ylab = "x(t+tau)", zlab = "x(t + 2 tau)") ``` If the package is not available in your local R installation you can install it with: ```{r install packages, eval=FALSE} install.packages("scatterplot3d") ``` # Learn more * Sugihara, G., May, R., Ye, H., Hsieh, C.H., Deyle, E., Fogarty, M. and Munch, S., 2012. Detecting causality in complex ecosystems. Science, 338(6106), pp.496-500. * DeAngelis, D.L. and Yurek, S., 2015. Equation-free modeling unravels the behavior of complex ecological systems. Proceedings of the National Academy of Sciences, 112(13), pp.3856-3857. * [Takens Theorem](https://www.youtube.com/watch?v=QQwtrWBwxQg&list=PL-SSmlAMhY3bnogGTe2tf7hpWpl508pZZ&index=2), a video by Sugihara's Lab. * Takens, Floris. Detecting strange attractors in turbulence. Lecture notes in mathematics 898.1 (1981): 366-381. * Deyle, E.R. and Sugihara, G., 2011. Generalized theorems for nonlinear state space reconstruction. PLoS One, 6(3), p.e18295. * Sauer, T., Yorke, J.A. and Casdagli, M., 1991. Embedology. Journal of statistical Physics, 65(3), pp.579-616. # Glossary ```{r include glossary, child = '../glossary.md'} ``` [^1]: To run the R code that creates this plot in your computer you need the `plotly` package.