The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.

ODEsensitivity

Dirk Surmann

2019-01-09

Introduction

The goal of sensitivity analysis is to examine how sensitive a mathematical model responds to variations in its input variables. Here we focus on the sensitivity analysis of ordinary differential equation (ODE) models via Morris screening.

If the assumption of a uniform distribution on the domain intervals doesn’t hold, the Morris screening method cannot be used and the variance-based Sobol’ method should be considered instead. In this case, simply switch from using the function ODEmorris to the function ODEsobol.

Analyse the Lotka-Volterra Equations

The Lotka-Volterra equations describe a predator and its prey’s population development and go back to Lotka (1925) and Volterra (1926). The prey’s population at time \(t\) (in days) will be denoted with \(P(t)\) and the predator’s (or rather consumer’s) population with \(C(t)\). \(P(t)\) and \(C(t)\) are called state variables. This ODE model is two-dimensional, but it should be noted that ODE models of arbitrary dimensions (including one-dimensional ODE models) can be analyzed with ODEsensitivity.

Model Definition

Now we define the model according to the definition in deSolve::ode().

Each of the five parameter names, their lower and upper boundaries, the initial values for the state variables and the timepoints of interest are saved in separate vectors:

Sensitivity Analysis

The sensitivity analysis of a general ODE model can be performed by using the generic function ODEsensitivity::ODEmorris().

Let’s take a look at the output LVres_morris.

The first row of each state variable contains a copy of all timepoints. The other rows contain the Morris sensitivity indices \(\mu\), \(\mu^\star\), and \(\sigma\) for all 5 parameters and all 51 timepoints.

Plotting

ODEsensitivity provides a plot() method for objects of class ODEmorris:

plot.ODEmorris() has two important arguments: pars_plot and state_plot. Using pars_plot, a subset of the parameters included in the sensitivity analysis can be selected for plotting (the default is to use all parameters). state_plot gives the name of the state variable for which the sensitivity indices shall be plotted (the default being the first state variable):

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.