pwSensitivityAnalysis3D
[sensX, sensY, sensZ] = pwSensitivityAnalysis3D(timepoint, reintegrate)
Applies a 3D sensitivity analysis.
sensX(i,j) Sensitivity of x variable i depending on parameter j
sensY y
sensZ z
timepoint Timepoint for which the sensitivity should be calculated.
If empty ([]), the mean value over time is used.
reintegrate If false, the trajectories of the last call to pwSensitivityAnalysis3D
are used to determine the sensitivity, e.g. for another time-point.
Note: The used couples should not be changed.
Default: true
Description
Applies a sensitivity analysis for all variables, observables and derived
variables and for all parameters. For very large models, the user can select
a subset of parameters in a graphical user interface. In contrast to the 1D
and 2D methods, no detailed trajectories are plotted. Sensitivity of a state
variable x_i on a parameter p_j is defined as
s_{ij}(t) = d(x_i(t))/d(p_j) * p_j / x_i(t)
for a certain time point t. Numerical approximation:
s_{ij} = (x_i(1.001 * p_j) - x_i(p_j))/(0.001 * p_j) * p_j / x_i(t)
= (x_i(1.001 * p_j) - x_i(p_j))/0.001 / x_i(t)
The factor 1.001 may be changed in the configuration:
config.analyses.sensitivity.factor
In this analysis, the sensitivity of the averaged concentration
for the integration time is calculated.
Note
The sensitivity of a variable can only be determined if its average
value or the value at the given timepoint is not zero.
Else, the result is a NaN value which is drawn
in 2D-matrix plots with the lowest available color, often dark blue.
Please use the output arguments to check for NaN values.
See also