CausalKinetiX (version 0.2.1)

generate.data.hidden: Hidden variable model

Description

Generate sample data from the hidden variable model.

Usage

generate.data.hidden(env = rep(1, 10), L = 15,
  par.noise = list(noise.sd = 0.01, only.target.noise = TRUE, relativ =
  FALSE), intervention = "initial_blockreactions",
  intervention.par = 0.1, hidden = TRUE, ode.solver = "lsoda",
  seed = NA, silent = FALSE)

Arguments

env

integer vector of length n encoding to which experiment each repetition belongs.

L

number of time points for evaluation.

par.noise

list of parameters that specify the added noise. noise.sd specifies the standard deviation of noise, only.target.noise specifies whether to only add noise to target and relative specifies if the size of the noise should be relative to size of variable (if TRUE standard deviation is given by par.noise$noise.sd*(x(t)-x(t-1))).

intervention

string specifying type of intervention. Currently three type of interventions are implemented "initial" (only intervene on intial values), "blockreactions" (intervene by blocking random reactions) or "intial_blockreactions" (intervene on both initial values and blockreactions").

intervention.par

if intervention is either "blockreactions" or "initial_blockreactions", the rate constant k_7 is set to a noisy version of k_4. This parameter specifies size of this noise, more precisely k_7 is set to k_4 + runif(1, -intervention.par, intervention.par).

hidden

boolean whether the variables H1 and H2 should be removed from output.

ode.solver

string specifying which ODE solver to use when solving ODE. Should be one of the methods from the deSolve package ("lsoda", "lsode", "lsodes", "lsodar", "vode", "daspk", "euler", "rk4", "ode23", "ode45", "radau", "bdf", "bdf_d", "adams", "impAdams", "impAdams_d", "iteration").

seed

random seed. Does not work if a "Detected blow-up" warning shows up.

silent

set to TRUE if no status output should be produced.

Value

list consisting of the following elements

simulated.data

D-matrix of noisy data.

time

vector containing time points

env

vector specifying the experimental environment.

simulated.model

object returned by ODE solver.

true.model

vector specifying the target equation model.

target

target variable.

Details

For further details see the references.

References

Pfister, N., S. Bauer, J. Peters (2018). Identifying Causal Structure in Large-Scale Kinetic Systems ArXiv e-prints (arXiv:1810.11776).

See Also

The functions generate.data.maillard and generate.data.targetmodel allow to simulate ODE data from two additional models.

Examples

# NOT RUN {
simulation.obj <- generate.data.hidden(env=rep(1:5, 3),
                                       L=15,
                                       par.noise=list(noise.sd=0.02,
                                                      only.target.noise=FALSE,
                                                      relativ=TRUE),
                                       intervention="initial_blockreactions",
                                       intervention.par=0.1)

D <- simulation.obj$simulated.data
fulldata <- simulation.obj$simulated.model
time <- simulation.obj$time
plot(fulldata[[1]][,1], fulldata[[1]][,2], type= "l", lty=2,
     xlab="time", ylab="concentration")
points(time, D[1,1:length(time)], col="red", pch=19)
legend("topright", c("true trajectory", "observations"),
       col=c("black", "red"), lty=c(2, NA), pch=c(NA, 19))
# }