问题:

I'm adapting MATLAB code to R and trying to generate a waveform using ARMA formula. Is there a simple R equivalent function for MATLAB's filter to take AR/MA coefficients to build a waveform?

npts = 100;
a = [1 0.6]; % AR coeffs
b = [1 0.25 3]; % MA coeffs
e = randn(npts,1); % generate gaussian white noise
waveform = filter(b,a,e); % generate waveform

回答1:

Hmm can't you achieve that with filter function in the package signal ?

require(signal)
a = c(1,0.6)
b = c(1,0.25,3)
e = rnorm(100)
waveform = filter(b,a,e)

回答2:

Yeah, you can do this usring arima.sim, e.g.

arima.sim(npts, model=list(ar=a, ma=b), rand.gen=rnorm)

Note that the model is checked for stationarity and the model you have above is not stationary. If you want something integrated you can specify the order of integration in the model.