<p>

Generating non-white random noise means producing arbitrary long time series with a given spectral density.
Such time series are needed for example for the following purposes:
</p>
<p>
	<ul>
		<li> To generate test data sets for programs that compute spectral densities, </li>
		<li> as inputs for various simulations.</li>
	</ul>
</p>
<p>
  One way of doing this is to apply digital filters (FIR or IIR) to white input noise.<br/>
  This approach is effectively implemented for the generation of <a href="ndim_ng.html"> multichannel noise </a>
  with a given cross spectral density. <br/>
  Multichannel transfer functions are identified by an automatic fit procedure based 
  on a modified version of the vector-fitting algorithm 
  (see <a href="zdomainfit.html"> Z-Domain Fit </a> for further details on the algorithm). <br/>
  Partial fraction expansion of multichannel transfer functions and the implementation of 
  <a href="sigproc_dfilt.html">filter</a> state initialization avoid the presence of unwanted 'warm-up period'.
</p>

<p>
	A different approach is implemented in LTPDA as <a href="franklin_ng.html">Franklin noise-generator</a>.<br/>
	It produces spectral densities according to a given pole zero model (see <a href="pzmodel.html">Pole/Zero Modeling</a>) and does not require any warm-up period. <br/>
</p>