Membrane noises, intrinsic ar extrinsic, are conspicuous and widespread entities that have an important role in neuronal information processing. Studies to elucidate the effects of noise on nerve cells can be dane with relative ease using mathematical models. Nevertheless, even simple models are shown to require very complicat ed mathematics which are valid for very limited conditions. Computer simulations are more adequate for these cases but they require a convenient numerical integration methodology to avoid erroneous ar paradoxical results. It is shown in this work that for a fourth arder Runge-Kutta a reasonable choice is to divide the noise sequence variance by the step size and to choose the intermediate input value required at each step equal to the present value. Thereafter, the methods used to simulate three different colóred Gaussian noises are presented. An approximately l/f noise was obtained as the output of an IIR filter excited by white noise. The IIR filter was designed using the least squares inverse method.Lorentzian noise was generated by an exact recursive relation. The third type of noise was obtained by passing white noise through a resonant low-pass filter. The prototype continuous-time filter was obtained from the literature and the corresponding IIR digital approximation was designed by the impulse response invariance method.

Membrane noises, intrinsic ar extrinsic, are conspicuous and widespread entities that have an important role in neuronal information processing. Studies to elucidate the effects of noise on nerve cells can be dane with relative ease using mathematical models. Nevertheless, even simple models are shown to require very complicat ed mathematics which are valid for very limited conditions. Computer simulations are more adequate for these cases but they require a convenient numerical integration methodology to avoid erroneous ar paradoxical results. It is shown in this work that for a fourth arder Runge-Kutta a reasonable choice is to divide the noise sequence variance by the step size and to choose the intermediate input value required at each step equal to the present value. Thereafter, the methods used to simulate three different colóred Gaussian noises are presented. An approximately l/f noise was obtained as the output of an IIR filter excited by white noise. The IIR filter was designed using the least squares inverse method.Lorentzian noise was generated by an exact recursive relation. The third type of noise was obtained by passing white noise through a resonant low-pass filter. The prototype continuous-time filter was obtained from the literature and the corresponding IIR digital approximation was designed by the impulse response invariance method.