A pseudorandom pink noise for the computer-based measurements of linear responses

We propose a nonbinary pseudorandom sequence for the measurement of linear responses. Compared with a maximal length shift register sequence $(m$ sequence), this sequence is more suitable for digital processing based on computers; with the use of discrete Fourier transforms, the response functions are reproduced without approximations from digitally sampled data because the input power spectrum is pink, i.e., completely flat in a given frequency range and 0 otherwise. In practice, the new sequence is simply the sum of harmonics with arbitrary phases and is produced readily with digital wave form generators. For reducing the peak power of this sequence, the amplitude distribution is better to be bimodal rather than Gaussian. For demonstrating its feasibility with common digital hardware, the magnetic resonance of Rb atoms in a sub-MHz region was measured successfully. With the use of the fast Fourier transform algorithm, our software task was only 0.4% of that for the cross-correlational calculation based on an $m$ sequence.