Multiple peaks in fft spectrum

[Diksha Prajapati]

Hello all,

I am trying to get fft of each cell corresponding to a particular frequency value. However I am getting multiple peaks in the fft spectrum of y-component of magnetization (plot attached). I referred to the 'Skyrmion excitation example'. Kindly refer to the code below.

https://colab.research.google.com/github/JeroenMulkers/mumax3-tutorial/blob/master/postprocessing.ipynb

# MODE IDENTIFICATION

mode1_idx = 7   # determined visually

mode2_idx = 58  # determined visually

mode1_freq = freq[mode1_idx]

mode2_freq = freq[mode2_idx]

print("Mode 1 frequency: %.2f GHz"%(mode1_freq/1e9))

print("Mode 2 frequency: %.2f GHz"%(mode2_freq/1e9))

# PLOT THE SPECTRUM

plt.plot(freq/1e9, spectr)

plt.axvline(mode1_freq/1e9, lw=1, ls='--', c='gray')

plt.axvline(mode2_freq/1e9, lw=1, ls='--', c='gray')

plt.xlim(0,fmax/1e9)

plt.ylabel("Spectrum (a.u.)")

plt.xlabel("Frequency (GHz)")

plt.show()

In this example there were only two peaks so it is convenient to plot them manually, howsoever in my case I am

getting multiple peaks, hence it is very time taking to first determine visually each peak and then to calculate fft of each cell

corresponding to each frequency value. Is there any way by which-

1. I can get value of frequency corresponding to each peak?

2. FFT of each cell corresponding to particular value of frequencies obtained above.

I would be very thankful if someone can help me with this.

With regards

Diksha

[Josh Lauzier]

Hi,

Instead of finding them visually, you can use a peak finding algorithm. There are many different types, so I would pick one off of Google that you like. It probably doesn't need to be very complex. Since you're already using python, something such as this built in scipy function might be convenient: https://docs.scipy.org/doc/scipy/reference/generated/scipy.signal.find_peaks.html

To select the modes, you can replace the portion of the code "mode1 = mz_fft[mode1_idx]" with a little for loop for each peak frequency value. You would probably also want to adjust the plotting to also be in a for loop.

Best,

Josh L.