Could anyone kindly point out why I get very different results?
There are many peaks which should not appear.In fact,there should be only one peak.
I am a python newbie and all comments about my code below are welcome.

The test data is here.enter link description here
You can directly wget https://clbin.com/YJkwr .
Its first column are the arrival times for a series of photons.A light source emits photons randomly. The total time is 55736s and there are 67398 photons. I try to detect some kind of periodicity of light intensity.
We can bin the time and light intensity is proportional to the photon numbers in every time bin.

I tried numpy.fft and lomb-scargle of scipy.signal to make its power spectrum,but got very different results.

fft

 import pylab as pl
 import numpy as np

 timepoints=np.loadtxt('timesequence',usecols=(0,),unpack=True,delimiter=",")
 binshu=50000
 interd=54425./binshu  
 t=np.histogram(timepoints,bins=binshu)[0]
 sp = np.fft.fft(t)
 freq = np.fft.fftfreq(len(t),d=interd)  
 freqnum = np.fft.fftfreq(len(t),d=interd).argsort()  
 pl.xlabel("frequency(Hz)")
 pl.plot(freq[freqnum],np.abs(sp)[freqnum])

lomb-scargle

 timepoints=np.loadtxt('timesequence',usecols=(0,),unpack=True,delimiter=",")
 binshu=50000
 intensity=np.histogram(timepoints,bins=binshu)[0].astype('float64') 
 middletime=np.histogram(timepoints,bins=binshu)[1][1:binshu+1]-np.histogram(timepoints,bins=binshu)[1][3]*0.5
 freq1=1./(timepoints.max()-timepoints.min())
 freq2=freq1*len(timepoints)
 freqs=np.linspace(freq1,freq2,1000)
 pm1=spectral.lombscargle(middletime,intensity,freqs)

 pl.xlabel("frequency(Hz)")
 pl.plot(freqs,pm1)
 pl.xlim(0.5*freq1,2*freq2)
 pl.ylim(0,250)
 pl.show()

*********************************
Thank you, Warren,I appreciate it.Thank you for your detailed reply.

You are right,there is an integration time,here it is around 1.7s.
There are many single exposures.Every exposure costs 1.7s
In one single exposure,we can not tell its arrival time precisely.

If the time series are like:
0 1.7 3.4 8.5 8.5

The integration time of the last two photons are 1.7s,not (8.5-3.4)s.So I will revise part of your code.

HOWEVER my question still remains. You adjust several parameters to get the 0.024Hz peak in lomb-scargle peak to some degree. And you use this to guide your parameters in fft.

If you do not know the the number 0.024, maybe you may use different parameters to get a different highest peak ?

How to guarantee every time we can get right the num_ls_freqs ? You can see if we choose different num_ls_freqs,the highest peak shifts.

If I have many timing series, every time I have to specify different parameters? And how to get them?