Statistical Methods
Separate angular frequency from cycles in an irregular-time periodogram
Define the task
Uneven measurement times require an explicit time-aware method. Retain actual timestamps and their units rather than pretending the observations are regularly spaced. A Lomb–Scargle periodogram fits sinusoidal components over trial frequencies; a large peak alone is not a calibrated significance statement or proof of a biological cycle.
Carry out the check
Specify the trial frequency grid, normalization, centering or floating-mean choice and any weights. SciPy's frequency input is angular frequency: multiply cycles per time unit by 2π. Examine the time span, large gaps and observation schedule because these affect what patterns can be distinguished. Regular-grid Nyquist rules do not transfer unchanged to arbitrary uneven sampling. Multiple searched frequencies and model assumptions matter for uncertainty.
Worked hypothetical example
For a hypothetical candidate period of 8 hours, frequency is 0.125 cycles per hour and angular frequency is π/4, about 0.785398 radians per hour. Supplying 0.125 directly as an angular frequency would describe a period of about 50.2655 hours instead. If timestamps are recorded in minutes, convert all frequency units consistently. Record the searched interval and grid, inspect a fitted curve against the timestamps, and validate any claimed periodicity on suitable independent evidence.
Method checked 15 November 2026 against SciPy documentation: lombscargle.