Short Period Wobble in RR CET?

Hello! I’ve gone back and started to look at data gather on November 21, 2023 through AAVSONet. Visually, I had thought I saw low amplitude period of about 1 hour in the roughly 12 hour curve.
I don’t remember much from my VStar course, but I thought I would post the graphs for feedback and guidance.
The first graph is the light curve for 11/21/2023; the second is the phase plot with period 0.55303; the third is the period analysis (DC DFT) ranging from 0.03 days to 0.07 days (giving a period of less than an hour to a little over 2 hours) with a resolution of 0.001.
The data shows several humps in the period analysis. I wonder if this is normal for good quality data regardless of the variable? That is, to have several small regular period fluctuations so that this would not be just for




RR CET ? Best regards. Mike

What is the cadence of the observations? You can get peaks (and aliases) from that.

-Kenneth

What you need to see here to evaluate the data quality and trends is the differential magnitudes of two (or more) comparison stars (comp1 minus comp2). That is what will allow you to estimate errors in the data for the variable.

\Brian

It seems to me the phase plot just repeats (twice) your posted light curve. The curve represents about one quarter of a cycle and the phase plot uses the same data. Therefore there isn’t enough data to ‘fold’ for the phase plot.

1 Like

Thank you, all, for your suggestions.
There are years of data for RR CET. I thought I would try to isolate data from AAVSONet since it should provide the highest quality data. I was not sure if the “wobble” in magnitude was due to my system, so that is why I had requested AAVSONet. The location that took the images was close to my location, so that helped minimize other potential confounders.
I’m interested in the low amplitude wobbles shown in the Period analysis. I’m not sure how to interpret it. Does this sinusoidal pattern represent a true short period low magnitude change that most short period variables might show? Could this be a “real” low amplitude variation in magnitude, but due to something else, like comp star changes as suggested by Brian. Could the the high frequency of images taken by AAVSONet that night (it was the only target, I believe) have led to a spurious result as Kenneth suggests? Lots of questions that I’m not sure how to evaluate.
RR CET is starting to be visible for extended periods now. I’ll try CV rather the BVI in order to get short cadence sequences. I’ll check the comps against one another as well. Best regards. Mike

As a reference / comparison, I took a look at the RR Cet photometry from TESS in similar time frame (TIC 344299442 in sector 70, i.e., 2023-09-21 to 2023-10-15) with a cadence of 200 s almost continuously.

The Lomb-Scargle Periodogram in the similar range (period of 0.03 - 0.07 d) only shows peaks that are harmonics.

The harmonics (blue dotted lines below) are easier to visualize in frequency view:

Phase plot of the TESS lightcurve:


It looks like you’re using VStar for analysis. If so, you can download the TESS data (link) and inspect / analyze them yourself as well.

1 Like

Thanks! I don’t know anything about variable star harmonics. Is this something to be expected for a star like RR CET? Best regards. Mike

Harmonics are frequently seen in the periodograms of lightcurves with stellar pulsations, including RR Lyraes such as RR CET here.

For RR CET, the base frequency is about 1.8 1/d (1 / 0.552941 d). In a periodogram in frequency view, one can spot equally-spaced peaks starting with the base frequency.

The same periodogram but zoomed out to include the base frequency:

(The peaks in the range of period of 0.03 - 0.07 d, or frequency of ~15 - 40 1/d are too small in this zoomed-out plot.)

Thanks! It does not sound like harmonics would be particularly interesting to folks. Best regards. Mike

I think the main message here is the short-term variations you are seeing on the declining part of the lightcurve are instrumental in nature — simply not visible in the TESS l.c. above that has very high internal precision.
It may be worth noting that fitting RR Lyrae lightcurves usually requires fairly high-order Fourier fits, like at least order-4, or even much higher, 7,9,11 depending on internal errors and the significance of the coefficients relative to the errors on those coefficients.

\Brian

Illustrating Brian’s point on fitting RR Lyrae lightcurve, the following is a plot of the model with 1 (basically sine curve), 4, and 7 coefficients:


(Edit: initial plot has an error. The plot has been replaced.)

By trial and error, it looks like one could use up to 24 23 coefficients (adding more doesn’t shrink the residuals):

Not enough data… to be doing Fourier power spectra generation based on one night with clearly only a partial light curve. Keep observing–it is job security perhaps.

TESS filters are very wide band which doesn’t exclude in-and-of-itself short-term stuff that might be seen in certain variable star types using narrower band filters. Google AI basic gives this as one possibility for short-term variability for RR Lyr types: Unknown Causes: The exact physical trigger remains debated, but leading theories point to magnetic fields, convection cycles, or nonlinear resonances between pulsation modes

My 2-cents! :money_mouth_face:

1 Like

Even using a 7 harmonic fit that is very poor fit. What is the explaination for the low model bias, even with 24 coefficient model (harmonics?)?

What software did you use?

The low model bias is an error on my part.

It’s based on Python Astropy LombScargle, with some additional logic to build up a model with multiple coefficients / harmonics.
It looks like in the additional logic in building up a model, the resulting model is not properly normalized.

The 7-coefficient model fit after the model is normalized:


I also repeat the same exercise using VStar. The result of a 7 coefficient model is comparable:

I’ve often wondered what astrophysical information is obtained by fitting a high order polyomial function to light curve data. Then I came across a paper describing the results of fitting cubic polynomials to RR Lyrae light curves: Reyner S. et al, MNRAS 407, p. 1081, 2010.

Thank you, all, for your comments. Years of data to see if any harmonics represent an actual physical change in RR CET. My goal in getting data from AAVSONet was to see if there was a difference between that data and my data with regards to this low amplitude “wobble,” that is, was there something in my system/data collection causing it. I’m still not sure that I can answer that questions! Best regards. Mike

On DEY_VAR’s point that TESS filters are very wide band, and the “wobble” might be only visible in some narrower band: I don’t know if that’s the case, but maybe you already have data to test it: e.g., if your existing data is in BVRI, are the “wobbles” visible in all bands or just in specific band?

Fourier models or any other simple model incl. piece-wise anything are meant to model the shape of the light curve of pulsating variable stars only. Sometimes used to generate time-of-maximum for O-C curve generation to monitor for stellar evolution through the H-R diagram instability region. Logical to use Fourier if nothing else as RR Lyr stars are pulsators of a type of harmonic, periodic non-linear oscillation.

If you want to model real physical processes doing a search using basic Google AI :robot: gives:

To generate an RR Lyrae light curve from real physical processes rather than empirical mathematical curves, astrophysicists use Non-linear Radiative Hydrodynamic (RHD) Stellar Pulsation Models.
Instead of treating the star as a collection of arbitrary sine waves, these models treat the star as a 3D or 1D physical sphere of gas governed by the fundamental laws of thermodynamics, gravity, and fluid dynamics.

1. The Core Physical Processes Simulated

To generate a true physical light curve, a computer program must model three main coupled physical mechanisms acting simultaneously on the star’s layers:

[ Gravity pulling inward ] <===> [ Gas & Radiation Pressure pushing outward ]
▲
│ (Regulated by)
[ Kappa Mechanism ]

  • The Thermodynamic Driving Engine (The $\kappa$-Mechanism): This is the physical “valve” that powers the pulsation. In the ionization zones of Helium (He II to He III), compressing the gas makes it more opaque ($\kappa$ increases) rather than hotter and more transparent. This traps radiation from the core, building up immense pressure.
  • Non-linear Hydrodynamics: Once the pressure builds up, the layer explodes outward. As it expands, it cools, becomes transparent, releases the trapped radiation (causing the steep rise in the light curve), and gravity pulls the layer back down to compress and restart the loop.
  • Time-Dependent Convection: Energy doesn’t just radiate; it moves via boiling gas currents (convection). Modern physical models must simulate how turbulent convection interacts with the expanding and contracting shockwaves.

2. The Governing Equations of a Real Physical Model

A physical code solves a system of partial differential equations for dozens of concentric spherical shells (“zones”) across the star. For a shell at radius $r$ with mass coordinate $M_r$, density $\rho$, pressure $P$, velocity $v$, and luminosity $L$, the core equations are:
$$\text{Conservation of Mass (Continuity):} \quad \frac{\partial r}{\partial M_r} = \frac{1}{4\pi r^2 \rho}$$
$$\text{Conservation of Momentum (Equation of Motion):} \quad \frac{\partial v}{\partial t} = -4\pi r^2 \frac{\partial (P + Q)}{\partial M_r} - \frac{GM_r}{r^2}$$
$$\text{Conservation of Energy:} \quad \frac{\partial E}{\partial t} + P \frac{\partial (1/\rho)}{\partial t} = -\frac{\partial L}{\partial M_r}$$

  • (Note: $Q$ is the artificial viscosity used to handle physical shockwaves traveling through the stellar atmosphere, and $E$ is internal energy).

3. Industry-Standard Physical Software Packages

Astronomers do not use Fourier series to create these; they run heavy structural software packages to let the physical equations evolve over time. The most widely used codes include:

  1. MESA (Modules for Experiments in Stellar Astrophysics): The modern global open-source standard. MESA includes a dedicated RSP (Radial Stellar Pulsation) module. You input basic physical parameters—Mass ($M$), Luminosity ($L$), Effective Temperature ($T_{\rm eff}$), and Metallicity ($Z$)—and MESA physically simulates the pulsations over time, naturally outputting a raw, physical light curve.
  2. The Warsaw Nonlinear Pulsation Code: Developed by Smolec and Moskalik, this hydrodynamical code is specifically optimized for classical pulsators like Cepheids and RR Lyrae to study non-linear behaviors, mode selections, and period changes.
  3. The Florida-Budapest Hydro Code: A foundational code used extensively to model the highly non-linear shock waves that propagate through RR Lyrae atmospheres during their rapid expansion phase.

Comparison: Physical Model vs. Fourier Model

Metric / Attribute Real Physical Model (e.g., MESA RSP) Fourier Model
Inputs Mass ($M$), Radius ($R$), Temperature ($T$), Composition ($X,Y,Z$) Amplitude ($A_k$), Phase ($\phi_k$), Period ($P$)
Underlying Math Partial Differential Equations (Hydrodynamics) Trigonometric Polynomials
Predictive Power High. Explains why the star pulsates and changes. Low. It only describes what the light curve looks like.
Computational Cost High (Minutes to hours of CPU time per star) Extremely Low (Milliseconds of CPU time)

If you want to explore running or understanding these models, let me know:

  • Would you like to walk through how to configure the physical inputs (Mass, Temperature, Metallicity) in a simulation code like MESA RSP?
  • Are you interested in how physical shockwaves in the star’s atmosphere create the specific asymmetric features seen on the final light curve?
    To generate an RR Lyrae light curve from real physical processes rather than empirical mathematical curves, astrophysicists use Non-linear Radiative Hydrodynamic (RHD) Stellar Pulsation Models.
    Instead of treating the star as a collection of arbitrary sine waves, these models treat the star as a 3D or 1D physical sphere of gas governed by the fundamental laws of thermodynamics, gravity, and fluid dynamics.

1. The Core Physical Processes Simulated

To generate a true physical light curve, a computer program must model three main coupled physical mechanisms acting simultaneously on the star’s layers:

[ Gravity pulling inward ] <===> [ Gas & Radiation Pressure pushing outward ]
▲
│ (Regulated by)
[ Kappa Mechanism ]

  • The Thermodynamic Driving Engine (The $\kappa$-Mechanism): This is the physical “valve” that powers the pulsation. In the ionization zones of Helium (He II to He III), compressing the gas makes it more opaque ($\kappa$ increases) rather than hotter and more transparent. This traps radiation from the core, building up immense pressure.
  • Non-linear Hydrodynamics: Once the pressure builds up, the layer explodes outward. As it expands, it cools, becomes transparent, releases the trapped radiation (causing the steep rise in the light curve), and gravity pulls the layer back down to compress and restart the loop.
  • Time-Dependent Convection: Energy doesn’t just radiate; it moves via boiling gas currents (convection). Modern physical models must simulate how turbulent convection interacts with the expanding and contracting shockwaves.

2. The Governing Equations of a Real Physical Model

A physical code solves a system of partial differential equations for dozens of concentric spherical shells (“zones”) across the star. For a shell at radius $r$ with mass coordinate $M_r$, density $\rho$, pressure $P$, velocity $v$, and luminosity $L$, the core equations are:
$$\text{Conservation of Mass (Continuity):} \quad \frac{\partial r}{\partial M_r} = \frac{1}{4\pi r^2 \rho}$$
$$\text{Conservation of Momentum (Equation of Motion):} \quad \frac{\partial v}{\partial t} = -4\pi r^2 \frac{\partial (P + Q)}{\partial M_r} - \frac{GM_r}{r^2}$$
$$\text{Conservation of Energy:} \quad \frac{\partial E}{\partial t} + P \frac{\partial (1/\rho)}{\partial t} = -\frac{\partial L}{\partial M_r}$$

  • (Note: $Q$ is the artificial viscosity used to handle physical shockwaves traveling through the stellar atmosphere, and $E$ is internal energy).

3. Industry-Standard Physical Software Packages

Astronomers do not use Fourier series to create these; they run heavy structural software packages to let the physical equations evolve over time. The most widely used codes include:

  1. MESA (Modules for Experiments in Stellar Astrophysics): The modern global open-source standard. MESA includes a dedicated RSP (Radial Stellar Pulsation) module. You input basic physical parameters—Mass ($M$), Luminosity ($L$), Effective Temperature ($T_{\rm eff}$), and Metallicity ($Z$)—and MESA physically simulates the pulsations over time, naturally outputting a raw, physical light curve.
  2. The Warsaw Nonlinear Pulsation Code: Developed by Smolec and Moskalik, this hydrodynamical code is specifically optimized for classical pulsators like Cepheids and RR Lyrae to study non-linear behaviors, mode selections, and period changes.
  3. The Florida-Budapest Hydro Code: A foundational code used extensively to model the highly non-linear shock waves that propagate through RR Lyrae atmospheres during their rapid expansion phase.

Comparison: Physical Model vs. Fourier Model

Metric / Attribute Real Physical Model (e.g., MESA RSP) Fourier Model
Inputs Mass ($M$), Radius ($R$), Temperature ($T$), Composition ($X,Y,Z$) Amplitude ($A_k$), Phase ($\phi_k$), Period ($P$)
Underlying Math Partial Differential Equations (Hydrodynamics) Trigonometric Polynomials
Predictive Power High. Explains why the star pulsates and changes. Low. It only describes what the light curve looks like.
Computational Cost High (Minutes to hours of CPU time per star) Extremely Low (Milliseconds of CPU time)

If you want to explore running or understanding these models, let me know:

  • Would you like to walk through how to configure the physical inputs (Mass, Temperature, Metallicity) in a simulation code like MESA RSP?
  • Are you interested in how physical shockwaves in the star’s atmosphere create the specific asymmetric features seen on the final light curve?

I guess it depends on what you want to do. That AAVSO spectroscopy data being collected may be useful in providing some of the inputs for a physical model fit to an observed light curve. A paper in the future for someone to do and something to do in our spare time! :star_struck:

Thanks! A bit beyond me. I was interested in knowing if there was something in my system that I needed to change/update that might have accounted for the “wobble.” RR CET is well observed but is also on the SPP list, so I’ll keep it as part of my program. Best regards. Mike