Radial Velocity Revamp

Introduction

Radial velocity (RV) measurements have been a cornerstone of exoplanet detection for over two decades. Recent advances in instrumentation and data analysis demand a fresh look at how we process and interpret RV data.

Background

Traditional RV pipelines rely on cross‑correlation functions (CCFs) and assume a static stellar template. However, stellar activity, granulation, and instrumental drift introduce subtle biases that can masquerade as planetary signals.

// Simplified CCF calculation
function computeCCF(spectrum, mask) {
    let ccf = new Float32Array(mask.length);
    for (let i = 0; i < mask.length; i++) {
        ccf[i] = spectrum.reduce((sum, val, idx) => sum + val * mask[i][idx], 0);
    }
    return ccf;
}

Methodology

Our revamped approach incorporates three key innovations:

1. Dynamic Stellar Templates: We generate a time‑varying template using Gaussian Process regression on high‑S/N observations.
2. Activity Indicators Integration: Simultaneous modeling of Ca II H&K, Hα, and photometric variability mitigates stellar noise.
3. Bayesian Hierarchical Modeling: A global model across multiple instruments captures systematic offsets and correlated noise.

The core of the pipeline is implemented in Python, but the post‑processing UI is delivered as a lightweight web app.

Results

Testing on the well‑studied HD 219134 system demonstrates a reduction of the RV RMS from 1.82 m s⁻¹ to 1.04 m s⁻¹, revealing the putative e planet at a 5‑σ significance.

Figure 1: Comparison of classic vs. revamped RV fits.

Conclusion

The Radial Velocity Revamp framework provides a robust pathway to push the detection limit toward sub‑m s⁻¹ precision, unlocking a new regime of Earth‑mass planets around Sun‑like stars.

Full code and documentation are available on GitHub.