py5vec : a modular Python package for the 5-vector method to search for continuous gravitational waves

L. D’Onofrio, F. Muciaccia, L. Mirasola, M. Pitkin, C. Palomba, P. Leaci, F. Safai Tehrani, F. Amicucci, L. Silvestri, L. Pierini · 2026

Read py5vec : a modular Python package for the 5-vector method to search for continuous gravitational waves by L. D’Onofrio, F. Muciaccia, L. Mirasola, M. Pitkin, C. Palomba, P. Leaci, F. Safai Tehrani, F. Amicucci, L. Silvestri, L. Pierini (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-15703.txt.

Introduction

D'Onofrio and co-authors release py5vec, a Python package for continuous wave searches. The package implements the 5-vector method for faint, steady gravitational wave signals. It also reviews the method's theory in full. Two extensions marginalize over noise variance and starting phase. The first gives a sturdier Student's t likelihood. The second guards against pulsar glitches. A modular layout splits data handling, demodulation, and inference. Searches on LIGO O4a data rebuilt injected signals well.

Catalog record

CREATORL. D’Onofrio, F. Muciaccia, L. Mirasola, M. Pitkin, C. Palomba, P. Leaci, F. Safai Tehrani, F. Amicucci, L. Silvestri, L. Pierini
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE56598
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.15703

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

py5vec : a modular Python package for the 5-vector method to search for continuous gravitational waves L. D’Onofrio, F. Muciaccia, L. Mirasola, M. Pitkin, C. Palomba, P. Leaci, F. Safai Tehrani, F. Amicucci, L. Silvestri, L. Pierini Abstract We present py5vec , a Python package for implementing and extending the 5-vector method, used to search for continuous gravitational wave (CW) signals. We also provide a comprehensive theoretical review of the 5-vector method and extend the relative likelihood formalism by marginalizing over the noise variance, resulting in a more robust Student’s t-likelihood, and over the initial phase to account for pulsar glitches. py5vec provides a modular architecture that separates data representation, signal demodulation, and statistical inference into independent abstract stages. It supports multiple input data formats and interoperates with existing Python software, providing a bridge between different approaches. For example, using a bilby -based interface, py5vec implements Bayesian parameter estimation within the 5-vector formalism for the first time. The modular design also allows for making exact multi-level and direct comparisons between other software, such as cwinpy and SNAG in MATLAB. In py5vec , we implement a multidetector targeted search for known pulsars, validated using LIGO data from the O4a run and hardware injections, demonstrating consistent reconstruction of signal parameters. This package therefore provides a flexible platform for current targeted searches and for future extensions to other CW search strategies. 1 Introduction Continuous gravitational waves (CWs) from rapidly rotating neutron stars represent one of the most challenging and promising targets for ground-based gravitational wave detectors. Unlike transient signals, CWs are weak, long-lived, and nearly monochromatic, requiring highly specialized data analysis techniques to extract them from detector noise [ 39 ] . Over the years, a variety of search strategies have been developed, depending on the knowledge of the source parameters (sky position, rotation frequency and orbital parameters); targeted and narrowband searches for known pulsars observed by radio/X-ray/Gamma-ray telescopes, directed searches for sources with only the sky position measured (for instance, neutron stars in supernova remnants), and all-sky surveys for completely unknown objects. Despite their differences, most CW searches share a common physical structure: a deterministic signal model with slowly varying phase evolution, amplitude modulation induced by the detector response, and a small set of unknown parameters that must be inferred or marginalized over. However, practical implementation of these searches and knowledge of source parameters often leads to pipelines that are optimized for a specific analysis strategy, making it difficult to compare methods or reuse components across different search paradigms (see [ 43 ] for a complete review of algorithms and methods). Within the class of targeted searches, the 5-vector method provides a compact and efficient frequency-domain formulation of the matched filter, based on the residual amplitude modulation of the signal due to the Earth’s sidereal rotation [ 12 , 11 ] . This approach has been extensively used in past analyses (e.g. latest results in [ 1 ] ) and has recently been connected to likelihood-based and Bayesian formulations of CW inference [ 20 ] . The existing implementation of the 5-vector method is embedded in a software environment, SNAG [ 24 ] , developed in MATLAB and closely tied to specific data formats, limiting its flexibility and extensibility. In this work, we present py5vec , a Python package that implements the 5-vector formalism within a modular and framework-independent architecture. First, we provide a comprehensive theoretical review of the state-of-the-art of the 5-vector method and the recent likelihood formulation. In addition, we generalize the likelihood formulation by relaxing the assumption of perfectly known Gaussian noise. By treating the noise variance as a nuisance parameter and marginalizing over it, we derive a Student’s t-likelihood formulation improving robustness against noise misestimation. We also extend the likelihood formalism by marginalizing over the initial phase to account for pulsar glitches. While the current implementation of py5vec focuses on targeted searches, the design is intentionally agnostic to the specific CW search class. The package is built around a minimal set of abstract data objects and transformations that encode the physical core of CW analyses, rather than a particular end-to-end pipeline. The guiding principles of py5vec are threefold: (i) separation of data representation, signal demodulation, and statistical inference; (ii) interoperability with existing, well-tested software packages commonly used in the CW community; and (iii) modularity that allows analysis components to be reused, replaced, or extended without modifying the core. Within this framework, the 5-vector method emerges as a specific realization of a more general approach, rather than as a closed and self-contained pipeline. By construction, this architecture naturally lends itself to future extensions. Alternative demodulation strategies, different detection statistics, or extension to different searches can be incorporated while preserving the same underlying data model and interfaces. As such, py5vec is not only a tool for current targeted searches, but also a flexible platform for the development, validation, and comparison of CW analysis methods across different search regimes. The current implementation of py5vec enables a complete multidetector targeted search for known pulsar allowing for the first time a Bayesian parameter estimation within the 5-vector formalism. Further developments are planned, including an extended comparison with established pipelines and application to real O4 [ 2 ] dataset from the LIGO-Virgo-KAGRA detectors [ 30 , 23 , 4 ] . The paper is organized as follows. In Section 2 , we provide a review of the state-of-art of the 5-vector method, while in Section 3 , we describe the new extension of the 5-vector likelihood formalism. In Section 4 , we describe the guiding principles and the general architecture of py5vec . The validation of the py5vec targeted search considering real data and hardware injections is described in Section 5 . Conclusion and future prospects are finally presented in Section 6 . 2 Review of the 5-vector method In this section, we provide a review of the 5-vector method, originally proposed in [ 12 ] . The method has been widely used in the CW community and has been extended to narrowband [ 11 ] , semicoherent [ 19 ] , directed [ 5 ] , and ensemble [ 16 ] searches. The purpose of this section is to provide a complete guide and a future reference for the 5-vector theory development, also in light of the recent work in [ 20 ] , which introduces a likelihood for the first time within the 5-vector formalism. 2.1 General principle The 5-vector method is generally defined as a matched filter in the frequency domain. After correcting the detector data for Doppler modulation due to the Earth’s motion and for the intrinsic spin-down of the source, the signal becomes nearly monochromatic. The time dependence of the detector response, induced by the Earth’s sidereal rotation, modulates the signal amplitude and redistributes the signal power into five discrete frequency components at \omega_{0}+k\Omega_{\oplus} , where \omega_{0} is the CW angular frequency, \Omega_{\oplus} is the Earth’s sidereal angular frequency and k=0,\pm 1,\pm 2 . According to the considered emission model, \omega_{0} is proportional to the source rotation angular frequency. In the single-harmonic emission model [ 26 ] , the CW angular frequency is exactly twice the source rotation angular frequency. In the complex formalism, firstly described in [ 12

Read the complete text