arXiv2024
Cosmic filaments, the most prominent features of the cosmic web, possibly hold untapped potential for cosmological inference. While it is natural to expect the structure of filaments to show universality similar to that seen in dark matter halos, the lack of agreement between different filament finders on what constitutes a filament has hampered progress on this topic. We initiate a programme to systematically investigate and uncover possible universal features in the phase space structure of cosmic filaments, by generating particle realizations of mock filaments with $\textit{a priori}$ known properties. Using these, we identify an important source of bias in the extraction of radial density profiles, which occurs when the local curvature $κ$ of the spine exceeds a threshold determined by the filament thickness. This bias exists even for perfectly determined spines, thus affecting $\textit{all}$ filament finders. We show that this bias can be nearly eliminated by simply discarding the regions with the highest $κ$, with little loss of precision. An additional source of bias is the noise generated by the filament finder when identifying the spine, which depends on both the finder algorithm as well as intrinsic properties of the individual filament. We find that, to mitigate this bias, it is essential not only to smooth the estimated spine, but to $\textit{optimize}$ this smoothing separately for each filament. We propose a novel optimization based on minimizing the estimated filament thickness, along with Fourier space smoothing. We implement these techniques using two tools, $\texttt{FilGen}$ which generates mock filaments and $\texttt{FilAPT}$ which analyses and processes them. We expect these tools to be useful in calibrating the performance of filament finders, thereby enabling searches for filament universality.