Check out our paper "Pattern finding in millimetre-wave spectra of massive young stellar objects" to explore the full story! Feel free to ask questions or share your thoughts aanda.org/articles/aa/...

Our approach is a robust, automated alternative to traditional rotational diagram fitting. This method is able to accelerate spectral analysis and yield physical and chemical information about MYSO evolution.

By training a simple random forest on the first few PCA components (as proof of concept), we demonstrated that you can directly derive the physicochemical state of a MYSO from its spectra. This streamlines spectral analysis and minimizes manual intervention.

Using a Gaussian mixture model in our reduced (embedded) parameter space, we identified these clusters. Principal component analysis (PCA) further supported an evolutionary pathway across the groups.

Our new analysis revealed three distinct MYSO populations:
(i) Cold, COM-poor sources
(ii) Warm, medium-COM-abundance sources
(iii) Hot, COM-rich sources
These groups hint at different evolutionary stages in massive star formation.

Their spectral line information showed that not all MYSOs share the same physico-chemical states. This points to plausible evolutionary pathways for MYSOs

They found several amazing complex organic molecules (COMs) like methanol, methyl acetylene, and methyl cyanide in these massive star forming regions

Using ALMA Band 6 data, we analyzed mm-wave spectra from 41 massive young stellar objects (MYSOs) sourced from diverse catalogs (N. Frimpong's PhD work).

We used dimensionality reduction techniques like Locally Linear Embedding (LLE) and Principal Component Analysis (PCA) to uncover hidden patterns in the spectra. This means we can get key physicochemical info without the manual hassle.

Traditionally, we manually extract spectral lines and fit them to build these diagrams. This is a process that’s usually time-consuming since each spectral line needs its own separate profile fit

Quick refresher: a rotational diagram plots the log-column density (divided by statistical weight) against the energy of a molecule’s rotational levels. This lets us infer the excitation temp. & column density of a chemical species. This is very helpful in star formation studies.

This is the amazing work of Yenifer Angarita @yangarita.bsky.social (UdeA 🇨🇴 graduate), whom I co-supervised while she was an MPhil student at Leeds Uni.🇬🇧. Coindicentally, she just finished her PhD at Radboud Uni. 🇳🇱 and is currently a Postdoc at Chalmers Uni. 🇸🇪. Ok now the paper ->

Paila, ya me congelé