🔭 astrophysics

Astrophysics Wrapped 2025: Year-in-Review of Every Astrophysics arXiv Paper from 2025

This paper presents a comprehensive 2025 year-in-review of all astrophysics arXiv submissions, analyzing metrics such as keywords, subfields, instruments, author affiliations, and citation impact to provide a holistic overview of the field's current state, hot topics, and global research landscape.

Original authors: Rommulus Francis Lewis, Hetansh Shah, Amruth Alfred

Published 2026-06-19

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Rommulus Francis Lewis, Hetansh Shah, Amruth Alfred

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the world of astrophysics as a massive, bustling library that never sleeps. Every single day, thousands of new books (research papers) are written and added to the shelves. In 2025, this library exploded with activity, adding over 18,600 new books.

This paper is like a "Year-in-Review" playlist for that library, similar to how Spotify tells you your top songs of the year. The authors, a team of researchers, acted as librarians who read every single title and summary from 2025 to tell us what the whole community was thinking, doing, and talking about.

Here is the story of 2025 in astrophysics, broken down into simple parts:

1. The Big Picture: A Library Overflowing

The main takeaway is that the library is getting crowded. In 2025, researchers published about 72 new papers every single day (excluding weekends).

The "Hot" Topics: If you walked into this library, you'd see the most crowded aisles are about Galaxies (how they form and change) and High-Energy Phenomena (explosions and intense energy).
The "Quiet" Aisles: Surprisingly, papers about Cosmology (the study of the universe's origin) were fewer in number, but they were the most "borrowed" (cited) by other scientists. It's like having a small section of rare, ancient books that everyone wants to read.

2. The Tools of the Trade: The Telescope Olympics

The authors counted how often different telescopes were mentioned in these papers.

The Gold Medalist: The James Webb Space Telescope (JWST) was the clear winner. It was mentioned in more papers than any other tool, largely because it sees the universe in infrared light (heat), which is currently the hottest way to look at the cosmos.
The Veteran: GAIA, a telescope that has stopped operating, is still being used constantly because scientists are still analyzing its massive archive of data.
The Dark Horse: The Einstein Probe didn't appear in many papers, but the few papers that did use it were cited heavily. It's like a small, quiet band that only plays one song, but everyone loves that one song so much they play it on repeat.

3. The "Who's Who": A Global Potluck

The paper looked at who is writing these books.

The Big Players: Scientists from the United States and China wrote the most papers.
The Team Dynamic: Most research is still done by "local teams." Imagine a group of friends from the same town working together on a project. While there are huge international teams (like the LIGO group that listens for gravitational waves), the average paper is written by a small group of people, mostly from the same country.
The Cost of Publishing: The authors did a shocking calculation. If you add up the fees scientists pay to publish their papers in journals, the astrophysics community spent about $17 million in 2025. That's enough to build a small observatory! They point out that while the public funds the science through taxes, they often have to pay again to read the results, and the scientists have to pay to publish them.

4. The "Spectral Fingerprint": What Colors Are We Looking At?

The authors created a "fingerprint" of the light scientists were studying.

Infrared is King: Because of the James Webb telescope, a huge chunk of research is focused on infrared light (heat).
Optical is Saturated: The visible light spectrum (the colors our eyes can see) is "saturated," meaning almost every possible topic in visible light has already been studied extensively. It's like a room where every wall is covered in paintings; there's no empty space left.

5. The "Wrapped" Awards: Fun Facts

Just like a year-end playlist, the authors gave out some quirky awards based on their data:

The Longest Book: A textbook on machine learning for astronomy that was 677 pages long.
The Most Pictures: One paper had 609 figures (charts and graphs) and 953 tables. It was essentially a giant data spreadsheet.
The Most Cited Paper: A report on the DESI telescope's data, which helps map the universe.
The Most Prolific Author: One scientist, J. Carretero, published 138 papers in a single year.

6. The "Fingerprint" of the Future

The paper concludes that while the topics change, the way scientists work hasn't. Whether you are studying black holes or exoplanets, you are just as likely to be working in a small local team as you are in a massive global collaboration.

The Bottom Line:
This paper is a mirror held up to the astrophysics community in 2025. It shows a field that is growing faster than ever, obsessed with the James Webb telescope, heavily focused on galaxies, and spending a fortune to share its discoveries. It's a celebration of the collective effort to understand the universe, reminding us that while we are all looking at different stars, we are all reading from the same library.

Technical Summary: ASTROPHYSICS WRAPPED 2025

Problem Statement
The field of astrophysics is experiencing an overwhelming increase in research output, evidenced by a year-over-year rise in papers submitted to the arXiv repository. In this "Golden Age of Astronomy," the sheer volume of daily submissions makes it exhausting for researchers to keep pace with progress outside their specific sub-fields. While previous annual summaries (e.g., Trimble et al.) provided qualitative overviews, the dataset has since more than doubled in size and become highly inhomogeneous, rendering manual summarization of every aspect of the field impractical. There is a lack of comprehensive, data-driven analyses that capture both broad trends and niche areas without requiring deep expertise in every sub-discipline.

Methodology
The authors conducted a quantitative analysis of all 18,660 papers uploaded to the Astrophysics section of arXiv in 2025. The methodology involved automated data collection and extraction using the following techniques:

Data Source: The dataset comprises primary submissions only, excluding replacements and cross-submissions. Metadata was scraped daily from arXiv, including titles, abstracts, comments, and LaTeX source files.
Citation Metrics: Citation data was retrieved via the NASA Astrophysics Data System (ADS). The authors defined four specific indices to mitigate bias regarding publication status and self-citations:
- All Articles Citation (AAC) and Excluding Self AAC (EAAC): Averages for all arXiv articles.
- Journal Articles Citation (JAC) and Excluding Self JAC (EJAC): Averages for articles confirmed as published in journals.
Content Extraction: A text-matching algorithm was applied to titles and abstracts to extract specific metrics:
- Redshift: Discrete values counted as single mentions; ranges were distributed across 50 equally weighted points to avoid undervaluing range-based studies.
- Telescopes & Instruments: String matching against a list of ~60 facilities, including instrument names (e.g., MIRI, WFC3).
- Keywords & Subfields: For papers lacking author-assigned keywords, three keywords were assigned by matching title/abstract terms against a dataset of the top 1,000 used keywords. Subfields were assigned based on a predefined list of ~100 terms.
- Geolocation: Country and region names were extracted from LaTeX files and cross-referenced with ADS author affiliations to ensure high completeness.
Collaboration Indices: The study introduced four indices to quantify collaboration:
- Local Collaborative Index (LCI): Number of authors from the same country/region as the first author.
- Local Collaborative Ratio (LCR): Ratio of local authors to total authors.
- Global Collaborative Index (GCI): Number of authors from different countries/regions.
- Non-repeated Global Collaborative Index (NGCI): Number of unique foreign countries/regions involved.
Validation: The authors performed manual checks on 100 randomly selected papers, reporting fidelity rates of ~95–97% for affiliations and content extraction, with an estimated coverage of 90–95% of the field.

Key Results

Publication Volume: In 2025, 18,660 papers were submitted (approx. 51/day including weekends), a significant increase from 16,333 in 2024. The most active months were October, September, and July.
Primary Subjects: "Astrophysics of Galaxies" was the most submitted category (4,761 papers). However, "Cosmology and Nongalactic Astrophysics" papers received the highest average citations (AAC: 4.83; JAC: 7.02).
Spectral & Redshift Trends: The "2025 Astrophysics Spectral Fingerprint" revealed the infrared band as the most studied region (driven by JWST), followed by the optical band. Redshift distributions showed a peak at $z=0$ , with a notable interest in the epoch of reionization ( $z \approx 6$ ).
Telescopes: The James Webb Space Telescope (JWST) was the most mentioned telescope. However, the Atacama Cosmology Telescope (ACT) and Dark Energy Spectroscopic Instrument (DESI) led in citation impact per paper. LIGO events were under-represented in raw telescope mentions due to naming conventions (e.g., "GWXXXXXX") but were identified via specific event pattern matching (193 papers).
Collaboration: Research is predominantly domestic. The median LCR is 0.86, indicating ~86% of authors on a typical paper are from the same country as the first author. The median GCI is 1.00. "Astrophysics" (the general category) showed the highest international collaboration, likely due to mega-collaborations, while "Solar and Stellar Astrophysics" was the most localized.
Institutions: The top three producing institutions were the Chinese Academy of Sciences, the University of California, and the Harvard & Smithsonian Center for Astrophysics.
Journal Costs: The authors estimated a total publishing cost of ~$17 million USD for the year (assuming Gold Open Access rates), averaging ~$2,400 per published paper. Only ~2.28% of published papers appeared in Diamond Open Access journals (free to read and publish).
Publication Rate: Approximately 62% of 2025 arXiv papers had been published in journals by December 2025. The authors estimate this rises to ~80% when accounting for papers currently in review.

Significance and Claims
The paper claims to be the first comprehensive, data-driven "year-in-review" of the entire Astrophysics arXiv category, moving beyond the limitations of manual summaries like those by Trimble et al. Its significance lies in:

Holistic Overview: Providing a "Spectral Fingerprint" and global collaboration maps that visualize the state of the field without requiring deep technical knowledge of every sub-discipline.
Community Reflection: Offering a "pat on the back" to the community to appreciate collective efforts rather than inciting competition. It aims to help students and professionals gauge the general state of the field and adjust their research trajectories.
Policy and Planning: The authors suggest these statistics can inform macro-decisions, such as grant funding allocation, the construction of new observatories, and the prioritization of research areas in decadal surveys (e.g., NASA's decadal surveys).
Transparency: By quantifying the high costs of publication and the low adoption of Diamond Open Access, the paper highlights economic barriers in scientific communication.

The authors explicitly state that the work is not a "cheat-sheet" for maximizing metrics but a tool for understanding the field's current trajectory and fostering healthy collaboration. They acknowledge limitations in text-matching fidelity and the exclusion of non-arXiv communities but assert that the data captures the overwhelming majority of the field's output.