arXiv · astro-ph.CO + astro-ph.IM · 2021–2024

Mapping the
frontier of cosmology.

An intelligence layer over real arXiv papers — finding the under-explored bridges between fields, tracking the instruments shaping the era, and reading what's emerging, honestly.

scroll ↓
01 — The Atlas

Every paper, placed.

Each point is one paper, embedded by meaning and clustered into topics. Search a term to light up matches; isolate a topic to see its territory.

02 — The Frontier

Where fields almost meet.

Topic pairs that sit close in meaning yet are rarely bridged in this corpus — candidate white space. Each pairs real evidence from both sides with a speculative, AI-proposed direction. “Rarely bridged” means within these 24k papers — not a claim of global novelty.

03 — The Instruments

The machines of the moment.

Facilities and missions tracked across the corpus. Toggle to compare — watch JWST rise after launch and the pulsar-timing arrays spike in 2023.

▲ Rising 2021→2023
▼ Fading
★ Most-cited in corpus
04 — The Pulse

What's emerging — honestly.

Acceleration finds real breakouts. It also fails in ways most "trend dashboards" hide. We show both.

✓ The signal that worked
#1

Pulsar timing arrays, caught before the surge.

Using only data through 2022, the pulsar-timing topic was the single fastest-accelerating topic in the field — and then surged 10.7× in 2023.

Real-world event: the NANOGrav 15-year data set (June 2023) — evidence for a nanohertz gravitational-wave background. The data flagged it before the announcement.

⚠ Why we don't sell a forecaster
−0.37

Across all topics, the same signal inverts.

Spearman correlation between pre-2023 acceleration and actual 2023 growth, over all 39 topics. Small topics mean-revert; the sample is thin.

Current top accelerators · full data