Issue #9 – The Other Half · Vienna Research Radar

Spotlight

A hurricane's heat exhaustion

Warm the sea and you get stronger hurricanes: the rule is intuitive, and broadly right. A tropical cyclone runs like a heat engine, drawing energy from the warm ocean surface and venting it high in the atmosphere, and its ceiling intensity is set by the temperature gap between those two ends. Warmer water raises the intake. But the same warming does something less obvious to the exhaust.

A study in Geophysical Research Letters from Caroline Muller's group at the Institute of Science and Technology Austria, with colleagues at Università degli Studi di Milano-Bicocca, works out the second half of that equation. In idealised simulations, first author Giousef Alexandros Charinti and colleagues track how warming near the top of the storm changes its outflow, and fold the effect into a modified version of the classic potential-intensity theory. The mechanism they isolate is specific: vertical advection near the tropopause is what warms the upper level, and a warmer outflow narrows the engine's working temperature gap, trimming the intensity that surface warming would otherwise deliver.

The sting is in the coupling. That upper-level brake is not fixed; it grows stronger as the sea surface warms. The same warming that stokes the intake also, through the storm's own circulation, heats the exhaust a little more, so the two effects rise together and partly offset.

Which is why "warmer seas, stronger storms" is a poor guide to how much stronger. The answer does not scale cleanly with sea-surface temperature; it turns on how the whole vertical temperature profile shifts, and on a competition the same warming sets off at both ends of the engine at once. The modified formula is a way to hold both ends in view.

→ Charinti, Davin, Polesello et al., Geophysical Research Letters, 2026

Radar Scans

As one cause for liver cancer recedes, another moves in

For decades, hepatocellular carcinoma in Europe tracked viral hepatitis: infect the liver, and cancer might follow. Vaccination against hepatitis B and a cure for hepatitis C have since pulled that cause down, and a European review co-led by Bernhard Scheiner at the Medical University of Vienna sets out what moved in to replace it. Alcohol and metabolic fatty-liver disease now drive a growing share of cases, often in people no surveillance programme was watching. The tumour has not gone away; its way in has changed, and screening built for the old map misses part of the new one.

→ Scheiner, Dekervel, Piscaglia et al., The Lancet Regional Health – Europe, 2026

The outlier was in the method

Strong coupling, a basic constant in physics, has up until now been one of the worst measured. One established method, reading it off the shape of the particle spray from electron–positron collisions, kept returning values that sat too low. A real anomaly, or a fault in the sums? A collaboration with André Hoang from University of Vienna, among its architects, with groups at Universidad de Salamanca and Massachusetts Institute of Technology and Harvard University, rebuilt the calculation. The outlier, it turned out, was in the arithmetic, not the physics.

→ Benitez, Bhattacharya, Hoang et al., SciPost Physics, 2026

Unearthing a Roman garrison without lifting a spade

Beside the old colonia at Saint-Bertrand-de-Comminges, in the French Pyrenees, a 2.7-hectare Roman military camp had been known only in fragments. Stefan Groh and @Helga Sedlmayer of the Austrian Academy of Sciences have now mapped it in its full glory with magnetics and radar: perimeter walls, projecting towers and inner buildings. The architecture dates it to the Severan era and ties it to legionary camps back in Noricum, at Enns and Albing.

→ Groh, Sedlmayer, Gladius, 2026

Funding Moves

European Research Council (ERC) Starting Grants 2026 announced

Twelve of the eighteen Austrian winners are based in and around Vienna, five at ISTA alone, each grant worth up to EUR 1.5 m. over five years: Valentina Ausserladscheider ( University of Vienna, the economics of climate-related loss), ESTER LAZZARI ( University of Vienna, mental health and reproductive behaviour), Daniela Kaufmann ( TU Wien, trustworthy automated reasoning), Jiří Pavelec ( TU Wien, robotic catalyst design), Ralf Schmidt ( Medical University of Vienna, direct cell reprogramming for therapies), Vlad Tereshenko (Medical University of Vienna, a bionic nerve interface for internal organs), Thomas Wolf ( Central European University, how vocal signals coordinate people), Denitsa Baykusheva (Institute of Science and Technology Austria, controlling collective quantum states), Jack Bravo (Institute of Science and Technology Austria, bacterial defences against plasmids), Ilaria Caiazzo (Institute of Science and Technology Austria, merging white dwarfs), Francesco Locatello (Institute of Science and Technology Austria, AI for causal analysis), and Charles Roques-Carmes (Institute of Science and Technology Austria, X-ray quantum imaging).

→ ERC Starting Grants 2026 results, European Research Council

WWTF Insight

This fortnight's Crossref data brought forward Science paper from the labs of Daniel Gerlich and Anton Goloborodko at IMP, IMBA - Part of Vienna BioCenter, credited to LS19-001. Textbooks draw the two sister chromatids, the identical copies made when a chromosome is duplicated, as aligned mirror images. Led by Flavia Corsi, the team shows they are not: the copies sit about 100 kilobases out of step, a built-in offset that is the physical fingerprint of an asymmetry already present in how DNA is copied. It matters because the sister chromatid is exactly what a cell reaches for to mend a broken strand. The template for repair is systematically misaligned, and that bears on how a genome stays stable.

Stray Signal

The cube that sends the LLMs spinning

Give someone a shape and a rotated copy and ask whether they match, and most of us solve it by turning the thing in our heads. Mental rotation is one of the oldest tasks in cognitive psychology. Jonas Lesigang and Jakob Pietschnig at the University of Vienna set the same puzzles to ChatGPT-5 and Gemini 3, with every prompting trick going, chain-of-thought, self-consistency, images fed in piece by piece, and compared them against 430 people. The people won, and not narrowly: across most tests the models scored between the 0th and 7th percentile of the human sample. Fluent in language as they are, they still cannot hold a cube in the mind's eye and turn it. There is a whole half of thinking the frontier has not reached.

→ Lesigang, Pietschnig, Journal of Intelligence, 2026