Cosmic dust opens window on ancient atmosphere
Peer-Reviewed Publication
Updates every hour. Last Updated: 29-Jul-2025 23:11 ET (30-Jul-2025 03:11 GMT/UTC)
Goethe University Frankfurt has been successful with several proposals in the Hessian State Offensive for the Development of Scientific and Economic Excellence (LOEWE): the religion research project “DynaRel” will receive €19 million in funding over the next four years as the only LOEWE Center in the 18th funding round. The new LOEWE Research Cluster “Lipid Space,” which investigates the role of lipophilic substances in tissue self-regulation, will also receive €4.3 million over four years. Goethe University is also a partner in the medical technology LOEWE Research Cluster “MultiDrug-TDM”, led by TU Darmstadt, and will also receive funding in the LOEWE-Exploration line for a physics project on the structure of water in nanopores.
Diffusion probabilistic models (DPMs) have achieved impressive success in high-resolution image synthesis, especially in recent large-scale text-to-image generation applications. An essential technique for improving the sample quality of DPMs is guided sampling, which usually needs a large guidance scale to obtain the best sample quality. The commonly-used fast sampler for guided
sampling is denoising diffusion implicit models (DDIM), a first-order diffusion ordinary differential equation (ODE) solver that generally needs 100 to 250 steps for high-quality samples. Although recent works propose dedicated high-order solvers and achieve a further speedup for sampling without guidance, their effectiveness for guided sampling has not been well-tested before. In this work, researchers demonstrate that previous high-order fast samplers suffer from instability issues, and they even become slower than DDIM when the guidance scale grows larger. To further speed up guided sampling, researchers propose DPM-Solver++, a high-order solver for the guided sampling of DPMs. DPM-Solver++ solves the diffusion ODE with the data prediction model and adopts thresholding methods to keep the solution matches training data distribution. Researchers further propose a multistep variant of DPM-Solver++ to address the instability issue by reducing the effective step size. Experiments show that DPM-Solver++ can generate high-quality samples within only 15 to 20 steps for guided sampling by pixel-space and latent-space DPMs.
Scientists have been able to recreate the extreme conditions found on icy moons in deep space - and revealed the unstable behaviour of water.
- The HeH⁺ ion was the first molecule to form in the early universe, before the first stars were born.
- Researchers have investigated reactions of HeH+ with hydrogen atoms under conditions similar to those in the early universe for the first time, revealing significant discrepancies with previous theories.
- The results significantly impact our understanding of the early chemistry of the universe and the processes that contributed to the formation of the first stars.