Search for: All records

Abstract Although trionychians have a rich fossil record, much of their fossil diversity is known from the Cretaceous and Paleogene, and little is known about their evolutionary history in the Neogene. We here describe cranial and shell material of trionychians from the Early Miocene Moghra Formation of Egypt that we attribute to a new carettochelyid taxon,Allaeochelys meylanisp. nov., and to theTrionyxlineage.Allaeochelys meylanisp. nov. fills a temporal gap between previously described taxa and exhibits a series of unique features, including greatly thickened cranial bones, a broad bony wall posterior to the orbit, a large fossa formed by the maxilla and premaxilla at the anterior third of the triturating surface, and a medial process on peripheral II.Allaeochelys meylanisp. nov. also documents the oldest occurrence ofCarettochelyidaeon the Afro-Arabian continent, while theTrionyxmaterial reported herein provides unambiguous evidence for the presence of this lineage on the Afro-Arabian continent no later than the Early Miocene. 

Connecting two surface-code patches may require significantly higher noise at the interface. We show, via circuit-level simulations under a depolarizing noise model with idle errors, that surface codes remain fault tolerant despite substantially elevated interface error rates. Specifically, we compare three strategies—direct noisy links, gate teleportation, and a CAT-state gadget—for both rotated and unrotated surface codes, and demonstrate that careful design can mitigate hook errors in each case so that the full code distance is preserved for both 𝑋 and 𝑍. Although these methods differ in space and time overhead and performance, each offers a viable route to modular surface-code architectures. Our results, obtained with stim and pymatching, confirm that high-noise interfaces can be integrated fault-tolerantly without compromising the code's essential properties, indicating that fault-tolerant scaling of error-corrected modular devices is within reach with current technology. 

Understanding the Holocene environmental and climatic changes in the area around the Nile Delta and how they correlate with human occupation patterns can provide valuable information on natural changes in the climatic system and modern environmental dynamics in addition to enabling the prediction of future climatic trends. There are limited high-resolution studies carried out on the desert regions of the Nile, even though the Nile has been a prosperous record of interaction between civilizations, desert, and the river system over a prolonged period of time. Two cores have been recovered from the vicinity of two lakes in the Wadi El-Natrun area, western Nile Delta, Egypt. Samples were subjected to XRF analysis at 5 cm intervals. Preliminary results indicate a major shift in the concentration of key elements like silicon (Si), calcium (Ca), iron (Fe), magnesium (Mg), potassium (K), and aluminum (Al) indicating pronounced lithofacies changes throughout the studied cores. These changes are related to the combined effects of changing climate and the subsequent change in the depositional environment. Pending carbon-14 dating results will enable the determination of exactly when such lithofacies shifts happened and possibly linking them to known global paleoclimatic events. These data pave the way for a more wholistic understanding of the Holocene climate and environment of the Nile Delta region. 

We introduce VoiceCraft, a token infilling neural codec language model, that achieves state-of-the-art performance on both speech editing and zero-shot text-to-speech (TTS) on audiobooks, internet videos, and podcasts. VoiceCraft employs a Transformer decoder architecture and introduces a token rearrangement procedure that combines causal masking and delayed stacking to enable generation within an existing sequence. On speech editing tasks, VoiceCraft produces edited speech that is nearly indistinguishable from unedited recordings in terms of naturalness, as evaluated by humans; for zero-shot TTS, our model outperforms prior SotA models including VALLE and the popular commercial model XTTS-v2. Crucially, the models are evaluated on challenging and realistic datasets, that consist of diverse accents, speaking styles, recording conditions, and background noise and music, and our model performs consistently well compared to other models and real recordings. In particular, for speech editing evaluation, we introduce a high quality, challenging, and realistic dataset named RealEdit.