Click here to go see the bonus panel!

Hovertext:
Also you can solve the halting problem by just checking the state at infinity.

Today's News:

Click here to go see the bonus panel!

Hovertext:
The weird part is that they're in a Burger King.

Today's News:

Click here to go see the bonus panel!

Hovertext:
The crank doesn't actually do anything, but it makes the humans happy.

Today's News:

Click here to go see the bonus panel!

Hovertext:
And when Babbage arrived the whole thing was actually running on a series of Pascalines.

Today's News:

Click here to go see the bonus panel!

Hovertext:
The key is to keep saying 'strictly speaking' over and over.

Today's News:

arXiv:2607.02360v3 Announce Type: replace-cross Abstract: Monocular spacecraft 6D pose estimation remains difficult under weak texture, thin structures, illumination variation, and occlusion. This article presents GAP-GDRNet, a geometry-aware RGB framework built on GDR-Net for a single-target synthetic spacecraft benchmark. The method strengthens the geometry-guided regression pipeline at two points. First, AFR is placed before dense geometric prediction to combine global structural attention with local weak-texture enhancement. Second, PGSA is inserted into Patch-PnP to relate downsampled geometric regions before final pose regression. Dense supervision is obtained from a Blender-based rendering and annotation process that provides masks, model-coordinate maps, camera intrinsics, and 6D pose labels. On the self-built spacecraft dataset, GAP-GDRNet achieves a rotation error of $1.96^\circ$, a translation error of 0.0165 m, and 95.16\% ADD@0.02 m, outperforming the reproduced GDR-Net baseline by 3.88 percentage points while running at 35.97 FPS. Tests on T-LESS and LM-O further show consistent gains over the reproduced baseline on textureless and occluded non-spacecraft objects.

arXiv:2606.25984v2 Announce Type: replace-cross Abstract: Large language models are increasingly deployed as investment research assistants, yet no benchmark tests whether they can accurately reconstruct and apply the specific procedural decision frameworks of expert investors. We introduce InvestPhilBench, a multi-layer benchmark spanning eight cognitive tiers, from principle identification (L1) to novel framework extrapolation (L8). The v0.6 release comprises 118 primary-source-verified principle cards, 25 decision-framework cards with explicit topology metadata, and 243 QA questions (197 dev / 46 held-out test). For reproducible scoring at scale we introduce the Benchmark Automated Scoring Pipeline (BASP), five algorithmic metrics, the Failure Mode Detection Protocol (FMDP) covering six failure modes, and Gate Reconstruction Accuracy (GRA), a per-gate metric for questions with gold reasoning programs. This release is primarily a benchmark-and-methodology contribution: its empirical study -- a four-model sanity wave on the 188-question development split (closed-book) -- is deliberately preliminary and stress-tests the metric design rather than ranking models. The wave shows a sharp provider-tier split (BASP 0.906 vs. 0.438), though these mixed-judge numbers are confounded upper bounds. The central methodological finding survives the caveat: the BASP composite saturates at the frontier (Claude L4 = 0.932) while GRA still exposes a procedural deficit (frontier L4 GRA ~0.77, L7 GRA 0.57-0.62) -- composite scoring rewards fluent prose and hides the procedural gap. On a 100-item expert-annotated gold set, the automated BASP composite tracks the human reference at Pearson r = 0.72 (MAE = 0.10). v0.6 also implements a unified judge and true model-in-the-loop retrieval/oracle conditions; the de-confounded multi-model leaderboard and full three-condition run are v1.0 deliverables.

From June 11 through July 19, NASA’s free interactive exhibit at FIFA Fan Festival™ brought space exploration to soccer fans in East Downtown Houston.  

More than 500,000 visitors explored the exhibit during the tournament, learning how the agency’s missions, research, and technology benefit life on Earth.  

The exhibit introduced a global audience to the Artemis program, Moon Base, the International Space Station, and the future of human space exploration.  

World Cup fans explored an Orion spacecraft exhibit, an International Space Station model, NASA’s Orion Crew Survival System suit, astromaterials, Hubble Space Telescope imagery, the Lunar and Mars Touch Table exhibit, a space-flown FIFA World Cup soccer ball, Space Center Houston activities, photo opportunities, and other hands-on experiences.  

Attendees also climbed into a lunar rover simulator to drive across a virtual lunar surface, explored a Moon Base display demonstrating how astronauts could live and work on the Moon during future Artemis missions, and examined lunar tools designed to support Artemis exploration.

The tournament welcomed special guests, as well. On June 20, Artemis II Commander Reid Wiseman and Pilot Victor Glover participated in World Cup activities as the stadium captains ahead of the Netherlands-Sweden match at NRG Stadium, where they received a standing ovation from more than 67,000 fans. 

While the weather prevented them from appearing on the FIFA Fan Festival main stage, the pair shared their experiences through media interviews.  

NASA also highlighted the connection between space exploration and soccer through a STEMonstration developed in collaboration with Adidas. Filmed aboard the space station, the demonstration shows how the same physics that govern motion in space also shape the game on Earth.  

Expedition 74 crew members kicked off FIFA Fan Festival with a prerecorded welcome message that aired on opening day. Throughout the tournament, the crew recorded greetings that played during matches and at Fan Festival events across the United States and Mexico. 

ESA (European Space Agency) astronaut Sophie Adenot joined the celebration by recording a “Viking Row” aboard the space station. ESA astronauts Andreas Mogensen and Marcus Wandt participated from the ground in a playful tribute connecting fans on Earth with life in orbit.  

After nearly 40 days of showcasing NASA’s missions to soccer fans from around the world, FIFA Fan Fest concluded with a final message from the Expedition 74 crew during the official closing ceremony. On the ground, NASA astronaut Mark Vande Hei took the Fan Festival stage to share his journey to becoming an astronaut, his first view of Earth from space, and how the focus required for a spacewalk mirrors the concentration soccer players need on the field. He then kicked the ceremonial soccer ball, bringing NASA’s participation in the festival to a close.  

“We could not have done this without the incredible NASA team who volunteered,” said Jessica Cordero, Johnson community engagement lead. “They connected with fans and shared the excitement of Artemis III and NASA’s return to the Moon. They are truly the best volunteer crew on the planet!” 

Explore NASA’s Exhibit 

Dr. Anthony Fauci, the face of America's COVID response, repeatedly invoked his Fifth Amendment right before a Senate committee on Wednesday. The hearing, chaired by Republican Sen. Rand Paul, was meant to probe the origins of COVID-19 and whether Fauci misled Congress and the public. Fauci called Paul's pursuit of him an "unhinged obsession." Ali Rogin reports. PBS News is supported by - https://www.pbs.org/newshour/about/funders. Hosted on Acast. See acast.com/privacy

A solar eclipse, the Perseids, bright Venus after sunset, and a deep partial lunar eclipse highlight August’s skywatching.

Skywatching Highlights

• Aug. 5: Last Quarter Moon
• Aug. 12: Total solar eclipse across northern Russia, Greenland, Iceland, northern Spain, and part of Portugal; partial eclipse in parts of the U.S.; New Moon
• Aug. 12-13: Perseid meteor shower peak under dark skies
• Aug. 14-16: Venus reaches greatest eastern elongation and shines low in the west after sunset
• Aug. 27-28: Partial lunar eclipse visible from much of North and South America and parts of Europe and Africa

Transcript

A solar eclipse, one of the year’s best meteor showers, Venus at its brightest in the evening sky, and a lunar eclipse to close out the month.

That’s “What’s Up” for August.

On Aug. 12, a total solar eclipse crosses northern Russia, Greenland, Iceland, and northern Spain, with a small corner of Portugal inside the path of totality.

In parts of the United States, from Alaska to North Carolina, the eclipse is partial. The Moon will take only a small bite out of the Sun, and the amount of coverage will vary with location.

Remember to watch safely. Use certified eclipse glasses or a safe solar viewer any time any part of the Sun is visible. Regular sunglasses are not safe. And never use binoculars, a telescope, or a camera without a solar filter made for the front of the optics.

Later that same night, the Perseid meteor shower will light up the sky, peaking the evening of Aug. 12 into the early morning hours of the 13th. And with a New Moon arriving on the 12th, the skies will be ideally dark.

The Perseids happen every year when Earth passes through a debris stream left behind by Comet Swift-Tuttle. As those tiny bits of comet dust hit our atmosphere at high speed, they burn up as bright streaks of light.

To view this meteor shower, look toward the northeast once it’s fully dark and watch for the constellation Perseus to clear the horizon. This is where the meteors originate, but let your eyes wander, because they can flash across any part of the sky.

For the best view, stay out late as the stars climb higher, find a dark open spot, and give your eyes 30 minutes to adjust.

Aug. 14-16, Venus reaches its greatest eastern elongation, which is its widest apparent separation from the Sun during this evening appearance.

Look low in the western sky shortly after sunset for the bright object that will outshine every star around it. Through a telescope after sunset, Venus will look close to half lit, like a tiny lunar phase.

On the night of Aug. 27, continuing into Aug. 28 for some time zones, the Full Moon slips through Earth’s shadow, resulting in a partial lunar eclipse. It will be visible from much of North and South America and parts of Europe and Africa.

At maximum eclipse, about 93% of the Moon’s diameter will be inside Earth’s dark central shadow, called the umbra. The Moon will not be completely covered, but it can look dramatically darkened, with a rusty, coppery tint along the covered edge.

Unlike a solar eclipse, a lunar eclipse is safe to watch with just your eyes. Binoculars or a small telescope can give you a closer view of Earth’s curved shadow moving across the Moon.

Here are the phases of the Moon for August.

You can stay up to date on all of NASA’s missions exploring the solar system and beyond at science.nasa.gov. I’m Raquel Villanueva from NASA’s Jet Propulsion Laboratory, and that’s What’s Up for this month.

Every spring in Palestine, Texas, the wide-open fields around NASA’s Columbia Scientific Balloon Facility fill with students and faculty from across Louisiana. They arrive carrying sensors, laptops, and carefully engineered payloads they have spent months preparing.  The goal is to send their experiments to the edge of space and return with meaningful data, while reflecting NASA’s long-standing commitment to develop future scientists and engineers.

NASA’s student programs have long served as an entry point for hands-on exploration, connecting students to the agency’s missions and giving them direct exposure to real aerospace environments. Among these initiatives, the Louisiana Aerospace Catalyst Experiences for Students, also known as LaACES, provides a robust undergraduate introduction to scientific ballooning.

Just after dawn on May 19, student-crafted instruments lifted off smoothly from the flight line, notching the 74th and 75th launches in a long-running collaboration between the facility and Louisiana Space Grant Consortium. The 2026 LaACES campaign consisted of 11 payloads that were designed and built by nine Louisiana university teams — Louisiana State University, Northwestern State, Southeastern Louisiana State, McNeese State, Loyola University, and Southern University, as well as and one high school team, St. Joseph’s Academy, during the weeklong event.

The teams’ scientific objectives included a wide of research, such as atmospheric science, cosmic ray detection, thermal management, ultraviolet characterization, stratospheric wind analysis, and solar cell performance.

Before arriving in Texas, students advanced their mission concepts through a structured sequence of design reviews modeled after NASA’s engineering lifecycle. From preliminary design to flight readiness, each team defended technical decisions, refined their payloads, and demonstrated their readiness for launch.

Once cleared for flight by program directors Doug Granger and Aaron Ryan, who oversee the LaACES initiative, the students shifted their work to Palestine, Texas, where the Columbia Scientific Balloon Facility supported the 2026 LaACES campaign. Facility personnel provided daily weather briefings, performed helium fills for both latex balloons, and offered launch-line guidance to ensure safe and successful operations. “We enjoy being able to support the students’ launches here and hope to help them cultivate a love for ballooning,” said Hugo Franco, operations manager at the balloon facility.

The two flight trains of associated student payloads, LACES-74 and LACES-75, on May 19 marked the start of the program’s first launch window. The balloons, capable of supporting 7- to 9-pound payloads, ascended to near-space altitudes and were monitored using various communication and tracking systems. Both missions completed their flights successfully and were recovered roughly 60 miles north of the launch site near Ben Wheeler, Texas.

“My most memorable experience during the course of this year was seeing the plots for the first time post flight,” said Savannah Matlock, a Louisiana State University student. “When we saw the data behave in the way we’d expected it to, and saw the science we were able to demonstrate, it makes it all worth it. It’s a great feeling to see all of your hard work pay off.”

Following recovery, student teams analyzed their sensor data and environmental measurements and then presented their findings to balloon facility engineers and technicians along with faculty mentors and peers.

Programs like LaACES mirror the operational environment of NASA’s broader Scientific Balloon Program. NASA scientific balloons are more than just “weather balloons”: The reality is far more sophisticated. NASA Columbia Science Balloon Facility supports the launch, tracking, and recovery of large scientific balloons capable of carrying advanced research instruments to the edge of space. These flights support investigations in astrophysics, atmospheric science, planetary research, technology demonstrations, and more.

By placing students in this ecosystem, LaACES acts as a bridge between academic learning and national research operations. Students witness firsthand how scientific experiments are prepared, integrated, launched, tracked, and recovered — experiences that can reshape career paths and open doors to future opportunities within NASA, academia, and aerospace industries.

Funded by NASA’s National Space Grant College and Fellowship program, LaACES is a statewide program of the Louisiana Space Grant Consortium. Since its inception, LaACES has supported more than 500 students across 13 higher‑education institutions, launching over 60 balloon flights and roughly 135 unique payloads. As NASA continues to advance scientific discovery, programs like LaACES help ensure a strong pipeline of future innovators. For these Louisiana students, watching their payload rise into the sky is more than a technical milestone — it is a defining moment, a spark that turns curiosity into possibility.

arXiv:2607.05583v2 Announce Type: replace-cross Abstract: Contemporary language models are dominated by the transformer architecture, which leverages self-attention mechanisms to enable more efficient, parallelized training across a wide set of documents and corpora. This has allowed transformers to effectively model data across a wide range of modalities and contexts. However, transformers, along with their conventional counterparts such as recurrent neural networks (RNNs) and convolutional neural networks (CNNs), often struggle to maintain efficiency when processing long contexts. We introduce ResonatorLM, a new mechanism that replaces attention with a physics-derived alternative. ResonatorLM treats token sequences as a single, driven one-dimensional latent field and replaces attention dot products with causal functions of damped resonators. We implement ResonatorLM on a traditional network architecture and test it on standard long-context modeling tasks. We find that in a small, 6M matched setting, training and prefill speedups increase with sequence length, decode speed reaches 6.47x compared to that of a standard, optimized transformer at 32K tokens, and accuracy reaches 61.31 percent (compared to 55.32 percent) on WikiText.

arXiv:2607.02615v2 Announce Type: replace-cross Abstract: Generating structured artifacts with Large Language Models - e.g.\ database queries, threat framework mappings, entity schemas - is relatively straightforward; however, making them reliable enough for production deployments presents challenges. We present TAG, a lightweight framework based on a core principle: \textit{LLMs generate, we validate}. This reframing shifts responsibility from generation quality to validation rigor. The framework rests on three key attributes: First, \textbf{test driven generation}: when tests fail, the LLM receives indicative error messages that expose why the output failed, enabling the LLM to understand its mistakes and refine subsequent attempts. Second, \textbf{deterministic and LLM-based tests}: deterministic tests catch heuristics that can be programmatically verified (schema, syntax, cross-reference), while LLM-based tests evaluate nuanced semantic and delicate features that resist programmatic inspection (intent alignment, logical consistency, domain correctness). Third, \textbf{expert-distilled judges}: LLM-based tests are calibrated to distill and replicate human expert decision distribution, transforming manual human quality gates into scalable, reusable evaluation proxies that reflect professional-grade validation standards. We demonstrate the framework on three artifact types in the security domain - KQL query generation, MITRE ATT\&CK mapping, and entity mapping - deployed in production at Microsoft Sentinel. We believe this framework can be applied beyond security to other artifact generation tasks, providing a path to reliable, high-quality outputs without sacrificing the efficiency gains of LLM generation.

NASA will hold a news conference at 2:30 p.m. EDT, Friday, Aug. 14, live from the agency’s Kennedy Space Center in Florida, and media and digital creators are invited to attend in person.

The announcement, held in the XLV Hangar at the Shuttle Landing Facility, will preview a new event at NASA Kennedy later this year tied to America’s 250th anniversary, showcasing American leadership in aviation, space exploration, and emerging technologies while bringing together the public, industry leaders, innovators, and the next generation of explorers.

Participants include:

• NASA Administrator Jared Isaacman
• NASA leadership
• Members of Congress

The agency will stream this news conference live through a variety of platforms:

https://www.nasa.gov/live

This event is open to U.S. media and digital creators. The request to attend must be received no later than 12 p.m. on Thursday, Aug. 6, to the Kennedy newsroom at: https://media.ksc.nasa.gov. NASA’s media accreditation policy is available online.

For more information about NASA’s missions, visit:

https://www.nasa.gov

-end-

George Alderman / Cheryl Warner
Headquarters, Washington
202-358-1600
george.a.alderman@nasa.gov / cheryl.m.warner@nasa.gov

Danielle Sempsrott
Kennedy Space Center, Fla.
321-298-8990
danielle.c.sempsrott@nasa.gov

arXiv:2607.03013v2 Announce Type: replace-cross Abstract: Images captured by consumer electronic devices, such as mobile phones and digital cameras, often suffer from low-light degradation due to sensor limitations and imaging pipelines, which degrades visual quality and affects downstream vision tasks. Existing methods based on Convolutional Neural Networks (CNNs) and Transformers have dominated current low-light image enhancement (LIE) due to their excellent ability to model hierarchical features. However, CNNs operate in local receptive fields that cannot model long-range dependencies, while Transformers overcome this problem but incur substantial computational costs. To address these challenges, we propose MambaLIE, a Scene Light Intensity-Boosted Low-Light Image Enhancement method based on a State Space Model (SSM). We first introduce scene light intensity to improve the structural distribution of illumination, which is then gated with the low-light input to guide enhancement. To better model the illumination while maintaining computational efficiency, we propose the Locally Enhanced State Space Model (LESSM) for efficient light enhancement. Our LESSM contains two branches: an SSM branch and a Local Enhanced branch, where the former is used to model the long-range dependencies with linear time complexity, while the latter is used to enhance local feature representations. Extensive experiments demonstrate that MambaLIE outperforms state-of-the-art CNN-based and Transformer-based LIE methods on four widely used synthetic benchmarks and five publicly available real-world benchmarks in terms of accuracy, speed, and model size, making it suitable for practical deployment on resource-constrained devices.

From serving in the U.S. Army to helping prepare astronauts to explore the lunar surface, Ike Theriot built a path to NASA through discipline and determination. 

Theriot serves as the Flight Operations Surface Testing lead in the Exploration Extravehicular Activity Operations Branch within the EVA, Robotics, and Crew Operations Division at NASA’s Johnson Space Center in Houston. He integrates flight operations testing needs across the Artemis program, ensuring teams have the data, procedures, and experience required to prepare astronauts to live and work on the Moon. He is currently transitioning from this role to focus on mission development, astronaut training, and mission execution for the next lunar landing. 

Theriot helps develop and execute simulated moonwalks at Johnson and at field locations across the U.S. and internationally. He also helped lead development of the Lunar Grid Reference System, designed specifically for human operations on the Moon and now incorporated into flight control systems. 

Theriot plans and executes testing in facilities such as NASA’s Neutral Buoyancy Laboratory and the Space Vehicle Mockup Facility. He often serves as a suited test subject and trains others to become test conductors and test leads, building the next generation of surface operations experts. 

In the Neutral Buoyancy Laboratory, the agency’s 40-foot-deep pool used to simulate microgravity, Theriot has supported International Space Station spacewalk rehearsals and lunar surface testing, helping refine procedures that astronauts will use in orbit and on the Moon. 

Theriot has supported testing of NASA’s government reference exploration spacesuit design and Axiom Space’s lunar spacesuit, the Axiom Extravehicular Mobility Unit, or AxEMU. In the Active Response Gravity Offload System, known as ARGOS, he practiced geologic sampling operations in the AxEMU, examining a collected sample before stowing it. The system offloads body weight to simulate lunar gravity, allowing teams to refine surface procedures and hardware performance before astronauts step onto the lunar terrain. 

He has also evaluated new lighting configurations during suited test runs supporting development of Artemis crew training curriculum, assessing how visibility affects safety and task execution in a simulated lunar environment. 

In addition, Theriot supports International Space Station spacewalks as a certified flight controller and was recently selected as a capsule communicator, or capcom. He has begun simulation training for that role, where he will serve as the voice between the Mission Control Center and astronauts in low Earth orbit. 

Theriot’s work extends far beyond NASA facilities, into remote environments that mirror the realities of lunar surface operations. During a field campaign at Kamestastin Crater in Labrador, Canada, Theriot trained alongside members of the Artemis II crew. Formed approximately 36 million years ago, the impact crater provides terrain similar to areas astronauts may encounter on the Moon and offers valuable geology training in a remote environment accessible only by boat. 

Theriot’s interest in space began while deployed overseas with the Army. In Iraq, he was tasked with setting up a satellite transceiver that failed on first power-up. After observing technicians perform field repairs, he began taking online engineering courses in his spare time, studying orbital mechanics, circuit design, cryptography, and space power systems. Later, while deployed to Afghanistan, he spent evenings stargazing and tracking the space station as it passed overhead. 

After leaving the Army, Theriot earned a mechanical engineering degree from the University of Houston and secured an internship at Johnson in 2016, rotating between engineering and flight operations teams. He accepted a full-time position in 2019, just as NASA announced plans to return astronauts to the Moon under the Artemis program. 

His most valued achievements involve setting others up for growth and success. Several years ago, Theriot created a Field Navigation Exercise at Challenger 7 Memorial Park in Houston to strengthen lunar surface navigation skills. As surface operations planning accelerated, he recognized many had limited field experience. He built the exercise and mentored others to lead it. 

More recently, he served as test conductor for a major lunar spacesuit test series in the Neutral Buoyancy Laboratory, spending months refining procedures and coordinating teams. On one occasion, the 19th Field Navigation Exercise and a Neutral Buoyancy Laboratory test run occurred on the same day. Both were led by people he had trained. Theriot took the day off. 

For him, that was success. 

Over time, his perspective on leadership has evolved. Early in his career, he believed the hardest challenges were technical. Now, he believes the greater challenge is providing the leadership and guidance that enable technical work to succeed.

For the mission to the Moon and beyond, Theriot hopes the agency demonstrates something larger than exploration alone. 

“Each day presents a new and interesting challenge, and I can feel myself growing to meet those challenges,” he said. “Through the Artemis missions, like the Apollo missions, I hope we can all show the world that impossible things can happen.” 

A new report is detailing the impact of cuts to USAID and the program known as PEPFAR, which has been credited with saving over 26 million lives since 2003. The study from the HIV/AIDS non-profit amfAR found that more than 1,700 treatment sites have closed in the wake of funding cuts. William Brangham discussed more with Dr. Atul Gawande, the assistant administrator of USAID under President Biden. PBS News is supported by - https://www.pbs.org/newshour/about/funders. Hosted on Acast. See acast.com/privacy

Firefighters are battling a series of blazes that began this weekend in Washington state and are continuing to burn in and around Spokane, the state's second most populous city. Tens of thousands of people have been displaced and hundreds of buildings have been destroyed. Stephanie Sy reports. PBS News is supported by - https://www.pbs.org/newshour/about/funders. Hosted on Acast. See acast.com/privacy

arXiv:2607.05061v2 Announce Type: replace-cross Abstract: Key-value (KV) cache growth is a major bottleneck in autoregressive decoding, as memory and bandwidth scale linearly with context length. Existing KV eviction methods often rely on static heuristics or proxy scores, which poorly track future token utility and cause brittle eviction as relevance shifts. To address this, we introduce KVpop, which learns a fixed-budget KV eviction policy by directly supervising the keep-or-drop decision. The scorer is trained against a novel future-attention target, computed efficiently without materializing dense attention maps. We further introduce a delayed memory-based scorer that, uniquely among learned eviction methods, defers scoring for a fixed number of steps to exploit near-future context. On AIME and HMMT mathematical reasoning, KVpop retains 98% of full-attention performance on Qwen3-4B at 75% KV cache compression and 97% at 88% compression, consistently outperforming established eviction baselines. Qwen3-8B shows even stronger results, reaching near-full teacher performance. These results show that supervising eviction with future-attention signals cuts memory costs while maintaining quality.

The longest total solar eclipse visible from land until 2114 will sweep across Spain and North Africa on August 2, 2027. Here’s how to choose where to watch it.

The post Plan for Next Year's Total Solar Eclipse: August 2, 2027 appeared first on Sky & Telescope.

NASA delivered the NavCube3-mini payload on July 13 to Intuitive Machines for integration into Altus-1, the company’s first lunar relay satellite, marking an important milestone in the development of future lunar communications and navigation services. The lunar relays are designed to enable communications and navigation for astronauts and rovers operating at the agency’s future Moon Base.

About half the size of a shoebox and weighing just 3.5 pounds, NavCube3-mini is a compact but powerful navigation receiver designed to use signals from Earth-based GPS and Galileo Global Navigation Satellite Systems (GNSS) at lunar distances. Operating on less than 20 watts of power, roughly the same as a laptop computer, it can determine a spacecraft’s precise position far beyond Earth orbit. The compact payload builds on a series of navigation technology advancements developed at NASA’s Goddard Space Flight Center in Greenbelt, Maryland, each extending GPS navigation to new record-breaking distances from Earth.

The payload will fly aboard Intuitive Machines’ Altus-1 lunar relay satellite, the first of a planned network of lunar relay satellites being developed under the company’s Near Space Network Services contract with NASA. The relays will provide communications and navigation support for missions operating at the Moon, including the challenging lunar South Pole region, where Artemis astronauts will land in 2028, and where direct communications with Earth can be difficult. By extending communications coverage and improving navigation services, the relay network will help realize NASA’s vision for a sustained human presence on the lunar surface.

Before being shipped to Intuitive Machines, NavCube3-mini underwent an extensive environmental and performance test campaign at NASA Goddard to verify it is ready for spaceflight. The environmental testing included vibration testing to simulate launch conditions, thermal vacuum testing in the extreme temperatures and vacuum of space, and electromagnetic compatibility testing to ensure the payload can operate reliably alongside other spacecraft systems without causing or experiencing electromagnetic interference. Performance testing was conducted before and after each environmental test using high-fidelity simulations of the GPS and Galileo signals the NavCube will encounter in lunar orbit, verifying functionality and performance throughout the testing campaign. 

NavCube3-mini will serve as a key technology demonstration aboard Altus-1, validating the use of GNSS-based navigation in the lunar region and providing valuable performance data to support the development of future lunar navigation infrastructure. This technology is part of NASA’s broader strategy to develop communications and navigation services that work across both commercial providers and NASA’s networks. These capabilities are designed to support a growing lunar ecosystem that includes orbiters, landers, rovers, and, eventually, astronauts living and working on the Moon.

The delivery of NavCube3-mini marks another step toward building the communications and navigation infrastructure needed for long-term lunar exploration. Through partnerships with commercial providers like Intuitive Machines, NASA is building a more connected and capable lunar environment. As activity around the Moon continues to grow, these capabilities will enable lunar spacecraft and explorers to operate more safely, efficiently, and autonomously.

arXiv:2607.06655v2 Announce Type: replace-cross Abstract: In this report, we present Pelican-VLA 0.5, a unified VLA model that integrates vision-language understanding, future-frame generation, and action prediction within a single architecture. Pelican-VLA 0.5 achieves attention-level generalization: without object annotations, segmentation masks, attention supervision, or task-specific fine-tuning, its action pathway already focuses on the manipulation-relevant object and contact region. This behavior persists across unseen scenes and unseen robot embodiments, and is substantially stronger than in other open-source VLA baselines. We verify that this ability originates from the learnable Bottleneck Token inserted between perception and action: by routing task-relevant visual information through a compact bottleneck, the tokens interface induces manipulation-centric attention during pre-training and remains effective across different policy structures, including a MoT-style architecture.

arXiv:2607.04983v2 Announce Type: replace-cross Abstract: This article is about the development of a fuzzy cognitive map using a local large language model. In the light of recent advances it is evident that large language models, and even local large language models are capable of extracting quantities from textual data. In other words, a local LLM like Qwen2.5-32B, or probably larger, can accept entities as prompt input and determine relevant quantitative data as the model output. In turn, this output can be utilized for the construction of a data driven fuzzy cognitive map. Hence, this implementation is achieved and then the model is thoroughly tested; Qwen2.5-32B is used and the data is extracted from hotel reviews from TripAdvisor. Furthermore, the extracted documents pass through the model unfiltered and then a fuzzy cognitive map is trained and evaluated. A case is made about Greek reviews where a star topology FCM is formed that indicates the preferences of the reviewers. Finally, external validation is performed to establish whether the fuzzy cognitive map can correlate the star rating of the review -an outcome outside the model's inference scope -with its predicted satisfaction.

arXiv:2606.00329v2 Announce Type: replace-cross Abstract: Recursive systems can enter collapse-like regimes -- self-reinforcing amplification, persistent recursion, and narrowing diversity that mask accelerating internal degradation -- before overt failure becomes visible. We introduce Loopzero, a claim-bounded benchmark framework for testing whether recursive failures follow a directional telemetry pattern: rising gain (G), recursive persistence (p), and declining diversity ($\delta$). The claim boundary is specified in Lean; the Lean artifact does not verify real telemetry, benchmark validity, or detector performance. We evaluate the bridge on two frozen public-artifact benchmarks: a segmented public-markets benchmark (Volmageddon 2018, COVID MWCB 2020) and a MovieLens-25M offline deterministic recommender replay. Detectors are evaluated under a locked equal-false-positive contract (FP $\in$ [0.03, 0.07], pre-registered) so all configurations face the same alert budget. Neither tested standard comparators nor Loopzero's pre-registered quantile detector achieved an accepted operating point. Directional witness alignment held on both canonical benchmarks, with adjacent-horizon and row-level limitations disclosed. Digitized Shumailov et al. (2024) LLM training-loop trajectories are directionally consistent with the pattern; matched-FP evaluation in that domain is deferred. The contribution is a reproducible, falsifiable benchmark framework for evaluating recursive-collapse warning claims under an explicit alert-budget contract -- non-acceptance reported as a first-class scientific outcome.

arXiv:2607.02885v2 Announce Type: replace-cross Abstract: Cognitive Behavioral Therapy (CBT) provides a structured framework for understanding a user's mental state by examining the interaction between cognitive and behavioral factors. However, out-of-the-box LLMs respond fluently and empathetically, yet collapse into validation & reflection, regardless of what the user actually needs. They know theoretical CBT (scoring up to 96% accuracy on licensing exam questions) but fail to apply it effectively. We explore this gap with a knowledge-guided framework that treats CBT dialogue as controlled affective reasoning: user narratives are decomposed into Beck's Cognitive Conceptualization structure, grounded in clinical SNOMED CT concepts validated via Natural Language Inference, and a Multiple Chain-of-Thought (MCoT) strategy selection between Validation & Reflection, Socratic Questioning, or Alternative Perspectives. To measure whether such guidance actually changes behavior, we introduce the Protocol Leverage Force (F), a behavior-level metric that captures how far an intervention shifts a model away from its default response. Across three open-weight LLMs and 14 RealCBT-derived case studies, evaluated with human experts, valence-arousal trajectories, and linguistic entrainment, F shows that simply introducing protocol definitions via single chain-of-thought prompting fails to change LLM behavior, while MCoT on these definitions guides strategy selection better. Still, the effect stays within 1% (approx. 1.2-1.3%), and all models remain biased toward Validation & Reflection. These results show CBT knowledge alone does not ensure effective application, giving the affective-computing community instrumentation to measure where LLMs fall short.

The average person isn't equipped to fact-check every video in their feed.

Prime Video is looking for a replacement for Ryan Hurst, who was initially cast as Kratos but dropped out of the series after an on-set injury.

null

In a presentation by the State Department, a generated map provides all the highlighted countries mislabeled. Jessica Donati reports for Reuters:

The State Department said ​it took “full responsibility” for the confusion caused and that the map had been produced by a team member who ​hastily changed the slide deck before the event.

The map showed Nigeria, where the U.S. currently has several hundred troops deployed, as landlocked in the Sahara Desert. Mozambique, which is in southeastern Africa, was relocated to the Horn of Africa, while Ivory ​Coast in West Africa was placed on the other side of the continent.

Trust the tools blindly and in haste. Regret it later.

Tags: generative, presentation, Reuters

With the new Spider-Man movie's opening now in the record books, the top boss at Marvel is already looking to the future.

arXiv:2606.16567v2 Announce Type: replace-cross Abstract: Neural ordinary differential equations (neural ODE) gained attention in safety critical settings such as continuous-time controllers for cyber-physical systems and classifiers integrated into automated decision pipelines, raising the question whether their behavior can be formally verified. Existing tools dedicated to neural ODE provide only a single reachability call without iterative input-set refinement, limiting the precision of their verdicts to whatever one reachability call can deliver. We present TNODEV, the first formal verifier for neural ODE that integrates a falsification checker, a fast interval-based reachability backend based on continuous-time mixed monotonicity, a verification and refinement loop with three input-set splitting heuristics, and a parallel scheduler in a single end-to-end pipeline. TNODEV supports safe-set inclusion verification on pure neural ODE, neural ODE in closed loop with a neural network controller and general neural ODE (GNODE), with the safe set specified either as an interval or as the half-space intersection induced by a target classification label. We evaluate TNODEV on a range of benchmarks across safe-set inclusion and classification-robustness properties, including a direct reachability comparison against NNV 2.0 and CORA and a verification comparison against NNV 2.0 on MNIST general neural ODE classifiers.

An AI analysis of 1.2 million satellite images shows that floating algae blooms are spreading across the global ocean. The rapid growth of massive seaweed mats could reshape marine habitats while bringing foul-smelling decay, ecological damage, and economic losses to coastlines.

On May 20, 2026, OpenAI made an announcement that shook the mathematical world. An internal AI model — one not available to the public — had come up with a counterexample to the “unit distance” problem, a conjecture made in 1946 by Paul Erdős, the prolific, itinerant Hungarian mathematician. Erdős posed thousands of questions, but this one was special: It was both simple to explain and…

Source

Jacob Tsimerman knows that many people see mathematics as a search for beauty. Over the course of his career, he has encountered beauty many times. But beauty is not what drives him. What he loves most is to solve hard problems, and to be the first to do so. “In my experience, and I think many people’s experience, math is very much a goal-oriented endeavor,” he said. “The truth-and-beauty stuff…

Source

It’s unclear when (and if) it will be publicly released.

Pseudoscience requires apathy about its victims, and that's the most grotesque part about it.

The post In 2014, Dr. Kelly Brogan and Sayer Ji Told People Natural HPV Infections Were “Beneficial, Not Deadly.” What’s Happened Since? first appeared on Science-Based Medicine.

null