We present a loss function for neural networks that encompasses an idea of trivial versus non-trivial predictions, such that the network jointly determines its own prediction goals and learns to satisfy them. This permits the network to choose sub-sets of a problem which are most amenable to its abilities to focus on solving, while discarding 'distracting' elements that interfere with its learning. To do this, the network first transforms the raw data into a higher-level categorical representation, and then trains a predictor from that new time series to its future. To prevent a trivial solution of mapping the signal to zero, we introduce a measure of non-triviality via a contrast between the prediction error of the learned model with a naive model of the overall signal statistics. The transform can learn to discard uninformative and unpredictable components of the signal in favor of the features which are both highly predictive and highly predictable. This creates a coarse-grained model of the time-series dynamics, focusing on predicting the slowly varying latent parameters which control the statistics of the time-series, rather than predicting the fast details directly. The result is a semi-supervised algorithm which is capable of extracting latent parameters, segmenting sections of time-series with differing statistics, and building a higher-level representation of the underlying dynamics from unlabeled data.

Microlensing observations suggest that the mass distribution of free-floating planets (FFPs) follows a declining power-law with increasing mass. The origin of such distribution is unclear. Using a population synthesis framework, we investigate the formation channel and properties of FFPs, and compare the predicted mass function with observations. Assuming FFPs originate from planet-planet scattering and ejection in single star systems, we model their mass function using a Monte Carlo based planet population synthesis model combined with N-body simulations. We adopt a realistic stellar initial mass function, which naturally results in a large fraction of planetary systems orbiting low-mass stars. The predicted FFP mass function is broadly consistent with observation: it follows the observed power-law at higher masses (10 \lesssim m/M_\oplus < 10^4), while at lower masses (0.1 < m/M_\oplus \lesssim 10) it flattens, remaining marginally consistent with the lower bound of the observational uncertainties. Low-mass, close-in planets tend to remain bound, while Neptune-like planets at wide orbits dominate the ejected population due to their large Hill radii and shallow gravitational binding. We also compare the mass distribution of bound planets with microlensing observations and find reasonably good agreement with both surveys. Our model predicts $\simeq 1.20$ ejected planets per star in the mass range of 0.33 < m/M_\oplus < 6660, with a total FFP mass of $\simeq 17.98~M_\oplus$ per star. Upcoming surveys will be crucial in testing these predictions and constraining the true nature of FFP populations.

The internal model principle, originally proposed in the theory of control of linear systems, nowadays represents a more general class of results in control theory and cybernetics. The central claim of these results is that, under suitable assumptions, if a system (a controller) can regulate against a class of external inputs (from the environment), it is because the system contains a model of the system causing these inputs, which can be used to generate signals counteracting them. Similar claims on the role of internal models appear also in cognitive science, especially in modern Bayesian treatments of cognitive agents, often suggesting that a system (a human subject, or some other agent) models its environment to adapt against disturbances and perform goal-directed behaviour. It is however unclear whether the Bayesian internal models discussed in cognitive science bear any formal relation to the internal models invoked in standard treatments of control theory. Here, we first review the internal model principle and present a precise formulation of it using concepts inspired by categorical systems theory. This leads to a formal definition of ``model'' generalising its use in the internal model principle. Although this notion of model is not a priori related to the notion of Bayesian reasoning, we show that it can be seen as a special case of possibilistic Bayesian filtering. This result is based on a recent line of work formalising, using Markov categories, a notion of ``interpretation'', describing when a system can be interpreted as performing Bayesian filtering on an outside world in a consistent way.

In a classic paper, Conant and Ashby claimed that "every good regulator of a system must be a model of that system." Artificial Life has produced many examples of systems that perform tasks with apparently no model in sight; these suggest Conant and Ashby's theorem doesn't easily generalise beyond its restricted setup. Nevertheless, here we show that a similar intuition can be fleshed out in a different way: whenever an agent is able to perform a regulation task, it is possible for an observer to interpret it as having "beliefs" about its environment, which it "updates" in response to sensory input. This notion of belief updating provides a notion of model that is more sophisticated than Conant and Ashby's, as well as a theorem that is more broadly applicable. However, it necessitates a change in perspective, in that the observer plays an essential role in the theory: models are not a mere property of the system but are imposed on it from outside. Our theorem holds regardless of whether the system is regulating its environment in a classic control theory setup, or whether it's regulating its own internal state; the model is of its environment either way. The model might be trivial, however, and this is how the apparent counterexamples are resolved.

We have identified iridium in an ~5 m-thick section of pelagic sediment cored in the deep sea floor at Site 886C, in addition to a distinct spike in iridium at the K-Pg boundary related to the Chicxulub asteroid impact. We distinguish the contribution of the extraterrestrial matter in the sediments from those of the terrestrial matter through a Co-Ir diagram, calling it the "extraterrestrial index" fEX. This new index reveals a broad iridium anomaly around the Chicxulub spike. Any mixtures of materials on the surface of the Earth cannot explain the broad iridium component. On the other hand, we find that an encounter of the solar system with a giant molecular cloud can aptly explain the component, especially if the molecular cloud has a size of ~100 pc and the central density of over 2000 protons/cm^3. Kataoka et al. (2013, 2014) pointed that an encounter with a dark cloud would drive an environmental catastrophe leading to mass extinction. Solid particles from the hypothesized dark cloud would combine with the global environment of Earth, remaining in the stratosphere for at least several months or years. With a sunshield effect estimated to be as large as -9.3 W m^-2, the dark cloud would have caused global climate cooling in the last 8 Myr of the Cretaceous period, consistent with the variations of stable isotope ratios in oxygen (Barrera and Huber, 1990; Li and Keller, 1998; Barrera and Savin, 1999; Li and Keller, 1999) and strontium (Barrera and Huber, 1990; Ingram, 1995; Sugarman et al., 1995). The resulting growth of the continental ice sheet also resulted in a regression of the sea level. The global cooling, which appears to be associated with a decrease in the diversity of fossils, eventually led to the mass extinction at the K-Pg boundary.

Origins of life research is marred by ambiguous questions and goals, creating uncertainty about when research objectives have been achieved. Because of numerous unknowns and disagreements about definitions and theories, the field lacks clear markers of progress. We argue that the origins community should focus on goals that have agreed-upon meaning and can be consensually categorized as achieved or unachieved. The origins community needs these goals to maintain coherence amongst a federation of problems with the shared, but nebulous aspiration of understanding the origins of life. We propose a list of challenges with clear 'Finish Lines'--explicit descriptions of what will be achieved if each goal is reached--similar to the X-prize model. The intent is not to impose top-down research directions, but to compel the community to coalesce around explicit problems of the highest priority, as physics, astronomy, and planetary science communities do when setting science objectives for missions and megaprojects. Even if the generated phenomena are not unequivocally life-like, demonstrating systems that achieve these goals will sharpen the distinction between life itself and the constellation of phenomena that co-occur with life. This document was originally submitted as a whitepaper to the 2025 NASA-DARES (Decadal Astrobiology Research and Exploration Strategy) call for whitepapers (this https URL).

When planets receive insolation above a certain critical value called the runaway threshold, liquid surface water vaporizes completely, which forms the inner edge of the habitable zone. Because land planets can emit a large amount of radiation from the dry tropics, they have a higher runaway threshold than aqua planets do. Here we systematically investigated the runaway threshold for various surface water distributions using a three-dimensional dynamic atmosphere model. The runaway threshold for the meridionally uniform surface water distribution increases from the typical value for the aqua-planet regime (~130% S0) to one for the land-planet regime (~155% S0) as the dry surface area increases, where S0 is the present Earth's insolation. Although this result is similar to the previous work considering zonally uniform surface water distributions, the runaway threshold for the land-planet regime is quite low compared to that of the previous work. This is because a part of the tropical atmosphere is always wet for the meridionally uniform case. We also considered the surface water distributions determined by the Earth's, Mars' and Venus' topographies. We found that their runaway thresholds are close to that for the meridionally uniform cases, and the amount of water at the boundary between an aqua- and land-planet regime is around 10% of the Earth's ocean. This clearly shows that the runaway threshold is not determined uniquely by the luminosity of the central star, but it has a wide range caused by the surface water distribution of the terrestrial water planet itself.

Direct-imaging spectra hold rich information about a planet's atmosphere and surface, and several space-based missions aiming at such observations will become a reality in the near future. Previous spectral retrieval works have resulted in key atmospheric constraints under the assumption of a gray surface, but the effect of wavelength-dependent surface albedo on retrieval has not been shown. We explore the influence of the coupling effect of cloud and wavelength-dependent surface albedo on retrieval performance via modeling suites of Earth-like atmospheres with varying cloud and surface albedo parameterizations. Under the assumption of known cloud scattering properties, the surface spectral albedos can be reasonably recovered when the surface cover represents that of Earth-like vegetation or ocean, which may aid in characterizing the planet's habitability. When the cloud scattering properties cannot be assumed, we show that the degeneracy between the cloud properties and wavelength-dependent surface albedo leads to biased results of atmospheric and cloud properties. The multi-epoch visible band observations offer limited improvement in disentangling this degeneracy. However, the constraints on atmospheric properties from the combination of UV band (R $\sim 6$) $+$ visible band (R $\sim 140$) are consistent with input values to within 1 $\sigma$. If short bandpass data is not available, an alternative solution to reduce the retrieval uncertainties would be to have the prior constraints on planetary cloud fraction with less than 20% uncertainty.

We introduce a method by which a generative model learning the joint distribution between actions and future states can be used to automatically infer a control scheme for any desired reward function, which may be altered on the fly without retraining the model. In this method, the problem of action selection is reduced to one of gradient descent on the latent space of the generative model, with the model itself providing the means of evaluating outcomes and finding the gradient, much like how the reward network in Deep Q-Networks (DQN) provides gradient information for the action generator. Unlike DQN or Actor-Critic, which are conditional models for a specific reward, using a generative model of the full joint distribution permits the reward to be changed on the fly. In addition, the generated futures can be inspected to gain insight in to what the network 'thinks' will happen, and to what went wrong when the outcomes deviate from prediction.

The source of impactors capable of producing the cratering asymmetry on Triton is open to debate. While hyperbolic comets are the likely source of impactors, the theoretically predicted crater distribution due to impacts from comets is incompatible with the observed crater distribution on Triton. In this study, we adopted a dynamical approach to investigate the crater distribution produced by three sources of impactors: Neptune-centric impactors on prograde, low-inclination orbits, Neptune-centric impactors on high-inclination orbits such as its irregular satellites, and heliocentric impactors with hyperbolic orbits. We employ N-body simulations which take into account the dynamics of the impacts. We found that none of the impactor sources alone are able to reconcile with the crater distribution on Triton. The crater distribution produced by a low-inclination Neptune-centric source is narrowly concentrated on Triton's leading hemisphere whereas the heliocentric impactors produce too many craters on the trailing hemisphere compared to the observed crater distribution. All three impactor populations produce craters on the trailing hemisphere of Triton. We derived the impact distribution of a mixed population of impactors using a Monte Carlo approach and found that if the population consists of mainly heliocentric impactors (70%), the resulting impact distribution fits the observed crater distribution on Triton's leading hemisphere well.

Natural evolution gives the impression of leading to an open-ended process of increasing diversity and complexity. If our goal is to produce such open-endedness artificially, this suggests an approach driven by evolutionary metaphor. On the other hand, techniques from machine learning and artificial intelligence are often considered too narrow to provide the sort of exploratory dynamics associated with evolution. In this paper, we hope to bridge that gap by reviewing common barriers to open-endedness in the evolution-inspired approach and how they are dealt with in the evolutionary case - collapse of diversity, saturation of complexity, and failure to form new kinds of individuality. We then show how these problems map onto similar issues in the machine learning approach, and discuss how the same insights and solutions which alleviated those barriers in evolutionary approaches can be ported over. At the same time, the form these issues take in the machine learning formulation suggests new ways to analyze and resolve barriers to open-endedness. Ultimately, we hope to inspire researchers to be able to interchangeably use evolutionary and gradient-descent-based machine learning methods to approach the design and creation of open-ended systems.

The search for life in the universe is a major theme of astronomy and astrophysics for the next decade. Searches for technosignatures are complementary to searches for biosignatures, in that they offer an alternative path to discovery, and address the question of whether complex (i.e. technological) life exists elsewhere in the Galaxy. This approach has been endorsed in prior Decadal Reviews and National Academies reports, and yet the field still receives almost no federal support in the US. Because of this lack of support, searches for technosignatures, precisely the part of the search of greatest public interest, suffers from a very small pool of trained practitioners. A major source of this issue is institutional inertia at NASA, which avoids the topic as a result of decades-past political grandstanding, conflation of the effort with non-scientific topics such as UFOs, and confusion regarding the scope of the term "SETI." The Astro2020 Decadal should address this issue by making developing the field an explicit priority for the next decade. It should recommend that NASA and the NSF support training and curricular development in the field in a way that supports equity and diversity, and make explicit calls for proposals to fund searches for technosignatures.

Type II orbital migration is a key process to regulate the mass and semimajor axis distribution of exoplanetary giant planets. The conventional formula of type II migration generally predicts too rapid inward migration to reconcile with the observed pile-up of gas giant beyond 1 au. Analyzing the recent high-resolution hydrodynamical simulations by Li et al. (2024) and Pan et al. (2025) that show robust outward migration of a gas accreting planet, we here clarify the condition for the outward migration to occur and derive a general semi-analytical formula that can be applied for broad range of planet mass and disk conditions. The striking outward migration is caused by azimuthal asymmetry in corotation torque exerted from cicumplanetary disk regions (connecting to horseshoe flow) that is produced by the planetary gas accretion, while the conventional inward migration model is based on radial asymmetry in the torques from the circumstellar protoplanetry disk. We found that the azimuthal asymmetry dominates and the migration is outward, when the gap depth defined by the surface density reduction factor of $1/(1+K')$ is in the range of $0.03 \lesssim K' \lesssim 50$. Using simple models with the new formula, we demonstrate that the outward migration plays an important role in shaping the mass and semimajor axis distribution of gas giants. The concurrent dependence of planets' accretion rate and migration direction on their masses and disk properties potentially reproduces the observed pile-up of exoplanetary gas giants beyond 1 au, although more detailed planet population synthesis calculations are needed in the future.

The search for life beyond the Earth is the overarching goal of the NASA Astrobiology Program, and it underpins the science of missions that explore the environments of Solar System planets and exoplanets. However, the detection of extraterrestrial life, in our Solar System and beyond, is sufficiently challenging that it is likely that multiple measurements and approaches, spanning disciplines and missions, will be needed to make a convincing claim. Life detection will therefore not be an instantaneous process, and it is unlikely to be unambiguous-yet it is a high-stakes scientific achievement that will garner an enormous amount of public interest. Current and upcoming research efforts and missions aimed at detecting past and extant life could be supported by a consensus framework to plan for, assess and discuss life detection claims (c.f. Green et al., 2021). Such a framework could help increase the robustness of biosignature detection and interpretation, and improve communication with the scientific community and the public. In response to this need, and the call to the community to develop a confidence scale for standards of evidence for biosignature detection (Green et al., 2021), a community-organized workshop was held on July 19-22, 2021. The meeting was designed in a fully virtual (flipped) format. Preparatory materials including readings, instructional videos and activities were made available prior to the workshop, allowing the workshop schedule to be fully dedicated to active community discussion and prompted writing sessions. To maximize global interaction, the discussion components of the workshop were held during business hours in three different time zones, Asia/Pacific, European and US, with daily information hand-off between group organizers.

Under what circumstances can a system be said to have beliefs and goals, and how do such agency-related features relate to its physical state? Recent work has proposed a notion of interpretation map, a function that maps the state of a system to a probability distribution representing its beliefs about an external world. Such a map is not completely arbitrary, as the beliefs it attributes to the system must evolve over time in a manner that is consistent with Bayes' theorem, and consequently the dynamics of a system constrain its possible interpretations. Here we build on this approach, proposing a notion of interpretation not just in terms of beliefs but in terms of goals and actions. To do this we make use of the existing theory of partially observable Markov processes (POMDPs): we say that a system can be interpreted as a solution to a POMDP if it not only admits an interpretation map describing its beliefs about the hidden state of a POMDP but also takes actions that are optimal according to its belief state. An agent is then a system together with an interpretation of this system as a POMDP solution. Although POMDPs are not the only possible formulation of what it means to have a goal, this nevertheless represents a step towards a more general formal definition of what it means for a system to be an agent.

Massive stars born in star clusters terminate star cluster formation by ionizing the surrounding gas. This process is considered to be prevalent in young star clusters containing massive stars. The Orion Nebula is an excellent example associated with a forming star cluster including several massive stars (the Orion Nebula Cluster; ONC) and a 2-pc size H{\sc ii} region (ionized bubble) opening toward the observer; however, the other side is still covered with dense molecular gas. Recent astrometric data acquired by the Gaia satellite revealed the stellar kinematics in this region. By comparing this data with star cluster formation simulation results, we demonstrate that massive stars born in the ONC center were ejected via three-body encounters. Further, orbit analysis indicates that $\theta^2$ Ori A, the second massive star in this region, was ejected from the ONC center toward the observer and is now returning to the cluster center. Such ejected massive stars can form a hole in the dense molecular cloud and contribute to the formation of the 2-pc bubble. Our results demonstrate that the dynamics of massive stars are essential for the formation of star clusters and H{\sc ii} regions that are not always centered by massive stars.

The climates of terrestrial planets with a small amount of water on their surface, called land planets, are significantly different from the climates of planets having a large amount of surface water. Land planets have a higher runaway greenhouse threshold than aqua planets, which extends the inner edge of the habitable zone inward. Land planets also have the advantage of avoiding global freezing due to drier tropics, leading to a lower planetary albedo. In this study, we systematically investigate the complete freezing limit for various surface water distribution using a three-dimensional dynamic atmospheric model. As in a previous study, we found that a land planet climate has dry tropics that result in less snow and fewer clouds. The complete freezing limit decreases from that for aqua planets (92% S0, where S0 is Earth's present insolation) to that for land planets (77% S0) with an increasing dry area. Values for the complete freezing limit for zonally uniform surface water distributions are consistently lower than those for meridionally uniform surface water distribution. This is because the surface water distribution in the tropics in the meridionally uniform cases causes ice-albedo feedback until a planet lapses into the complete freezing state. For a surface water distribution using the topographies of the terrestrial planets, the complete freezing limit has values near those for the meridionally uniform cases. Our results indicate that the water distribution is important for the onset of a global ice-covered state for Earth-like exoplanets.

We report the discovery of one super-Earth- (TOI-1749b) and two sub-Neptune-sized planets (TOI-1749c and TOI-1749d) transiting an early M dwarf at a distance of 100~pc, which were first identified as planetary candidates using data from the TESS photometric survey. We have followed up this system from the ground by means of multiband transit photometry, adaptive-optics imaging, and low-resolution spectroscopy, from which we have validated the planetary nature of the candidates. We find that TOI-1749b, c, and d have orbital periods of 2.39, 4.49, and 9.05 days, and radii of 1.4, 2.1, and 2.5 $R_\oplus$, respectively. We also place 95\% confidence upper limits on the masses of 57, 14, and 15 $M_\oplus$ for TOI-1749b, c, and d, respectively, from transit timing variations. The periods, sizes, and tentative masses of these planets are in line with a scenario in which all three planets initially had a hydrogen envelope on top of a rocky core, and only the envelope of the innermost planet has been stripped away by photoevaporation and/or core-powered mass loss mechanisms. These planets are similar to other planetary trios found around M dwarfs, such as TOI-175b,c,d and TOI-270b,c,d, in the sense that the outer pair has a period ratio within 1\% of 2. Such a characteristic orbital configuration, in which an additional planet is located interior to a near 2:1 period-ratio pair, is relatively rare around FGK dwarfs.

We derive a single-cell level understanding of metabolism in an isogenic cyanobacterial population by integrating secondary ion mass spectrometry (SIMS) derived multi-isotope uptake measurements of Synechocystis sp. PCC6803 with a statistical inference protocol based on Liebig's law of the minimum, the maximum entropy principle, and constraint-based modeling. We find the population is structured in two metabolically distinct clusters: cells optimizing carbon yield while excessively turning over nitrogen, and cells which act reciprocally, optimizing nitrogen yield and excessively turning over carbon. This partition enables partial heterotrophy within the population via metabolic exchange, likely in the form of organic acids. Exchange increases the feasible metabolic space, and mixotrophic cells achieve the fastest growth rates. Metabolic flux analysis at the single-cell level reveals heterogeneity in carbon fixation rates, Rubisco specificity, and nitrogen assimilation. Our results provide a necessary foundation for understanding how population level phenotypes arise from the collective contributions of distinct individuals.

A fundamental goal of astrobiology is to detect life outside of Earth. This proves to be an exceptional challenge outside of our solar system, where strong assumptions must be made about how life would manifest and interact with its planet. Such assumptions are required because of the lack of a consensus theory of living systems, or an understanding of the possible extent of planetary dynamics. Here we explore a model of life spreading between planetary systems via panspermia and terraformation. Our model shows that as life propagates across the galaxy, correlations emerge between planetary characteristics and location, and can function as a population-scale agnostic biosignature. This biosignature is agnostic because it is independent of strong assumptions about any particular instantiation of life or planetary characteristic--by focusing on a specific hypothesis of what life may do, rather than what life may be. By clustering planets based on their observed characteristics, and examining the spatial extent of these clusters, we demonstrate (and evaluate) a way to prioritize specific planets for further observation--based on their potential for containing life. We consider obstacles that must be overcome to practically implement our approach, including identifying specific ways in which better understanding astrophysical and planetary processes would improve our ability to detect life. Finally, we consider how this model leads us to think in novel ways about scales of life and planetary replication.