Recent Additions: New and Updated Entries in the Past 60 Days
Displaying entries 46 through 60 of 126 found.

Ecosmomics: Independent Complex Network Systems, Computational Programs, Genetic Ecode Scripts

Gabrielli, Andrea, et al. Network Renormalization. arXiv:2412.12988. Enrico Fermi Research Center, Rome, IMT School for Advanced Studies, Lucca, University of Leiden and Universitat de Barcelona physicists including Ángeles Serrano begin to methodically scope out how this reliable physical approach can now be effectively applied to life’s many complex network vitalities, which has mostly eluded prior success. Their contribution so far involves a new informational content and the presence of chimeras and criticalities.

Renormalization group (RG) theories were developed to describe system configurations with many degrees of freedom, along with the associated model parameters and coupling constants. They also can identify critical points of phase transitions. Usually, the RG builds on the notions of homogeneity, symmetry, geometry and locality to define metric distances, scale transformations and self-similar coarse-graining. However, the strong heterogeneity of real-world networks complicates renormalization procedures. In this review, we discuss past attempts, the important advances, and the ochallenges on the road to network renormalization. (Excerpt)

Millan, Ann, et al. Topology shapes dynamics of higher-order networks. Nature Reviews Physics. February, 2025. System physicists in Spain, Sweden, Japan, the USA, UK, Belgium and Germany including Filippo Radicchi and Ginestra Bianconi add a further finesse to our Earthuman understandings of nature’s reticulate anatomy and metabolism which can apply to hyper intricate phases of world weather and deep neural learnings.

Higher-order networks capture the many-body interactions present in complex systems. The new theory of topological dynamics can enhance our understanding of such areas as climate phenomena and AI algorithms. It encodes the dynamics of a network through topological signals assigned not only to nodes but also to edges, triangles and cells. Recent findings show that topological signals lead to the emergence of distinct types of dynamical state and collective phenomena including pattern formation and percolation. These results offer insights into how topology shapes dynamics and how dynamics learns topology. (Excerpt)

Zhang, Zhang et al. Coarse-graining network flow through statistical physics and machine learning.. Nature Communications. 16/1605, 2025. We cite this entry by Beijing Normal University, Indiana University and University of Padua theorists including Manlio De Domenico as an example of new abilities to root complex system phenomena in deep physical substrates by way of an AI assistance.

Information dynamics plays a crucial role in complex systems from cells to societies. Recent advances in statistical physics have been able to find key network properties but large system sizes have computational issues. We use graph neural networks to identify coarse-graining groups to achieve a low computational complexity for practical applications. Our method offers multiscale compression perspective that preserves information flow in biological, social, and technological networks better than other methods mostly focused on network structure. (Excerpt)

, . Stepney, Susan. Physical reservoir computing: a tutorial. Natural Computing. November 2024.. Natural Computing.. November, 2024. The University of York computer scientist (search) provides a latest succinct explanation of this increasingly popular procedure especially as quantum versions become available. See, for example, A Reservoir-based Model for Human-like Perception of Complex Rhythm Pattern by Zhongju Yuan, et al at arXiv:2503.12509.

This tutorial covers physical reservoir computing which first defines what it means for a physical system to compute, rather than evolve under the laws of physics. It describes the underlying computational Echo State Network (ESN) model, and explains why the it is suitable for direct physical implementation. The entry then describes how to characterise a physical reservoir in terms of benchmark tasks, and task-independent measures, along with optimising configuration parameters, and exploring the space of potential configurations. (Excerpt)

Reservoir computing is derived from recurrent neural network theory that maps input signals into dimensional spaces through a non-linear system called a reservoir The first key benefit is that training is performed only at the readout stage. The second is that the computational power of natural systems, both classical and quantum, can reduce the relative cost.

Flamm, Christoph, et al. Computation in chemical graph rewriting networks. Journal of Physics: Complexity. 6/1, 2025. CF and Peter Stadler, University of Vienna and Daniel Merkle, Algorithmic Cheminformatics Group, Bielefeld University discuss perceptive ways to investigate and identify the computational capabilities of ‘constructive’ chemistry.

transformations underlying the turn-over of their molecular components. In chemical reaction networks, computation may refer to two main aspects: concentrations of molecules, and molecular structures. The latter can be modeled by a chemical rewriting system acting on structural formulae, i.e. labeled graphs. We investigate graph rewriting and show that it can emulate Turing machines. and the computational capabilities of ‘constructive’ chemistry. (Excerpt)

Sidl, Leonhard, et al. Computational complexity, algorithmic scope, and evolution. Journal of Physics: Complexity. 6/1, 2025. University of Vienna and University of Leipzig bioinformatic researchers including Peter Stadler consider better ways by which life’s metabolic processes can be perceived and understood in terms of program operating systems. See also Computation in chemical graph rewriting networks by Christoph Flamm, et al in the same issue.

Biological systems are widely regarded as performing computations. Here we explore the idea that evolution confines biological computation to subsets of instances that can be solved with algorithms that are 'hardcoded' in the system. We use RNA secondary structure prediction as a developmental program to show that the salient features of the genotype–phenotype map remain intact even if 'simpler' algorithms are employed that correctly compute the structures for small subsets of instances. (Abstract)

Biological systems are often perceived to perform computations that solve complex problems at low computational cost as an evolutionary adaptation such as the central ‘information metabolism’ of a cell, i.e. DNA replication, transcription, and translation of RNA to proteins. Beyond these transformations of encoded information processing in neural systems, it is often unclear what exactly a biological system is computing. (1)

Returning to the question whether the concepts developed in theoretical computer science to describe computational complexity can also apply to computation in biological systems, we arrive at an affirmative answer. (11)

Koplenig, Alexander, et al. Human languages trade off complexity against efficiency.. PLoS Complex Systems.. February, 2025. • Over 42 pages and 150 references. Leibniz Institute for the German Language, Mannheim system linguists demonstrate how the historic corpora of spoken and enscribed Rosetta-like cipher narratives can indeed be seen to exhibit nonlinear, systematic, network themes and schemes.

• From a cross-linguistic perspective, language models are worthwhile because they can be trained on volumes of linguistic input. In this paper, we study different versions from statistical to neural networks from a database of 3 billion words across 6,500 documents in over 2,000 languages. We use the trained models to estimate entropy rates and a complexity measure derived from information theory. To compare entropy rates we use a machine learning approach to account for both language- and document-specific traits, as well as phylogenetic and geographical relationships. We then confirm by systematic differences in entropy rates, i.e. text complexity, across many corpora. (Excerpts)

Monakhov, Sergei and Holger Diessel. Complex Words as Shortest Paths in the Network of Lexical Knowledge.. Cognitive Science. 48/11, 2024. Friedrich-Schiller University, Jena system linguists carry out a latest, comprehensive analysis of the English language to show how it is wholly characterized by complex network topologies and emergent behaviors. See also Composition as Nonlinear Combination in Semantic Space: A Computational Characterization of Compound Processing by Tianqi Wang and Xu Xu in this Journal (49/2, 2025) for similar findings in Chinese script. In regard, an extensive 2025 verification of this deeper, common ecode, textual dimension is again achieved, which then by turns implies a natural literary narrative.

Lexical models diverge on how to represent complex words. Under the morpheme-based approach, each morpheme is treated as a separate unit, while in the word-based methods, morphological structure is derived from complex words. In this paper, we propose a computational model for word-based networks to view how complex words are learned, stored, and processed. Our study shows that complex words can be segmented into morphemes through the shortest pathway and novel terms are often formed along optimal paths. Our empirical results are tested by a usage-based grammar which reveals that network science provides a deep language structure. (Excerpt)

ity of the network. In conclusion, network science provides a powerful framework for analyzing language. In this paper, we have focused on central aspects of morphological productivity. However, if we think of language as an encompassing network, the network approach can also be applied to many other phenomena in phonology, morphology, and syntax. This approach is consistent with the way psychologists and neuroscientists analyze the human mind and brain and resonates with the emergentist view of grammar. (28)

Arroyo, Jose. et al. Arroyo, Jose, et al. Toward a General Theory for the Scaling and Universality of Thermal Responses in Biology.. arXiv:2503.05128. Santa Fe Institute system theorists including Pablo Marquet, Christopher Kempes, and Geoffrey West post a chapter for the forthcoming volume Scaling in Biology: A New Synthesis from SFI Press. The technical paper goes on to discern still another instance of innate, recurrent self-similarities with regard to energetic gradients in metabolisms.

We developed a theory showing that under appropriate normalizations and rescalings, temperature response curves show a remarkably regular behavior and follow a general, universal law. The impressive universality of temperature response curves remained hidden due to curve-fitting models not well-grounded in first principles. In addition, this framework can help explain the origin of thermal scaling relationships in from biology to ecosystems. Here, we summarize the background, predictions, implications, and extensions of this theory. (Abstract)

Importantly, our framework can be used for predicting scenarios of global warming, disease spread, and industrial applications. It provides a general equation that can be integrated into theoretical ecology and evolution, such as Major Transitions in Evolution. It also allows us to better understand the impacts of climate change at global scales, whereby mutation rates and mortality of viruses will likely increase, given their convex temperature response curves. (16)

Deco, Gustavo, et al. Deco, Gustavo, et al. Complex harmonics reveal low-dimensional manifolds of critical brain dynamics. Physical Review E. 111/014410, January, 2025. Universitat Pompeu Fabra, Barcelona and Oxford University open another window to view a neural-like nature which evolves and proceeds to attain a twintelligence (herein a reciprocal poise) and effective cognizance by way of this inherent self-organized complementarity and familiarity. We also note that the paper appears in a traditional physics journal as the two realms grow together and reunite as one.

The brain needs to perform time-critical computations to ensure survival, for which nonlocal, distributed computation at the whole-brain level make possible by self-organized criticality. These responses accord with Schrödinger's wave equation, so as to form a complex harmonics decomposition (CHARM) framework to express the complex network dynamics that are the key computational engines of critical brain dynamics. (Excerpt)

Korbel, Jan, et al.. Microscopic origin of abrupt mixed-order phase transitions. Nature Communications. 16/2628, 2025. Veteran system theorists JK and Stefan Thurner, Complexity Science Hub, Vienna and Shlomo Havlin, Bar-Ilan University, Israel extend their 21st century studies to wider and deeper perceptions o an inherent natural criticality at each and every turn. Into April, many entries like this are now reaching a critical number which strongly implies a phenomenal independent existence of a universal mathematic source code that exemplifies itself in a genetic sense everywhere.

We suggest a possible origin for abrupt mixed-order transitions in physical systems by way of three Ising interaction models. We identify a microscopic origin driven by long-term cascades of changes. We calculate the critical exponents for the cascading, magnetization, convergence, and fluctuations of single-trajectory critical temperature. Our findings can shed light on the microscopic mechanisms behind many abrupt transitions in nature and technology. (Excerpt)
n this paper, we studied the origin of mixed-order phase transitions in the case of the Ising model with three types that change the interaction network. We studied three models: (i) the Ising model with molecule formation, (ii) the Potts model with hidden states, and (iii) the Truncated Ising model. In each of them, the their interactions changed the order of the phase transition from a second-order transition in the standard Ising model to an abrupt first-order, critical transition. (10)

Gabriel, Nicholas, et al. Connecting the geometry and dynamics of many-body complex systems with message passing neural operators. arXiv:2502.15913. George Washington University and Brown University system mathematicians including Neil Johnson describe a real connection all the way from deep physical phenomena to cerebral and public realms by way of novel renormalization theories. Once again, a deep grounding in substantial, generative dynamics is achieved as they continuously instantiate and exemplify themselves everywhere.

The relationship between scale transformations and dynamics established by renormalization group techniques is a cornerstone is in effect from fluid mechanics to elementary particle physics. Integrating these methods into neural operators for many-body complex systems could enhance their utility and uncover a multiscale organization. In this regard, we introduce a scalable AI framework, ROMA (Renormalized Operators with Multiscale Attention), for learning evolution operators and apply it to large systems with 1M nodes, long-range interactions, and Kuramoto oscillators. (Excerpt)

Hecker, Nikolai, et al. Enhancer-driven cell type comparison re. Enhancer-driven cell type comparison reveals similarities between the mammalian and bird pallium. Science. February 14, 2025. Twenty VIB Center for AI & Computational Biology. Leuven, Belgium biologists apply the latest neuroimage techniques to discern a constant recurrence of genomic sources and cerebral architectures throughout the Metazoan creatures. Once again our Earthumen acumen reveals how nature’s long developmental course reuses in kind the same patterns and processes.

Despite vast diversity in behavior and cognition, a consistent similarity in brain structures and even gene expression is being found to exist across the amniote group of reptiles, birds, and mammals. Three papers in this issue explore the development and evolution of the brain telencephalon. Rueda-Alana et al used single-cell resolution and mathematical modeling to investigate sensory circuits in chicken, gecko, and mouse. Zaremba et al generated a spatial cell atlas in chicken pallium. Hecker et al developed deep learning models to identify telencephalon cell types in chicken, human, and mouse. (Editorial)

Combinations of transcription factors govern the identity of cell types, which is reflected by genomic enhancer codes. We used deep learning to compare cell types in the telencephalon across amniotes. To this end, we resolved transcriptomics data of the chicken telencephalon. Enhancer codes of avian mesopallial neurons are most similar to those of mammalian deep-layer neurons. (Hecker Excerpt)

Poggialini, Anna, et al. Networks with many structural scales: a Renormalization Group perspetive. arXiv:2406.19104. We cite this work by Università “Sapienza” Rome and Universidad de Granada (Miguel Munoz) system physicists as an example of the increasing avail of this foundational theory in many far removed cerebral, bioregion and public phases. As the second quotes advises, by the mid 2020s such a consistently apt utiliety can then be seen to imply a true universal invariance. See also, e.g., Laplacian renormalization group: heterogeneous coarse-graining by Guido Caldarelli, et al in the Journal of Statistical Mechanics: (August 2, 2024). Gabriel, Nicholas, et al. Connecting the geometry and dynamics of many-body complex systems by Nicholas Gabriel, et al. at (arXiv:2502.15913).

Scale invariance profoundly influences the dynamics and structure of complex systems from critical phenomena to network architecture. Here, we propose a precise definition of scale-invariant networks by leveraging the concept of a constant entropy-loss rate in a renormalization-group coarse-graining setting. This approach differentiates between scale-free and scale-invariant networks, revealing characteristics within each class. We then survey genuine networks to show that the human connectome exhibits true scale invariance. (Excerpt)

The network paradigm captures essential attributes of real-world complex systems, offering a natural framework for studying entangled interconnected systems across disciplines like neuroscience [1], ecology [2], and epidemiology [3], among others [4]. Understanding the evolution-ary dynamics of complex networks, as they adapt their connectivity patterns to achieve diverse goals, is crucial to understanding their long-term stability or other features influencing functional roles and performance [5, 6]. Notably, amidst the multitude of potential network structures, one organization ubiquitously arises in natural systems: the scale-free architecture. (1)

Brixi, Garyk, et al. Genome modeling and design across all domains of life with Evo 2. biorxiv. February 21, 2025. Some fifty computational geneticists at the Arc Institute (Google); Stanford University, NVIDIA, Liquid AI, UC Berkeley, Columbia University and the University of California, San Francisco post a latest version of this their human person – AI planet initiative to expand and enhance a wider scope of (epi)genetic encodings. See also Semantic mining of functional de novo genes from a genomic language model by Aditi Merchant at bioRxiv (December 18, 2024). In early regard, the concept that some Turing-type, genomic-like, LLModel code script is indeed running as it programs and informs an ecosmic genesis becomes increasingly evident.

While the sequencing, synthesis, and editing of genomic codes have transformed biological research, intelligently composing new biological systems requires deeper understandings of their immense complexity. We introduce Evo 2, a foundation model trained on 9.3 trillion DNA base pairs from a genomic atlas spanning all domains of life. Evo 2 learns from DNA sequences to accurately predict the influences of genetic variation. Guiding Evo 2 via inference-time search also enables controllable generation of epigenomic structure, which we demonstrate for the first time. (Excerpt)

Our work shows that a generative model of genomic language enables a machine learning model to achieve generalist prediction and design capabilities across Metazoan life. By learning the statistical properties of DNA via a trillion tokens of genomic sequences, Evo 2 can predict mutational effects on protein function, ncRNA function, and organismal fitness. (17) Biological foundation models capable of composing novel systems to advance biomedical innovation, but also raise safety, security and ethical considerations. Aligned with Responsible AI Biodesign (Google), we assessed and mitigated concerns prior to open source publication. (18)

New AI breakthrough can model and design genetic code across all domains of life. A team of scientists from UC Berkeley, Arc Institute, UCSF, Stanford University and NVIDIA have developed a machine learning model trained on the DNA of over 100,000 species across the entire tree of life. The model, called Evo 2, can identify patterns in gene sequences across disparate organisms that would typically need years to uncover. In addition to identifying disease-causing mutations in human genes, Evo 2 can design new genomes that are as long as that of simple bacteria. (UC Berkeley Center for Computational Biology)

Previous   1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9  Next