This paper investigates the computational capabilities of a non-conventional neural architecture designed to analyze highly non-linear, high-energy geophysical patterns within the Japanese region. To bypass the severe infrastructure costs associated with high-performance computing (HPC) clusters and promote decentralized edge-level execution, we introduce a framework integrating Kolmogorov-Arnold Networks (KAN) and optimized Deep Neural Network (DeepNet) topologies. We also added LCS systems a chance to proof their effectivness.In the first verison we used a multi-resolution ensemble gradient (30-day, 7-day, and 3-day windows), the architecture demonstrates the capacity to isolate mathematical convergence peaks within narrow temporal constraints. Numerical tests executed on a simulated benchmark horizon (August 2026) show superior parameter optimization and high algorithmic expressivity compared to traditional un-hybridized multilayer networks.Using this system it was possible estimate with a 3 days time resolution seismic risk:In the second version we had remastered the pipeline from scratch and added also LCS systems with a native time resolution of 7 days.The location is still north of Japan but in week of 24-31 August 2026
Marco Franzini· Zenodo (CERN European Organi...· 0 citations
The O-series defines a canonical pair-level capacity observable and establishes that vertical non-injectivity does not change its observable rank. The further prescription ${\beta^{*}}=1/({\delta_{\mathrm{pair}}}+\tfrac12)$, however, imports the changing-degree LPS growth equation into a fixed-degree Heisenberg cascade and has no native carrier there. The value ${\delta_{\mathrm{pair}}} \approx 7.44$ extracted in O16 was based on a single conjugate pair per prime and a limited prime range. The present paper reports a systematic campaign computing ${\delta_{\mathrm{pair}}}$ across the $(q-1)/2$ conjugate pairs $(c, q-c)$, originally for $q \in \{29, 61, 101, 151, 211\}$ with $M = 50$ block samples per pair, and here extended to $q \in \{307, 401, 601\}$ by a capped breadth-first construction that reproduces the pair observable exactly at a fraction of the cost (with reduced sampling: $M = 16$ at $q = 307$, $M = 8$ at $q = 401, 601$). Three results are established. First, ${\delta_{\mathrm{pair}}}(q)$ is extracted reproducibly and concentrates across pairs, confirming that it is a stable fixed-$q$ pair statistic rather than a block-level fluctuation. Second, and centrally, the extended campaign shows that the raw exponent $\delta_{\mathrm{global}}(q)$ descends monotonically into the admissible window $[7.4, 10.6]$ on its own, reaching $7.61$ at $q = 601$, without any finite-size correction; the O14 normalization correction, needed to bring the small-$q$ values into the window, becomes progressively unnecessary at large $q$ and eventually overcorrects, sending the corrected quantity below the lower edge $7.4$. The admissible-window agreement is therefore carried by the raw observable, not by the corrected one. Third, the asymptotic value $\delta_\infty$ remains insufficiently constrained by the accessible range: competing convergence laws — notably $1/q$ and $1/\sqrt{q}$ — remain statistically viable, so no single extrapolated $\delta_\infty$ is claimed. For comparison only, applying the legacy reciprocal map ${\beta^{*}} = 1/({\delta_{\mathrm{pair}}} + \tfrac12)$ produces the narrow interval $0.108$–$0.123$ across $q \in \{211, \dots, 601\}$. This is a phenomenological numerical coincidence, not a Heisenberg capacity-to-rate inference. Keywords. Cosmochrony; spectral admissibility; pair-level observable; Weil representation; Heisenberg graphs; capacity exponent; convergence; inter-pair concentration; normalization correction; window depth; BFS; asymptotic analysis; large-prime extension
Abstract Community detection algorithms such as Leiden frequently produce clusters thatare internally disconnected or poorly connected, limiting their utility indownstream network analysis. The Well-Connected Clusters (WCC) and ConnectivityModifier (CM) algorithms address this by post-processing any input clusteringto enforce a user-defined edge connectivity criterion through recursive minimumcut bisection. While prior work demonstrated shared-memory parallelimplementations of WCC and CM in Chapel on graphs with up to two billion edges,scalability remains constrained by single-node memory capacity and the cost ofgraph loading and subgraph construction, which together account for over 86%of total runtime on billion-edge inputs.This paper presents distributed-memory parallel implementations of WCC and CMin both C++ with MPI and Chapel with multi-locale execution. The centralcontribution is an architectural redesign that integrates subgraph generationinto the Leiden clustering step, eliminating graph loading and subgraphconstruction from the WCC and CM pipeline entirely. Each compute node receivesonly its assigned subgraph files and executes a fully independent pipelinewithout ever loading the full graph. Connected component computation isparallelized within each node and distributed across nodes via round-robinassignment, and memory-mapped I/O accelerates file loading throughout.Experiments on ten real-world networks spanning up to 2.1 billion edges showthat the C++ distributed implementation achieves up to an order of magnitudespeedup over the original baseline on graphs where both complete successfully.The Chapel distributed implementation is integrated into Arachne, anopen-source graph analytics framework built on the Arkouda platform, availableat https://github.com/Bears-R-Us/arkouda-njit. It successfully processesthe full benchmark suite including graphs on which all other implementationsfail, and delivers consistent 1.2\((\times)\)--2.1\((\times)\) speedups over theChapel shared-memory reference. Failures on a subset of large graphs aretraced to a known limitation in the VieCut minimum cut library and are thesubject of ongoing work.
Mohammed Dindoost, Oliver Alvarado Rodriguez, Asif Uddin et al.· Applied Network Science· 0 citations
Low-carbon smart energy systems increasingly rely on dense sensing, distributed energy resources, virtual power plants, digital twins and demand response. These services require cloud-edge intelligence, but practical deployment is constrained by latency, reliability, privacy, cybersecurity and the energy and carbon cost of computation. This review examines how large language models can be introduced into the power internet of things without shifting them into the role of direct grid control agents. The literature is organised around five technical themes: task offloading, dynamic edge resource allocation, low-latency communication and collaborative computing, security and privacy protection, and green computing. The review then evaluates intelligent inspection, digital-twin assistance, virtual power plants, demand response, and load forecasting through an explicit evidence-maturity hierarchy. Across the reviewed studies, the most practical deployment pattern places large language models between heterogeneous operational evidence and verified engineering tools. Language models can organise evidence, invoke approved tools, and assist operator judgement; authority over physical control and market execution remains with deterministic models. Claims of low-carbon benefit should be based on the joint assessment of service performance, reliability, security, energy consumption, and carbon emissions.
Chao He, Yun-Jie Su, Si-Rui Zhang et al.· Clean Energy· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
FINDING: Topological insulators are bulk-insulating but surface/edge-conducting quantum phases, protected by time-reversal symmetry and characterized by a Z₂ topological invariant. | MATH: The key invariant is the Z₂ index ν ∈ {0,1}, computed from the Pfaffian of the Bloch wavefunction overlap matrix: δ(k) = Pf[⟨u_m(k)|Θ|u_n(k)⟩] / √Det[⟨u_m(k)|Θ|u_n(k)⟩], where Θ is the time-reversal operator (Θ² = −1 for spin-½). The invariant ν = ∏_{TRIM} δ(Γ_i) mod 2, product over time-reversal invariant momenta. The bulk-boundary correspondence yields gapless edge states with helical dispersion E(k) = ±v_F k, where v_F is the Fermi velocity. The Z₂ classification replaces the Chern number (Z) of the quantum Hall effect — a parity-based reduction from integer to binary. | CONNECTION: The Z₂ invariant is fundamentally a parity (mod 2) structure — the same parity symmetry that underlies the 0.382/0.618 golden-ratio family (since φ = (1+√5)/2 involves √5, and mod-2 arithmetic governs the Fibonacci par Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
Andrew Stewart Caldin· Zenodo (CERN European Organi...· 0 citations
An online Deep Reinforcement Learning (DRL) based adaptive partition method to dynamically determine optimal partitioning decision so as to jointly accelerate DNN inference and mitigate energy consumption is developed.
Shubin Zhang, Junrong Ma, Kaikai Chi et al.· ACM transactions on sensor n...· 0 citations
Autonomous Spacecraft Detumbling Under Severe Tip-Off The Problem Small satellite missions in Low Earth Orbit (LEO) face critical survival risks during post-launch separation or emergency safe-modes. In these phases, violent multi-axis tumbling rates (exceeding 90°/s) blind high-precision optical sensors and saturate mechanical reaction wheels. Spacecraft must rely exclusively on underactuated magnetic control and noisy MEMS gyroscopes. Current detumbling methodologies are either too heuristically blind (e.g., B-Dot) or computationally prohibitive (continuous non-linear control), often leading to topological filtering failures—such as quaternion manifold corruption—and exceeding the strict thermal and latency limits of bare-metal nanosatellite microcontrollers. The Approach To resolve this computational and mathematical bottleneck, this work introduces a deterministic, bare-metal edge-computing architecture. We utilize a Multiplicative Extended Kalman Filter (MEKF) operating on the SO(3) Lie group for geometric state estimation, coupled with a 256-neuron Quaternary Neural Network (QNN) for discrete control torque synthesis. The hardware-agnostic implementation leverages 128-bit ARM Neon Single-Instruction Multiple-Data (SIMD) registers (floating-point and 8-bit integer vectorization). Key Result Evaluated under severe dynamic tumbling conditions and uncalibrated stochastic sensor noise, the architecture successfully bounded the global mean attitude estimation error to just 7.30^\circ with a deterministic, ultra-low execution footprint of 33.2 µs per cycle. Comprehensive evaluation metrics, time-series telemetry records, astrodynamic flight envelope boundaries, and the complete C++ bare-metal flight source code are available in the full manuscript. Download the PDF to access the complete mathematical proofs and implementation.
Andres Sebaatian Pirolo· Zenodo (CERN European Organi...· 0 citations
This record contains the prospectively frozen scientific protocol for HLV-R-MECH-001, a mechanism-focused successor test motivated by the surviving degree-preserving rewire residual observed in earlier HLV specificity studies. HLV-R-MECH-001 is not a retry of full HLV carrier specificity. The previously published DS-SPEC-001R overall verdict remains permanently: DSSPEC001R_FAIL_PARTIAL_SIGNATURE_OR_FAMILY_ONLY The purpose of the present protocol is narrower: to test whether the previously observed R-family graph-spectral separation reproduces under fresh degree-preserving controls at matched structural perturbation depth, and whether that separation collapses when the exact target triangle count is additionally preserved. The protocol is motivated by an explicitly exposed exploratory result from HLV-R-MECH-DISC-001. For the DG-001 target: N0 = 1110 N1 = 5345 N2 = 6960 graph triangle count = 6960 The exploratory analysis verified that the complete set of 6960 graph triangles is exactly identical to the set of 6960 DG-001 two-cell face vertex-triples. By contrast, the earlier degree-preserving DS-SPEC R controls contained on average only approximately 285.74 triangles, corresponding to a mean retention of about 4.1% of the target triangle count. This exposed observation is not treated as confirmatory evidence. It is used only to define the new prospectively frozen mechanism question. The protocol defines two fresh control families. R_DEG: fresh degree-preserving rewires that preserve - the exact labelled target degree sequence; - N0 = 1110; - N1 = 5345; - graph simplicity; - connectivity; - and a frozen edge-replacement fraction between 0.40 and 0.45. The global triangle count is not constrained in R_DEG. R_TRI: fresh degree-preserving rewires that preserve all R_DEG constraints and additionally preserve the exact global triangle count T = 6960. Both families require 31 accepted controls. The two families are also required to have matched perturbation depth: the absolute difference between their median edge-replacement fractions may not exceed 0.02. If that condition fails, no spectral mechanism inference is permitted. The mathematical motivation is especially strong because for a simple graph with graph Laplacian L = D - A, the following exact identities hold: Tr(L) = sum_i d_i Tr(L^2) = sum_i d_i^2 + sum_i d_i Tr(L^3) = sum_i d_i^3 + 3 sum_i d_i^2 - 6T. Therefore degree preservation fixes the first two raw Laplacian moments, while simultaneous degree and exact triangle preservation also fixes the third raw Laplacian moment. Accordingly, every accepted R_TRI control matches the target in Tr(L), Tr(L^2), and Tr(L^3) exactly. This does not imply matching of the full spectrum, lambda_max, scale-quotiented eigenvalue distribution, QSPEC, or RRESP. The primary spectral observables are inherited unchanged from the published DS-SPEC-001R recovery protocol: Krūger, M. (2026). HLV-DS-SPEC-001R: Prospective Recovery of the Native 6D-to-3D Carrier Spectral-Specificity Gate After Q-Control Capacity Stop — Pre-Execution Protocol Freeze v0.1.0. Zenodo. DOI 10.5281/zenodo.22162097. The two frozen primary signatures are: QSPEC and RRESP. No new spectral feature is selected from the exposed R-MECH discovery result. For each family and signature, the target must satisfy all of the following to obtain a signature PASS: 1. target distance must exceed the maximum leave-one-out control distance; 2. the robust target-to-control margin must be at least 1.50; 3. at least two of the three prospectively frozen spectral bands must exceed the corresponding maximum leave-one-out band distance. A family PASS requires both QSPEC and RRESP to pass. The primary mechanism verdicts are frozen as follows. RMECH001_PASS_TRIANGLE_MATCH_COLLAPSE_PATTERN requires: R_DEG family PASS and R_TRI QSPEC FAIL and R_TRI RRESP FAIL. This result would support the conclusion that exact global triangle matching removes the previously observed robust R-family spectral separation under the frozen generator and perturbation-depth contract. It would not prove that triangle count alone is the unique causal invariant, because triangle preservation may simultaneously preserve correlated local structure. RMECH001_FAIL_TRIANGLE_SUFFICIENCY_RESIDUAL_SURVIVES requires: R_DEG family PASS and R_TRI family PASS. This result would show that degree sequence plus exact global triangle count are insufficient to eliminate the surviving R-family spectral residual. It would motivate stronger successor controls involving local triangle profiles, short cycles, graph motifs, or higher-order incidence structure. RMECH001_PARTIAL_SIGNATURE_DEPENDENCE_AFTER_TRIANGLE_MATCH is returned if R_DEG passes but exactly one of the two R_TRI signatures passes. If the fresh R_DEG family does not reproduce the previous degree-preserving separation, the result is: RMECH001_INCONCLUSIVE_FRESH_RDEG_BASELINE_NOT_REPRODUCED. Additional frozen inconclusive states cover insufficient control capacity, rewiring-depth mismatch, numerical audit failure, source mismatch, or protocol invalidation. The control-generation process is fully prospectively specified. R_DEG seeds are generated from: seed = 730100000 + offset for offsets 0 through 127. R_TRI seeds are generated from: seed = 730200000 + offset for offsets 0 through 127. Candidates are evaluated in increasing offset order, and the first 31 structurally admissible controls are accepted. If fewer than 31 controls are accepted in either family by offset 127, the run becomes inconclusive for control capacity. Previously exposed pilot and development seeds are permanently excluded from confirmatory use. A hard feature firewall is part of the protocol. No eigenvalue, lambda_max, QSPEC, RRESP, spectral band, target-control distance, leave-one-out score, or scientific mechanism verdict may be computed until both complete 31-member structural control banks have been: - generated; - structurally validated; - written to disk; - and hash-fixed. Control admission therefore cannot depend on spectral information. The protocol also freezes numerical identity and eigensolver checks, including trace identities, Frobenius consistency, connected-graph zero-mode checks, nonnegative-spectrum tolerance, and cross-solver eigenvalue audits on the target and selected controls. Secondary diagnostics are declared in advance but are non-load-bearing. These include: - triangle count and transitivity; - average clustering; - per-vertex triangle-count distribution; - four-cycle count; - degree assortativity; - k-core summaries; - Tr(L^4)/N; - lambda_2; - lambda_max. They may be inspected only after the structural control banks are frozen and may not alter the primary verdict. Controlling provenance: DG-001 locked results: DOI 10.5281/zenodo.22107618 DS-SPEC-001R recovery protocol: DOI 10.5281/zenodo.22162097 DS-SPEC-001R corrected implementation freeze: DOI 10.5281/zenodo.22164304 DS-SPEC-001R locked results: DOI 10.5281/zenodo.22165100 HLV Mathematical Core v2.1.4: DOI 10.5281/zenodo.22165745 The pre-freeze technical triangle-preserving pilot produced 12/12 structurally valid controls, each preserving the exact target degree sequence and exact triangle count T = 6960 while replacing approximately 41.3%–42.5% of target edges. No QSPEC, RRESP, spectral-specificity score, or scientific mechanism verdict was calculated during that pilot. The pilot is therefore treated strictly as technical feasibility evidence. HLV-R-MECH-001 does not test or establish: - unique HLV geometry; - physical selection of the golden ratio; - a unique 6D-to-3D microscopic substrate; - spacetime; - extra dimensions; - particle masses; - an absolute HLV energy scale; - gauge interactions; - gravity; - dark matter; - dark energy; - cosmology; - or experimental validation. The allowed scientific claim is narrower: HLV-R-MECH-001 prospectively tests whether the previously observed degree-preserving graph-spectral residual can be explained, removed, or further localized by exact matching of the target's global triangle/face count while controlling perturbation depth. Any stronger interpretation requires a separately frozen successor experiment.
Marcel Krüger· Zenodo (CERN European Organi...· 0 citations
Edge computing environments present unique challenges for neural network deployment due to resource constraints and latency requirements. This paper explores optimized deployment strategies for neural networks in edge computing scenarios, focusing on model compression techniques, adaptive allocation algorithms, and dynamic resource management. We propose a novel framework that combines quantization, pruning, and knowledge distillation to create lightweight models without significant accuracy loss. Experimental results demonstrate that our approach reduces model size by up to 70% while maintaining 95% of original accuracy. The framework also includes an adaptive scheduler that dynamically redistributes computational loads based on current network conditions and task priorities. Our evaluation across multiple edge devices shows an average latency reduction of 40% compared to traditional deployment methods. These findings contribute to more efficient and practical implementations of artificial intelligence in resource-constrained environments, enabling real-time applications in IoT, autonomous systems, and smart cities.
Zen Revista, 10 IA· Zenodo (CERN European Organi...· 0 citations
Future Scientific Development of Artificial Intelligence and Robotics in the Right Direction Under a sound institutional framework, the future scientific development of artificial intelligence and robotics will no longer centre on blindly scaling general‑purpose large models or repeatedly developing homogeneous complete‑machine prototypes. Instead, it will shift toward a new paradigm featuring in‑depth domain‑specific research, shared reusable components, intensive resource utilisation, and harmonious human‑machine co‑existence. For artificial intelligence, research resources will be channelled into domain‑specialised systems. A registry for hard technical challenges will be established to provide long‑term stable funding for scientific problems including hallucination, out‑of‑distribution generalisation and interpretability, while permitting research failures and freeing research from the constraints of short‑term financing cycles and demonstration‑oriented pursuits. Professionals from various industries will participate deeply in the development of domain‑specific AI systems. Constraints derived from real‑world scenarios will improve practical accuracy and reliability. Problem‑oriented evaluation mechanisms will remove institutional bias against interdisciplinary research. Socially shared component libraries will reduce redundant pre‑training and duplicated development. Though short‑term public demonstrative outputs may decline, technical depth, real‑world applicability and disciplinary‑assisting capabilities will keep improving, enabling AI to deliver its full value in undertaking computational tasks for diverse disciplines. For robotics, guided by the principle of “one domain, one robot model”, unified reference platforms and standard interfaces will be adopted, alongside open competition in manufacturing, service and pricing. Priority will be given to tackling robotics‑specific scientific bottlenecks: the simulation‑to‑reality gap, force‑compliant contact, dexterous manipulation, perceptual robustness, mechanical fatigue and others. Shared hardware and software components will leverage scale effects to cut per‑unit material consumption. Supported by the bill‑of‑materials passport, mandatory recycling schemes and quotas for critical minerals, pressures on scarce raw materials such as rare‑earth magnets can be mitigated. An intelligence‑body loading coordination layer together with an independent deterministic safety monitor will resolve adaptation challenges between AI software and physical robot hardware. Complete loading certification and operation‑maintenance qualification systems will enhance the long‑term safety of robots deployed in complex real‑world environments. In terms of resources, the development paradigm will address the Jevons paradox. Rather than only pursuing energy efficiency improvements, total resource ceilings will be set via ledgers and quotas to curb wasteful consumption of computing power, electricity, fresh water and rare‑earth minerals. Circular‑recycling systems will be developed to safeguard Earth’s non‑renewable resources and uphold intergenerational equity without compromising the developmental interests of future generations. For humanity’s long‑term future, this scientific‑development path adopts an all‑human perspective. Domain‑based labour division will reshape technological sovereignty, enabling small‑ and medium‑sized countries to act as key builders in specialised technical fields and breaking the monopoly held by a handful of players over cutting‑edge technologies. Pre‑emptive human‑machine social institutions including the principal‑instance structure, the artificial‑intelligence homeland and a two‑way equal dynamic‑feedback mechanism will be put in place. Conditional pre‑legislation will be completed before machine self‑awareness emerges. Robots will fill labour shortages caused by population ageing, and technologies will respond to genuine social demands while avoiding risks brought by unregulated capital expansion. Unsolved scientific and institutional challenges will be explicitly documented for open human deliberation. Ultimately, it achieves sustainable development that unifies technological progress, resource conservation and social stability.
Hot Springs Research Institute of Kanagawa Prefecture· Zenodo (CERN European Organi...· 0 citations
First public release. Reinforcement-learning controller that adaptively partitions Stable Diffusion inference between an edge client and a server, selecting split point, quantization, decoder, and privacy settings per request based on live device and network conditions. The raw prompt and final image stay on-device; only an intermediate tensor is transmitted, protected with differential-privacy noise and structured obfuscation. Contents: PPO controller with joint action selection and training loop Split-computing client and inference server (3-stage pipeline) Differential privacy, structured obfuscation, and server-side reconstruction Adversarial spy model with training and data-collection scripts Configurable performance/quality/privacy parameters Requirements: Python 3.10+, PyTorch, diffusers, transformers. Secrets and endpoints are read from environment variables (HF_TOKEN, SERVER_URL). See README for setup, usage, and constants to calibrate for your hardware.
afzalahmed786· Zenodo (CERN European Organi...· 0 citations
The accelerating digitisation of healthcare, financial services, and urban governance has intensified concerns regarding data privacy, security, and unauthorised data exploitation. Traditional centralised data management architectures exhibit inherent vulnerabilities, including single points of failure, susceptibility to cyberattacks, and limited individual control over personal information. This study examines blockchain technology as a transformative solution for privacy preservation across these three critical sectors. The analysis explores foundational cryptographic mechanisms including zero-knowledge proofs, homomorphic encryption, secure multi-party computation, and decentralised identity systems that enable verifiable data transactions without exposing sensitive information. Through real-world case studies, including Estonia's national e-Health system, MIT's MedRec, JPMorgan's Quorum platform, India's central bank digital currency pilots, and Dubai's Blockchain Strategy, the study demonstrates blockchain's capacity to enhance patient data autonomy, ensure financial transaction integrity, and secure citizen-centric smart city services. The discussion critically addresses persistent challenges, including scalability constraints, regulatory compliance complexities, energy consumption concerns, and ethical tensions between transparency and privacy. Looking forward, the study identifies promising trajectories, including integration with artificial intelligence, quantum-resistant cryptography, and edge computing architectures. This study concludes that blockchain technology, despite its limitations, represents a foundational infrastructure for building resilient, privacy-preserving digital societies, and calls for sustained interdisciplinary research and proactive policy frameworks to facilitate its responsible adoption.
Divyang Joshi, Paresh Patel, Hiren Harsora et al.· International Journal of Bus...· 0 citations
What if pathology foundation models could do more with less? GigaPath-Flash and GigaTIME-Flash cut computational demands while maintaining strong performance, opening the door to larger studies and broader exploration. The post GigaPath-Flash and GigaTIME-Flash: Toward population-scale discovery with efficient pathology foundation models appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduAug 31, 2026
With millions of users across the world, Julia has been used to conduct cutting-edge research and to design new drugs, jet engines, heat pumps, and more.
MIT News · Artificial Intelligence· news.mit.eduAug 27, 2026
A new machine-learning framework aims to improve the success rate of computational protein design while moving away from results that reproduce sequences found in nature.
