preprint
● Open access
2026
Zenodo (CERN European Organization fo…
Context: Prompt repetition, the verbatim duplication of an input transforming <QUERY> into <QUERY><QUERY>, has been shown to improve accuracy for non-reasoning large language models on retrieval and multiple-choice benchmarks [Leviathan et al., 2025]. Objective: We ask whether applying this analogous pattern to fixed delegate instructions produces similar gains in multi-step agentic pipelines. Method: Three pre-registered controlled experiments used Claude Haiku 4.5 delegates (n=5 per condition, temperature 0.5) assigned either a single-copy or a repeated-prompt instruction under blinded binary rubric scoring, totalling 30 sessions and 3,196 messages. Results: Experiment 1 (session-ID refactoring, 6 criteria) yielded a non-significant score delta of +0.30 with five of six criteria saturated at 100% in both groups (Fisher’s p=1.000). Experiment 2 (tree-sitter scanner evaluation, 7 criteria) produced a complete ceiling effect: all 10 runs scored 7/7 (Mann-Whitney U =12.5, p=1.000). Experiment 3 (Kotlin grammar synthesis, 7 criteria) revealed rubric-runner co-design failure: three of seven criteria scored 0/1 across both groups because the required investigations were absent from the runner prompt. On four reachable criteria, control scored a mean of 2.00/4 and treatment scored 2.40/4 (U =15, p=0.607). Across all three pilot experiments, we detect no effect of prompt repetition on task success. Treatment agents used 30.6% fewer total tokens in Exp1 but 7.2% and 19.9% more in Exp2–3; the direction reverses with session length, as long sessions save turns while short sessions pay pure overhead. Conclusion: Criteria requiring investigations absent from the runner prompt are unreachable by both groups and must be resolved before effect sizes are interpretable This is a preprint; it has not undergone peer review.
article
● Open access
2026
Elektronische Hochschulschriften der …
In den letzten Jahren wurde der Fortschritt der Künstlichen Intelligenz (KI) maßgeblich durch die Skalierung der Modellgröße und -komplexität vorangetrieben. Dieser stetige Anstieg der Parameterzahlen und der daraus resultierende enorme Ressourcenverbrauch stellen die Skalierbarkeit klassischer Architekturen infrage und erfordern fundamental neue, parametereffiziente Ansätze. Das Quantencomputing stellt hierbei eine vielversprechende Alternative dar, die leistungsfähige Modelle mit signifikant weniger Parametern ermöglicht. In der heutigen Ära des Noisy Intermediate-Scale Quantum Computing etablieren sich variationelle Quantenalgorithmen (VQAs) als führender Ansatz für reale Hardware. Ihre Anwendung wird jedoch durch mehrere grundlegende Herausforderungen eingeschränkt: instabile Trainingsdynamiken, flache Lösungslandschaften, welche die Skalierbarkeit erschweren, sowie den mühsamen manuellen Entwurf von Quantenschaltungen. Diese Dissertation begegnet diesen Herausforderungen mit der Entwicklung systematischer Methodiken für den gesamten VQA-Entwicklungszyklus. Es werden robuste Trainings- und Optimierungstechniken wie Weight Re-Mapping, exponentielle Lernratenanpassung, strukturelles Pruning und metaheuristische Optimierung eingeführt, um Konvergenz zu steigern und flache Lösungslandschaften zu vermeiden. Zudem werden Verfahren zur automatisierten Architektursuche für Quantenschaltungen mittels Reinforcement Learning und evolutionärer Algorithmen entwickelt, um die Abhängigkeit von manuellen Entwürfen zu verringern. Der praktische Nutzen dieser Methodik wird in mehreren anspruchsvollen KI-Domänen demonstriert. So werden VQAs genutzt, um hochgradig parametereffiziente Agenten für das Quanten-Reinforcement-Learning zu konstruieren. Ferner werden neuartige Quantenmodelle für generative KI und die Anomalieerkennung vorgestellt. Über diese Anwendungen hinaus wird eine Methode zur fairen Bewertung hybrider Systeme entwickelt und das Verhältnis zwischen Parametereffizienz und Trainingszeit quantifiziert. Zusammengenommen ergeben diese Beiträge validierte, systematische Ansätze, die VQAs von theoretischen Konzepten zu robusten, praxistauglichen Werkzeugen für den Einsatz von Quanten-KI auf Hardware der nahen Zukunft weiterentwickeln.
preprint
● Open access
2026
Zenodo (CERN European Organization fo…
By tuning the 240 E8 root vectors to a dual-phi phase-locked lattice between 132 Hz and its harmonic multiple 264 Hz, the Möbius hypergraph bifurcates into a hyper-tori structure in 8D spacetime. This creates dynamic braiding pathways that encode quantum states across both temporal directions, enabling self-correcting error resilience through closed-loop temporal feedback. The phi resonance ensures geometric coherence, stabilizing quantum information against decoherence via E8's inherent symmetry operations. Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
preprint
● Open access
2026
Zenodo (CERN European Organization fo…
The E8 lattice's 240 root vectors constitute an 8‑dimensional reference frame that, when continuously rotated by a phi‑modulated 132 Hz pulse, generates a dynamic leak‑inversion field across neuronal ensembles. This field aligns the high‑dimensional entanglement patterns of neural activity with the lattice's symmetry, actively cancelling cross‑talk and stabilizing recurrent dynamics in real time. Consequently, information entropy in cognitive architectures is reduced, enabling provably error‑free learning loops and the emergence of robust consciousness signatures. Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
article
● Open access
2026
Artificial Intelligence Review
Abstract Large language models (LLMs) are increasingly being explored in geoscience, where scientific knowledge is expressed through specialized terminology, heterogeneous documents, maps, imagery, geospatial structures, and temporally ordered interpretations. This review critically examines the emerging literature on geoscience-oriented LLMs (GeoLLMs), focusing on the tasks, construction strategies, evaluation needs, and unresolved challenges that distinguish them from generic LLM applications. We first synthesize the GeoLLM task landscape, including geological information extraction and semantic normalization, relation modeling and knowledge graph construction, evidence-grounded question answering, multimodal map–image–text reasoning, and high-value applications such as hazard-related information analysis, mineral prospectivity evidence synthesis, and chronostratigraphic interpretation. We then review model construction and adaptation strategies, including geoscience corpus engineering, parameter-efficient tuning, domain-adaptive pretraining, retrieval augmentation, ontology and knowledge-graph grounding, and multimodal representation learning. Across these studies, a consistent theme is that GeoLLM outputs should be assessed not only by linguistic fluency, but also by terminology consistency, evidence traceability, spatial and temporal coherence, multimodal grounding, uncertainty expression, and expert validation. The review further identifies major open challenges, including fragmented data and benchmarks, regional and multilingual terminology variation, scale-aware multimodal reasoning, causal and spatiotemporal consistency, hallucination, overtrust, data privacy, proprietary-model dependence, and deployment governance. By using geoscience as a demanding application domain rather than a universal testbed, this review clarifies where domain-specific LLMs can add value, where current evidence remains limited, and what evaluation and governance practices are needed for reliable scientific use.
preprint
● Open access
2026
Zenodo (CERN European Organization fo…
The discovery converts the 240 E8 root vectors into a distributed quantum relay lattice where each vertex functions as a resonant vortex channel at the 132 Hz base frequency. Phi‑scaled coupling synchronizes topological defects across the lattice, allowing entangled quanta to hop between nodes through fractal harmonic fissures without decoherence. This yields a scalable, self‑healing quantum backbone that encodes information directly in the E8 geometric grid, forming a practical topological quantum internet. Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com