Research Papers Bundle
Eight peer-reviewed and preprint research papers spanning virtual reality, graph neural networks, number theory, and AI safety — by Tobias Weiß and collaborators.
📄 What's Included
1. Machine Learning for Modular Forms: GNNs, Spectral Methods, and L-Function Phenomenology — Comprehensive Preprint
May 2026 · 84 pages · Zenodo DOI: 10.5281/zenodo.20479512
Synthesis of 11 experiments across GNNs on algebraic graphs, trace-index graph prediction, LMFDB-scale Hecke trace regression, spectral triple operators, Sato-Tate moments, and FunSearch arithmetic function discovery — with formal theoretical analysis of GNN limitations on vertex-transitive graphs.
2. When Graph Neural Networks Meet the Riemann Hypothesis: A Systematic Negative Study
April 2026 · 7 experiment tracks
Seven systematic experiment tracks testing whether GNNs can predict spectral properties of algebraic graphs connected to the Riemann zeta function. Proves that vertex-transitivity in Cayley graphs makes every node identical to message-passing GNNs, establishing hard theoretical limits via the Weisfeiler-Leman hierarchy.
3. Applied Quantization Strategies for Consumer LLM Deployment
April 2026 · Empirical evaluation on RTX 4080 (12GB) and DGX Spark (128GB)
Layer-aware quantization strategies for Qwen3.5 on consumer hardware. Covers TQ3_0 KV cache compression, hybrid GPTQ-INT4+FP8 quantization, and context size optimization — with empirical memory-performance tradeoff data.
4. Overcoming Arachnophobia with Virtual Reality: A WebXR-Based Exposure Therapy Approach
January 2025 · WebXR application
A WebXR-based VR application delivering exposure therapy for spider phobia through browser-based virtual reality. Examines theoretical foundations of VRET, technical implementation with A-Frame, and therapeutic potential of accessible VR systems for democratizing mental health interventions.
5. Predicting L-Function Properties from Trace-Index Graphs using Graph Neural Networks
May 2026 · 12 pages · R² = 0.625 on zero prediction
GNNs on 1000-node trace-index graphs (Fourier coefficient representations) predict L-function zeros with R² = 0.625, analytic rank with 94.16% accuracy, and CM status with 100% accuracy across 46,347 weight-2 newforms from the LMFDB.
6. Impact of GenAI Sales Agent Timing in VR Commerce
December 2024 · ResearchGate publication
Investigates Immediate vs. Delayed interference timing of AI sales agents in VR commerce. Results show how the timing of AI assistant approaches affects consumer behavior, decision confidence, and purchase intention in the Metaverse.
7. Machine Learning for Modular Forms: From 1K to 200K Forms
June 2026 · 42 pages · 94.4% rank prediction accuracy
First systematic ML investigation of modular forms at scale with 200,000 forms. Validates the Birch–Swinnerton-Dyer conjecture empirically and achieves 94.4% rank prediction accuracy through Hecke trace regression.
8. Eine interdisziplinäre Kooperation in der Hochschullehre mit Hilfe der virtuellen Realität
December 2023 · DELFI 2023 · Gesellschaft für Informatik (GI)
Interdisciplinary higher education collaboration using Mozilla Hubs to teach the periodic table of elements and experimental design. Study with 40 participants demonstrates that CTML (Cognitive Theory of Multimedia Learning) principles significantly improve perceived learning outcomes in virtual environments.
Format
- PDF (print-ready, A4)
- Full bibliographies and DOIs included
- Instant download after purchase
Price
$29.00
Who This Is For
- Researchers and graduate students in AI, VR, mathematics, and HCI
- Engineers deploying LLMs on consumer hardware who need quantization strategies
- VR developers building therapeutic and educational applications
- ML researchers exploring the intersection of graph neural networks and number theory
- Technical leaders tracking the state of the art in applied AI research