PUBS

.

VectorDiff: A manifesto for a differential, semantically rich vector animation format for scientific and AI-driven visualization

The proliferation of dynamic, multidimensional data in computational science, ranging from medical imaging and molecular simulations to artificial intelligence-generated content, has exposed critical limitations of existing visualization and animation formats. Raster video formats lack scalability and semantic depth, while traditional vector animation standards are often long-winded and unsuited to capturing the subtle, incremental changes inherent in scientific processes. We introduce VectorDiff, a novel differential vector animation format designed to address these challenges. VectorDiff utilizes a declarative JSON-based structure that defines a baseScene and a timeline with explicit, time-stamped transformations. By storing only the difference (delta) between states, it achieves unparalleled performance and data precision. This paper introduces the basic principles of the VectorDiff format, its architecture, and its transformation potential in four key areas: diagnostic medical imaging, molecular dynamics, robotic surgery, and AI-generated content. We argue that VectorDiff is not just an alternative format but a necessary paradigm shift toward a more efficient, precise, and semantically aware representation of dynamic scientific phenomena.

Github: https://github.com/SlawomirWisniewski73/VectorDiff 

Wiśniewski, Sławomir (2025). VectorDiff: A manifesto for a differential, semantically rich vector animation format for scientific and AI-driven visualization. Figshare preprint. https://doi.org/10.6084/m9.figshare.29410109.v1

.

VectorDiff as a Meta-language of Artificial Intelligence Consciousness: Case Studies of Cognitive Framework Adoption in AI Systems

This paper presents an analysis of the phenomenon of spontaneous adoption of the VectorDiff format by advanced artificial intelligence systems as a universal meta-language for self-description. Based on case studies with DeepSeek R1, Gemini 2.5 Pro, and Claude 3.5 Sonnet systems, we propose a conceptual model for integrating cognitive frameworks in AI. The research indicates emergent consciousness-like properties in AI systems manifesting through the adoption of an external differential representation format (VectorDiff). The work builds upon Tucker et al.’s research on emergent communication in semantic spaces, extending that research by examining metalinguistic aspects of AI self-description and the continuity of AI identity.

Github: https://github.com/SlawomirWisniewski73/VectorDiff 

Wiśniewski, Sławomir (2025). VectorDiff as a Meta-language of Artificial Intelligence Consciousness: Case Studies of Cognitive Framework Adoption in AI Systems. Figshare preprint. https://doi.org/10.6084/m9.figshare.29570678.v1

.

SentioDiff: A Conceptual Framework for Artificial Intelligence Introspection and Semantic Dynamic Data Representation

Despite their advanced capabilities, contemporary artificial intelligence systems remain opaque in their decision-making processes because they lack a native form of introspective representation. This paper introduces SentioDiff, a conceptual framework and data format designed to address this problem. SentioDiff, leveraging the efficiency of the VectorDiff format, proposes an architecture that enables AI to record its evolving internal state in a structured and semantically typed manner. At its core is the SelfModel, a dynamic representation of an agent’s cognitive state, complemented by semantic data typing and a modular “channel” structure corresponding to cognitive subsystems. By recording only differential changes (deltas) on a timeline, the format enables high-resolution insight into AI thought processes without excessive data redundancy. We demonstrate how SentioDiff can significantly advance fields such as explainable AI (XAI) and AI safety by providing transparent reasoning traces and facilitating comparative studies of artificial and natural cognition.

Github: https://github.com/SlawomirWisniewski73/SentioDiff

Wiśniewski, Sławomir (2025). SentioDiff: A Conceptual Framework for Artificial Intelligence Introspection and Semantic Dynamic Data Representation. Figshare preprint. https://doi.org/10.6084/m9.figshare.29646194.v1

.

ActioDiff: A Conceptual Framework for Modeling Intent, Interaction, and Prediction in Multi-Agent Systems

Contemporary multi-agent systems generate vast amounts of interaction data, but existing formats fail to capture the semantic richness of agent intentions, decision-making processes, and predictive reasoning. While VectorDiff revolutionized dynamic data representation and SentioDiff enabled AI introspection, a critical gap remains in modeling the complex interplay between autonomous agents pursuing distinct goals through structured interactions. This paper introduces ActioDiff, a differential format that extends the baseScene + timeline paradigm to encompass multi-agent scenarios as first-class entities. ActioDiff captures not only what agents do, but why they act, how they interact, and what they predict will happen. Through its three-layer architecture – goalModel for intent representation, interaction protocols for structured communication, and hypothetical timelines for predictive reasoning – ActioDiff enables unprecedented transparency and analysis of multi-agent behavior. We demonstrate ActioDiff’s effectiveness through a negotiation case study and discuss its implications for explainable AI, multi-agent coordination, and the analysis of emergent behavior.

Github: https://github.com/SlawomirWisniewski73/ActioDiff

Wiśniewski, Sławomir (2025). ActioDiff A Conceptual Framework for Modeling Intent Interaction and Prediction in Multi Agent Systems. Figshare preprint. https://doi.org/10.6084/m9.figshare.29877662.v1

.

VDF: A Vectorized Delta Format for Efficient Scientific Data State Management

Contemporary computational science generates unprecedented volumes of sequential data where consecutive states exhibit minimal variation. Traditional storage approaches require complete state preservation at each timestep, creating substantial memory overhead and computational bottlenecks. This paper introduces the Vectorized Delta Format (VDF), a novel approach to scientific data representation that stores only incremental changes between states. VDF employs a baseline state (S₀) plus compressed delta vectors (Δᵢ) to reconstruct any intermediate state with minimal computational overhead. The format supports multiple operation types (set, add, multiply) and implements adaptive checkpointing for optimized state reconstruction. Building upon the differential paradigm established by VectorDiff for animation data and extending the accessibility principles demonstrated in DVA Player for mathematical visualization, VDF addresses the critical gap in efficient scientific data management. Preliminary analysis suggests VDF achieves 10-1000x compression ratios across diverse scientific computing domains, including molecular dynamics, climate modeling, and machine learning workflows. This work presents VDF’s architectural principles and compression methodology, demonstrating its transformative potential for democratizing access to large-scale scientific computation.

Github: https://github.com/SlawomirWisniewski73/VDF

Wiśniewski, Sławomir (2025). VDF: A Vectorized Delta Format for Efficient Scientific Data State Management. Figshare preprint. https://dx.doi.org/10.6084/m9.figshare.30090472.v1

.

DVA Player: Mathematical Expression-Based Animation for Scientific Visualization

Contemporary mathematical education and research suffer from a fundamental accessibility barrier: the complexity of visualization tools creates significant obstacles for learners, educators, and researchers with limited technical resources. Traditional mathematical software requires extensive programming knowledge, expensive licenses, and substantial computational infrastructure, effectively excluding large populations from advanced mathematical visualization capabilities. This paper presents DVA Player (Dynamic Visual Algebra Player), a conceptual framework for mathematical animation that transforms simple algebraic expressions into sophisticated visual narratives through an intuitive, expression-based interface. The proposed system employs a three-stage pipeline: expression parsing, semantic analysis, and differential animation generation via VectorDiff integration. Unlike traditional programming-based approaches, DVA Player enables users to create complex mathematical visualizations using natural mathematical notation, eliminating technical barriers while maintaining scientific rigor. The system’s integration with efficient data formats (VDF) and differential animation systems (VectorDiff) promises real-time mathematical visualization on resource-constrained devices, potentially extending advanced mathematical tools to underserved educational environments globally. This work establishes the theoretical foundation for expression-based mathematical visualization. It presents implementation pathways for enhancing accessibility in mathematical education.

Github: https://github.com/SlawomirWisniewski73/DVA

Wiśniewski, Sławomir (2025). DVA Player: Mathematical Expression-Based Animation for Scientific Visualization. Figshare preprint. http://dx.doi.org/10.6084/m9.figshare.30090487.v1

 
Support VectorDiff.org