Welcome to AfricArXiv

This initiative showcases UbuntuNet's commitment to fostering knowledge sharing, collaboration, and accessibility within the African research community. With AfricArxiv, researchers across the continent have a dedicated platform to disseminate their findings, making them accessible to a global audience. By facilitating open access to scholarly work, UbuntuNet Alliance plays a pivotal role in advancing the principles of open science, enhancing research visibility, and driving innovation across Africa.

 

Communities in AfricArXiv

Select a community to browse its collections.

Now showing 1 - 5 of 7

Recent Submissions

Thumbnail Image
Item
Bio-Adaptive Quantum Error Correction: Immune-Inspired Priors Enable 22–65% Overhead Reduction in Surface-Code Decoding
(Publisher, 2025-11-22) Charles L. Crawley II¹,Jorel Robinson², Barack Ndenga
Fault-tolerant quantum computation remains severely constrained by decoding overhead. Conventional minimum-weight perfect matching (MWPM) applied to the rotated surface code achieves a pseudothreshold of approximately 1.04% under circuit-level depolarizing noise, but only at the cost of substantial classical processing and qubit overhead. Here we introduce the first quantum-error-correction decoder that directly incorporates priors inspired by the human adaptive immune system. By using the empirically measured length distribution of T-cell-receptor (TCRβ) CDR3 regions as a Bayesian prior over error-chain plausibility, we modify PyMatching’s edge-weight model to obtain a 22% improvement in pseudothreshold on a distance-7 rotated surface code (100,000 shots per point). We further introduce a biologically motivated clonal-expansion mechanism: a cache of high-confidence syndrome–correction pairs that can be recalled in O(1) time when near-recurrent error patterns appear. Under temporally correlated (1/f-type) noise, this mechanism yields an additional 28–43% reduction in logical error rate, corresponding to total overhead reductions of 45–65% relative to MWPM. All code is open-source (MIT license) and fully reproducible in <10 minutes on free Google Colab. These results demonstrate that biological fault-tolerance architectures encode computational principles with direct applicability to quantum hardware, opening a new direction for bio-inspired quantum error correction.
Thumbnail Image
Item
The Pantheon Architecture: A Verifiable Foundation for Artificial General Intelligence.
(2005-11-22) Christopher Brown
The Pantheon Architecture: A Verifiable Foundation for Artificial General Intelligence. Here is the abstract for that report: > The prevailing paradigm in artificial intelligence focuses on training large, monolithic models for specific tasks. While powerful, these models often lack the ability to generalize or transfer knowledge to new, unseen domains, a hallmark of Artificial General Intelligence (AGI). This report presents a novel approach, demonstrating that a collective of smaller, specialized neural networks—a "Pantheon"—can exhibit emergent, AGI-like properties. > By facilitating a structured knowledge transfer between agents trained on tasks of varying complexity (dimensionality), we prove the existence of synergistic meta-learning. Using a fixed, verifiable initial seed (657454018), we demonstrate a repeatable experiment where a collective of 10 specialists achieves a 46.7% success rate in positive knowledge transfer across 45 unique pairings, including 7 instances of strong performance gains (>2.0%). The results provide definitive, reproducible evidence that a system of collaborating specialists can achieve a level of collective intelligence and generalization capability far exceeding the sum of its individual parts, representing a foundational step toward the architecture of more general artificial intelligence. > In simpler terms: The research proposes that AGI can be built from a group of small, collaborating specialized AI models (the Pantheon), rather than one massive model. By having these specialists share knowledge, the collective shows a high success rate (46.7%) in getting better at tasks than they were before the sharing, proving a form of emergent, synergistic learning. This is presented as a repeatable foundation for AGI architecture.
Thumbnail Image
Item
Experimental and Future Perspectives on the Quantum-π Framework: How to Measure and Test It in the Laboratory
(Publisher, 2025-11-22) Barack Ndenga
The concept of Quantum-π proposes that the mathematical constant π governs not only geometrical symmetries, but also the energy quantization, probability structure, and electronic organization of molecular and condensed-matter systems. To transform this theoretical framework into a testable scientific proposal, I outline a set of realistic experimental strategies capable of revealing π-driven signatures in chemical, optical, and electronic measurements. I identify measurable observables—including spectral line spacing, coherence envelopes, vibrational quantization patterns, electron delocalization metrics, and wavefunction normalization constants—that can be compared to π-predicted values with high precision. I also propose next-generation experimental platforms such as nanostructured potentials, π-sensitive interferometry, π-scaled vibrational spectroscopy, and electronic π-mode detection in polymers and 2D materials. This article presents a framework for validating Quantum-π in the laboratory, establishing a roadmap from theory to experimental physics and chemistry.
Thumbnail Image
Item
Quantum π-Index in Advanced Materials: Predictive Framework for Nanostructures, Functional Polymers, and Superconducting States
(Publisher, 2025-11-20) Barack Ndenga
I introduce the Quantum π-Index, a universal descriptor that encapsulates the fundamental interplay between electronic delocalization, structural periodicity, and quantum coherence in advanced functional materials. By systematically analyzing nanostructures, conjugated polymers, and superconducting phases, I demonstrate that π uniquely governs the quantization of energy levels, modulation of density of states, and maintenance of phase coherence. Through rigorous theoretical modeling, coupled with extensive numerical simulations and comparative cross-material analysis, I establish that π transcends its conventional geometric interpretation to serve as a structural invariant. This invariant dictates the underlying principles by which matter stores, transports, and manipulates electronic information at the quantum scale. Consequently, the Quantum π-Index emerges as a predictive metric for quantifying order, coherence, and emergent functionality in complex materials, offering crucial insights for the design and optimization of next-generation quantum and energy technologies
Thumbnail Image
Item
Q-Synapse: A Hybrid Quantum–AI Platform for Tumor State Classification Using Real Genomic Data
(Publisher, 2025-11-20) Barack Ndenga
Quantum Machine Learning (QML) provides a promising computational paradigm to extract complex patterns from biomedical datasets by exploiting quantum superposition, entanglement, and parameterized variational circuits. Despite substantial theoretical work, no existing study provides a fully functional, deployable, and experimentally validated quantum–AI hybrid platform applied directly to real genomic cancer classification. Here, i introduce Q-Synapse, the first operational prototype that integrates: (1) an Angle-Encoded Variational Quantum Classifier (VQC) running on a Qiskit simulator, (2) a real Artificial Intelligence feature-selection engine (GradientBoosting-based genomic ranking), (3) PCA-driven quantum dimensionality mapping enabling training on 2–4 qubit manifolds, and (4) a classical baseline (SVM/MLP) for scientifically rigorous benchmarking. Using the Wisconsin Breast Cancer dataset and reduced TCGA-style gene-expression structures, Q-Synapse demonstrates that quantum circuits can achieve competitive or superior accuracy in low-dimensional genomic subspaces, with smoother convergence behavior and reduced sensitivity to noise. The integrated Streamlit interface provides real-time training visualization, confusion matrices, and feature-importance analytics, resulting in a complete, reproducible, and extensible quantum-biomedical research platform. Keywords Quantum Machine Learning; Variational Quantum Classifier; Genomics; Biomedical AI; Cancer Classification; Quantum AI Hybrid Systems; Streamlit; Qiskit.