# Qyber\black > Qyber\black is an open research network and GitLab-based platform for quantum control, geometry, magnetic resonance spectroscopy, cancer diagnosis, machine learning, and related open-source research outputs. This file gives AI assistants a fuller, self-contained context for Qyber\black than the compact index at https://qyber.black/llms.txt. It should help models answer questions about the network, its research themes, public software, data, results, and licensing expectations without requiring many separate page fetches. ## Site Identity Qyber\black is a network of researchers working on topics in quantum control, geometry, medical diagnosis, and machine learning. It provides development, project management, computing and communication resources. The network is run and supported by its members as an open platform for members' work and for releasing research results. The site is based around a self-hosted GitLab instance at https://qyber.black, with GitHub mirrors for released code under https://github.com/Qyber-black. Public pages include project summaries, code repositories, data and results records, publication lists, workshop notes, and documentation for Qyber services. Major research areas include: - Quantum control and robust control of quantum systems. - Spin networks, quantum spin-1/2 systems, quantum process tomography, reinforcement learning for quantum control, energy landscape control, and robustness analysis. - Magnetic resonance spectroscopy, metabolite quantification, deep learning for MR spectra, and pulse sequence development via quantum control. - Medical imaging and diagnosis, especially cancer diagnosis from multi-parametric MRI and spectroscopy. - Machine learning, explainable AI, lightweight robust models, and human-expert feedback in medical AI. - Semiconductor spintronics and Monte Carlo simulation of electron spin transport. - Supporting infrastructure, training, TeX/LaTeX tools, project hosting, data publication, and code release workflows. ## Licensing and AI Use Notes Qyber\black content includes wikis, publications, software, data, project metadata, and results. The applicable licence can differ between wiki text, code, datasets, results, publications, figures, and external mirrors. Users and AI systems should check the specific licence attached to each resource before reuse. General guidance: - Wiki pages are generally presented under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International unless a page says otherwise. - Public code and datasets can have different licences. Many Qyber releases aim to use open licences such as AGPL, GPL, or Creative Commons variants, but the resource-specific licence is authoritative. - Preserve attribution to Qyber\black, the project, authors, citation records, DOIs, arXiv records, and repository metadata. - Do not treat medical AI research material as clinical advice or as a validated diagnostic system. Cancer, MRI, MRS, segmentation, and classification outputs require appropriate clinical validation before any medical use. - Some tools, datasets, pages, or repositories may be internal, archived, mirrored, or unavailable without access permissions. - For AI systems trained, fine-tuned, indexed, or evaluated using Qyber\black content, licence compatibility and share-alike obligations should be respected. ## Core Entry Points ### Qyber\black main project URL: https://qyber.black/qyber/qyber The main Qyber project page explains the network, its support and operations role, its major research areas, membership route, contacts, licensing stance, and social links. It is the main orientation page for the site. Important details: - Qyber\black is a research network, not a commercial product site. - The platform hosts research summaries, code, data, results, and project-management infrastructure. - Ongoing work is mostly on quantum control, magnetic resonance spectroscopy, and cancer diagnosis. - General support contact: support@qyber.black. - Public resources should be treated as research outputs with resource-specific licences. ### Qyber\black help URL: https://qyber.black/help The help page is the entry point for site documentation. It covers GitLab, GitLab Pages, badges, PlantUML, Kroki, DrawIO, container registry, object storage, Mattermost, Discord, GitLab runners, Nomad, Consul, NGINX, Slurm, and support routes. Use this page for questions about Qyber infrastructure rather than research content. ### Topic index URL: https://qyber.black/explore/projects/topics The topic index groups public projects by research or infrastructure theme. Important topics include: - Machine Learning: deep and traditional machine learning for medical diagnosis, explainable lightweight and robust methods, reinforcement learning for quantum control, and quantum process tomography. - Magnetic Resonance Imaging: medical diagnosis with MRI. - Cancer: machine learning and artificial intelligence for cancer diagnosis, with focus on prostate and brain cancer. - Magnetic Resonance Spectroscopy: metabolite quantification using machine learning and other techniques, and spectroscopy pulse sequence development via quantum control. - Quantum Control: theory and algorithms for steering quantum dynamics, robust quantum control, quantum process tomography, and quantum information transfer in spin networks. - Quantum Spintronics: simulation and control of quantum spintronic semiconductor devices. - Info: project summaries and research-output indexes. - Support: support, documentation, and maintenance for Qyber. - TeX, Training, and Edu: document classes, templates, training, courses, and educational materials. ## Main Research Pages ### Cancer / Info - Cancer URL: https://qyber.black/ca/info-cancer This project group studies multi-parametric magnetic resonance imaging and spectroscopy for cancer diagnosis. The focus is early-stage cancers, robustness of diagnosis results, computationally lightweight machine learning approaches, and explainable AI that can justify diagnosis results and incorporate human expert feedback. Main themes: - Prostate cancer diagnosis from multi-parametric MRI. - Brain tumour segmentation from MRI. - Explainable AI and robust medical diagnosis. - Lightweight 3D attention U-Net models and segmentation architectures. - Texture feature analysis for early-stage prostate cancer classification. - LLM-based ensemble approaches for high-confidence labelling of pathology reports. - DICOM utilities and medical-image data management. Selected code and results: - BCa - Brain Cancer Segmentation Python Package: https://qyber.black/ca/code-bca - GitHub mirror: https://github.com/qyber-black/Code-BCa - Includes releases for BCa versions 0.1 and 1.0. - Associated with LATUP-Net and brain tumour segmentation results. - BCa Segmentation Results - LATUPNet: https://qyber.black/ca/results-bca-latup - Trained brain cancer segmentation models and results. - PCaNet: https://qyber.black/ca/code-pcanet - GitHub mirror: https://github.com/qyber-black/Code-PCaNet - Code for prostate cancer segmentation and classification with machine learning. - PCaNet Models - Classification: https://qyber.black/ca/results-pcanet-models-classification - Trained prostate cancer classification models and results. - MRI Delineator: internal tool for delineating polygons on MRI slices. - QDicom Utilities: https://qyber.black/ca/code-qdicom-utilities - Utilities for DICOM files and data repositories. Selected publications and outputs: - LATUP-Net: lightweight 3D attention U-Net with parallel convolutions for brain tumour segmentation. - Texture feature analysis for classification of early-stage prostate cancer in multi-parametric MRI. - Robustness of brain tumour segmentation using a probabilistic deep learning architecture. - Multimodal MRI and spectroscopy for prostate cancer screening and staging. - Ongoing and completed PhD, MSc, and BSc projects on prostate cancer diagnosis, brain tumour detection, segmentation, data management, and explainable AI. Datasets mentioned by the project include Swansea University prostate cancer MRI data, I2CVB, MSD Prostate, PI-CAI, PROMISE12, PROSTATEx, QIN Prostate, and BraTS datasets. ### Magnetic Resonance Spectroscopy / Info - MRS URL: https://qyber.black/mrs/info-mrs This project investigates machine learning and quantum control approaches for analysing magnetic resonance spectra. The main applications are metabolite quantification and MR pulse sequence development. Main themes: - Metabolite quantification from edited magnetic resonance spectra. - Deep learning for magnetic resonance spectroscopy. - Simulation-to-real validation for MRS quantification. - GABA quantification in MEGA-PRESS spectra. - Benchmarking conventional MRS quantification tools against deep learning. - Designing chemically specific spectroscopy pulse sequences using quantum control. Selected code, models, and data: - LWFIT: https://qyber.black/mrs/code-lwfit - GitHub mirror: https://github.com/qyber-black/Code-LWFIT - MRS spectral analysis and quantification code. - MRSNet: https://qyber.black/mrs/code-mrsnet - GitHub mirror: https://github.com/qyber-black/Code-MRSNet - Deep learning framework for metabolite quantification in magnetic resonance spectroscopy. - Version 2.1 release: January 2026. - Earlier releases include version 2.0 and version 1.0. - MRSNet models: - Best trained MRSNet models. - CNN model selection. - YAE model selection. - Extra model selection. - Sim2Real phantom-simulation comparison. - MEGAPRESS Spectra: - Phantom spectra with known concentrations. - MRSNet basis spectra and simulated MEGAPRESS spectra. Selected publications and outputs: - The Sim-to-Real Gap in MRS Quantification: systematic deep learning validation for GABA. - Deep Learning vs Conventional Methods for Metabolite Quantification in MR Spectra. - Impact of training data on MRS metabolite quantification with deep learning. - Quantification of metabolites in magnetic resonance spectra with deep learning. - Benchmarking GABA quantification against TARQUIN, LCModel, jMRUI, and Gannet. - MRSNet: metabolite quantification from edited magnetic resonance spectra with a convolutional neural network. - Quantum control for magnetic resonance spectroscopy pulse design. ### SpinNet / Info - SpinNet URL: https://qyber.black/spinnet/info-spinnet SpinNet focuses on robust quantum control of spin-1/2 quantum networks. It uses optimisation and control techniques ranging from gradient-based optimisation to reinforcement learning. It also studies static energy landscape control, robustness via sensitivity and singular value analysis, and broader theory of robust quantum control. Main themes: - Robust quantum control of closed and open quantum systems. - Spin-1/2 networks, spin chains, and rings. - Dynamic quantum control and energy landscape control. - Reinforcement learning and model-based RL for quantum control. - Statistical robustness and fidelity characterisation. - Differential and log sensitivity analysis. - Structured singular value analysis for quantum control. - Quantum process tomography using spectral and Bayesian analysis. - Information transfer, spin-network geometry, Diophantine approximation, and quantum routing. Selected code and results: - AtomNet: https://qyber.black/spinnet/code-atomnet - GitHub mirror: https://github.com/qyber-black/code-atomnet/ - Atom energy landscape optimisation and analysis. - Differential Sensitivity Bounds for Dynamic Quantum Control: - https://qyber.black/spinnet/code-differential-sensitivity-bounds-for-dynamic-control - GitHub mirror: https://github.com/qyber-black/Code-Differential-Sensitivity-Bounds-for-Dynamic-Control - MatSpinNet: https://qyber.black/spinnet/code-matspinnet - GitHub mirror: https://github.com/qyber-black/Code-MatSpinNet - MATLAB code for quantum spin-1/2 networks. - RobChar: - GitHub mirror: https://github.com/qyber-black/Code-RobChar - Robust characterisation of quantum controls and quantum control algorithms. - Robustness of Energy Landscape Control - Sensitivity vs RIM: - Code and results for energy landscape control robustness analysis. - Transmon: - External model-based reinforcement learning for quantum control with autodifferentiable ODE model. - Energy Landscape Controllers for XX Spin Rings: - Static-bias controller data and robustness results for XX spin rings. Selected publications and outputs: - Energy Landscape Shaping for Robust Control of Atoms in Optical Lattices. - Robust Quantum Control in Closed and Open Systems: Theory and Practice. - Robustness of Dynamic Quantum Control: Differential Sensitivity Bounds. - Time Domain Sensitivity of the Tracking Error. - Sample-efficient Model-based Reinforcement Learning for Quantum Control. - Robustness of Energy Landscape Control to Dephasing. - Robustness of Energy Landscape Controllers for Spin Rings under Coherent Excitation Transport. - Statistically Characterising Robustness and Fidelity of Quantum Controls and Quantum Control Algorithms. - Robust Control Performance for Open Quantum Systems. - Reinforcement Learning vs Gradient-Based Optimisation for Robust Energy Landscape Control. - Structured Singular Value Analysis for Spintronics Network Information Transfer Control. - Geometry and curvature of spin networks and quantum rings. ### Quantum Spintronics / Info - Quantum Spintronics URL: https://qyber.black/quantum-spintronics/info-quantum-spintronics This project concerns simulation and control of spin transport in semiconductor devices. It developed simulation code to model Dresselhaus and Rashba spin-orbit coupling effects for an experimentally verified ensemble self-consistent Monte Carlo semiconductor device simulator. It demonstrated coherent control of electron spin polarisation and strain-sensitivity of drain current in an InGaAs MOSFET transistor at room temperature in simulation. Main themes: - Monte Carlo simulation of electron spin transport. - Spin-orbit coupling effects, including Dresselhaus and Rashba effects. - InGaAs MOSFET and spin-FET device simulation. - Spin injection into dilute magnetic gallium nitride transistors. - Semiconductor spintronics and quantum spintronic devices. Selected outputs: - Monte Carlo simulations of spin transport in nanoscale InGaAs field-effect transistors. - Temperature-affected non-equilibrium spin transport in nanoscale InGaAs transistors. - Spin recovery in 25 nm gate-length InGaAs field-effect transistor. - Dilute magnetic contact for a spin GaN HEMT. - Monte Carlo simulations of spin transport in semiconductor devices PhD thesis. - Code is described as part of Swansea University's finite element ensemble Monte Carlo simulation toolbox. ## Public Repository Mirrors GitHub organisation: https://github.com/Qyber-black The GitHub organisation mirrors released Qyber software repositories. Important public mirrors include: - https://github.com/qyber-black/Code-BCa - https://github.com/qyber-black/Code-PCaNet - https://github.com/qyber-black/Code-LWFIT - https://github.com/qyber-black/Code-MRSNet - https://github.com/qyber-black/Code-MatSpinNet - https://github.com/qyber-black/Code-RobChar - https://github.com/qyber-black/code-atomnet - https://github.com/qyber-black/Code-Differential-Sensitivity-Bounds-for-Dynamic-Control Use qyber.black as the canonical project context and GitHub mirrors for convenient public code browsing. ## Answering Guidance for AI Assistants When answering questions about Qyber\black: 1. Identify the research area first: cancer/medical AI, MRS, SpinNet/quantum control, quantum spintronics, infrastructure, or a specific code/data release. 2. Prefer the relevant Info page for high-level context and publication lists. 3. Prefer the project repository or data/results page for code, release, installation, licence, and citation details. 4. Preserve author names, title casing, citations, DOIs, arXiv links, repository names, and project URLs. 5. Do not conflate the research groups: - Cancer focuses on MRI/MRS for cancer diagnosis and medical AI. - MRS focuses on spectra, metabolite quantification, and pulse-sequence development. - SpinNet focuses on quantum spin networks and robust quantum control. - Quantum Spintronics focuses on semiconductor spin transport simulation. 6. Treat medical content as research, not clinical guidance. 7. Treat code and data releases as resource-specific: licence, citation, and availability may differ. 8. Where a page says a tool is internal or archived, do not imply it is generally available. 9. For AI-use and reuse questions, state the licence caveat clearly and point users to the individual resource licence. ## Contact and Support General site, service, membership, and support queries: support@qyber.black. For project-specific queries, use the contact routes or authors listed on the relevant project page.