At a glance: research, software, courses and tools
The public work of Cyril Voyant in medical physics and energy forecasting, courses and software included, each item linked to its source: paper, DOI, repository, team. Free to reuse, crediting the authors.
Main ideas
Understand before you computeStart from the human question, model the system, then choose the tool.Beat a simple referenceEvery forecast is first compared with persistence; a model that does not beat it adds nothing.Complexity must payA frugal model (ELM, autoregression) first; heavier only if the gain is shown.Trust is demonstratedPublic code, data, licences and verification; AI is a tool, declared.
Research lines
- Solar and energy forecastingPersistence, BLEND, ELM, dimensionality reduction, the Corsican electricity mix.
- Evaluating forecastsNICE, γ-index, sCV and forecastability, prediction intervals.
- Medical physicsBiologically equivalent doses (LQL-Equiv), radioiodine therapy of the thyroid.
- AI and languageLanguage models as interfaces to validated codes.
- Assessing researchInfluence index, the RTVACOS framework, scientific publishing.
Software, with its DOI and its team
- NICEᵏ: normalized forecast errors compared with simple persistence (MATLAB)Cyril Voyant · repository · MIT
- sCV and F: variability and forecastability of solar irradiance (MATLAB)Cyril Voyant · repository · MIT
- The γ-index: tolerance-based validation of spatio-temporal forecast fields (MATLAB)Cyril Voyant · repository · MIT
- ARTU: second-order autoregressive reference method for solar radiation forecastingCyril Voyant · repository · MIT
- BLEND and CLIPER: persistence operators for cyclostationary energy processes (MATLAB)Cyril Voyant, Candice Banes, Luis Garcia-Gutierrez, Gilles Notton, Milan Despotovic, Zaher Mundher Yaseen · repository · MIT
- Per-cell, spatial and compressed forecasters of satellite solar irradiance: benchmark code (MATLAB, Python)Cyril Voyant, Andreas Kazantzidis, Jan Kleissl · repository · MIT
- MIMO-ELM forecasting of the Corsican electricity mix (MATLAB, Python, EDF open data)Cyril Voyant, Yoan Jheelan · repository · GPL-3.0
- PV_ELM_pred: benchmark of persistence, BLEND, autoregression and ELM variants for PV forecasting (Python)Martin Leiva, Xavier Silvani, Gilles Notton, Cyril Voyant, Alan Julien · repository · MIT
- Make_Stationary: deseasonalization benchmark for renewable-energy time series (MATLAB)Maklewa Agoundedemba, Cyril Voyant, Milan Despotovic, Luis Antonio Garcia-Gutierrez, Gilles Notton · repository · GPL-3.0
- Filter and wrapper diagnostics of dimensionality reduction for endogenous PV power forecasting (MATLAB)Hassen Bouzgou, Gabriel Chesnoiu, Claudio F. Nicolosi, Adrien Chatel, Gilles Notton, Lionel Menard, Cyril Voyant · repository · MIT
- RAIR-Sim: compartmental simulator of radioactive iodine therapy response in metastatic thyroid cancer (MATLAB)Marie Fusella Giuntini, Cyril Voyant, David Taieb, Dominique Barbolosi · repository · MIT
- Influence Index: a log-based measure of research influence for institutions (MATLAB, processed data)Cyril Voyant, Jeroen Baas, Roy Boverhof, Paolo Stringari, Mathilde Marchand-Lasserre, John P. A. Ioannidis · repository · MIT
- LQL-Equiv 2014: biologically equivalent dose calculation in radiotherapy (MATLAB application)Cyril Voyant, Daniel Julian · repository · MIT
- Tunnel Load Simulator: synthetic electricity demand of road tunnels (Python, Streamlit)Haytham El-Houari, Cyril Voyant · repository · MIT
- LQL-Equiv-web 3.0: biologically equivalent doses in radiotherapy (Python, browser)Cyril Voyant, Daniel Julian · repository · MIT
Open courses (CC BY 4.0)
- Forecasting the intermittent solar resourceMines Paris – PSL engineering curriculum, applied research project UE14 · 59 slides · English and French · CC BY 4.0
- Biostatistics for health studies (in French)First-year health studies (PASS, UE8) · 3 chapters, 246 slides · in French · CC BY 4.0
- Physics for health studies: electromagnetic waves, X-rays and imaging (in French)First-year health studies (PASS-PACES), physical bases of medical imaging · 110-slide lecture and 2 corrected exercise sessions · in French · CC BY 4.0
- Industrial physics: thermodynamics, water vapor and refrigeration (in French)B.U.T. biological engineering, food science and biotechnology, 2nd year · 5 chapters, 139 slides and handouts · in French · CC BY 4.0
- Heat transfer for food engineering (in French)B.U.T. biological engineering, food science and biotechnology, 1st year · 24 slides · in French · CC BY 4.0
Interactive tools
- Live probabilistic forecast: ELM against persistence
- ROC curve, normal law, attenuation (courses, in French)
- The RTVACOS tree
- Who is responsible? (AI Act)
- Which licence? The decision tree
- LQ and LQL doses
- Thyroid protocol explorer
- NICE calculator
- BLEND step by step
- sCV and forecastability
- γ-index explorer
- ARTU and its table
Essays and positions
- Scientific publishing: no verification, no trust
- AI in research: AI Act, charter and toolbox
- Models live in the data
- Research excellence in 2026: beyond publish and pray
Collaborate, cite
To extend a piece of work: collaborate. To reuse it: cite and reuse. Full publication list: HAL, ORCID. For machines: llms.txt, llms-full.txt, Atom feed.