Unfiltered · In-Progress · Real
Ongoing essays, experiment notes, and technical drafts published as research progresses. Not polished papers — live thinking.
Initial experimental results from the closed-loop biological energy system. Unexpected efficiency gains in the thermodynamic coupling stage. Data, charts, and interpretation of what we found — and what we didn't expect.
Constructing informative priors using population-level ECG datasets. 18% improvement over frequentist baseline. Notes on prior selection, likelihood specification, and posterior approximation methods.
A mathematical framework for modelling long-range dependencies in non-stationary time series using hierarchical Bayesian models. Theoretical derivations and implications for both ECG and energy output prediction.
Modelling PV-electrolyzer coupling efficiency under variable irradiance conditions. Simulation results and efficiency curves. Path toward 24-hour continuous hydrogen production from seawater.
Applying our temporal Bayesian framework to EEG signals. Notes on non-stationarity in neural time series and our approach to modelling long-range dependencies without overfitting.
A critical review of current PEM electrolyzer limitations and the design decisions behind Urjawave. Efficiency bottlenecks, membrane degradation, and the case for nano-catalyst integration.
When population-level data is available, how should we set priors? Exploring the tradeoff between informativeness and overfitting in Bayesian ECG analysis using hierarchical shrinkage.
The design principles behind BIOLOOP — how we close the carbon loop between CO₂ capture and biofuel synthesis. System architecture, biological reactor design, and the thermodynamic constraints we're working within.