The Power of Accurate Uncertainty Quantification

At Chatelet Resources we combine state-of-the-art statistical models with probabilistic inference to solve high-value problems which sit at the crux between physics, experiment and human intuition: problems such as predicting the flow of hydrocarbons from shale oil and gas wells, optimizing hydrofracture stimulations, monitoring micro-earthquakes, finding and delineating gold deposits and figuring out the role of shape changes in evolutionary biology.  We work on complex systems, building rigorous frameworks in which science, data and expertise mathematically interact.

 

​Probabilistic Inference does not predict the future exactly.  Instead, it is the study of what could happen, and why.  Our models give industry leaders the ability to make clear-eyed, rational decisions in the face of risky ventures, secure in the knowledge that they understand the odds of success.



We do it by carefully building what we call a Knowledge Representation for every new domain.  Based on a combination of physical and professional knowledge and statistics derived from historic data, our models describe how systems actually work.  We then design a problem-specific Inference Algorithm which uses the Knowledge Representation to evaluate the probability of millions of possible scenarios – searching for not just the best answers to particular problems, but all possible explanations for a given set of data.    When we return results, it is always in terms of a probabilistic spread – the equivalent of asking a million experts for their opinion, figuring out which answers really do explain the data, and then being able to see what they have in common, including which ones have the highest probability, what the average is, and where uncertainty lies.

​

Each of our projects is spun into a separate Limited Liability Company and commercialized through partnership with other visionary enterprises who are leaders in their field.