op_infer

Estimate model quantities from incomplete data and assess fitting methods. Private pre-alpha research software.

I am developing op_infer to connect scientific models to parameter inference, calibration checks and study design.

The software is private, pre-alpha and unreleased. The implementation includes Bayesian fitting through established inference libraries, simulation-based calibration, and replicated studies that test recovery on simulated data. Broader interfaces and application-specific validation remain in development.

The research asks how measurement assumptions and fitting procedures affect what can be learned about a scientific system. Simulation-based calibration assesses posterior inference using repeated data generated from the assumed model and prior. It does not establish that those assumptions describe observed data. Identifying a quantity from data also depends on the model, measurement process and available observations.

op_infer builds on OP System for model specification, OP Engine for numerical simulation, and trade-study for replicated evaluation. Their public documentation and releases provide the foundations for this work.

op_infer capabilities include fitting through established libraries, calibration checks on simulated data, and replicated study comparisons. Private, pre-alpha and unreleased.
Working research capabilities; broader interfaces and application-specific validation remain in development.

Vector PDF · Research in progress · Figure sources and downloads