Two ways to explore the mathematics of geometallurgical testwork selection: an approachable challenge and a detailed specimen-to-model planning workflow.
INTERACTIVE CHALLENGE / FICTIONAL BIF
Beat the Sampler
Public educational demonstration
Can your geological intuition choose a more informative set of tests than a sampling algorithm?
Explore fictional core, assays and test costs. Choose up to three additional Bond Ball Mill Work Index (BBWi) tests with an invented AUD 4,500 budget, then compare your selection with Random, Representative and Information Gain strategies.
The useful question isn't just who “wins”, but why the most geologically interesting specimen may not be the best addition to a full, budgeted programme.
Locking a basket does not purchase tests or reveal the 12 missing BBWi values. Results compare model-expected information, not realised prediction RMSE, calibrated uncertainty or financial savings.
INTERACTIVE PLANNER / FICTIONAL RESEARCH WORKFLOW
Geometallurgical Testwork Optimiser
Public synthetic demonstration
Which existing physical drill-core specimens are worth testing next, and why?
Stage A checks declared material availability, laboratory eligibility and geological coverage under a budget. Stage B shows a separate fictional laboratory campaign where purchasing and revealing a selected BBWi result updates a Gaussian-process model before another Information Gain selection.
The two stages make physical specimen rules, coverage requirements, model uncertainty and cost accounting auditable rather than hiding them in a single opaque score.
The model is marked LIMITED and the sample-stock and lab declarations are fictional. This is not validated mine-site optimisation or a demonstration of superior real-world prediction. The public upload route is restricted; do not enter confidential, client-owned or licensed data.
