Getting started¶

Start here to set up numgeo-ACT and run your first calibration.

What you need¶

numgeo-ACT orchestrates element-test simulations and an optimizer around your laboratory data. To run a calibration you need three ingredients:

1. numgeo: the finite-element solver that simulates each laboratory test. numgeo-ACT calls it as an external program, so the numgeo executable must be available on your system PATH. See the numgeo documentation for installation.
2. A Python environment with numgeo-ACT and its dependencies (NumPy, SciPy, pandas, openpyxl, matplotlib and Pillow). See Installation & requirements.
3. Your experimental database: a single Excel workbook containing one sheet per laboratory test, formatted so the reader can interpret it. See Experimental data and download the template.

The idea in one paragraph¶

You tell numgeo-ACT which constitutive model to calibrate and which of its parameters are free (to be optimized) versus fixed. Each free parameter is constrained by a lower and an upper bound. The selected optimiser then proposes parameter sets, simulates every selected laboratory test with numgeo, and measures how well each simulation matches your experiment using a similarity measure. These per-test scores are combined into a single objective value. Weights control the contribution from each test. The optimiser minimises this value and returns the best parameter set, together with a report and comparison plots.

Where to go next¶