Visualizing Simulation Error
Simulation results carry discretization, convergence, and model error; visualizing these keeps a rendered field from being mistaken for truth.
A rendered field is not reality
A simulation produces numbers on a grid using approximations: finite resolution, finite iterations, and a physical model that is itself an approximation. The resulting field can look crisp and authoritative while carrying real error. Visualizing that error is what separates honest simulation reporting from over-claiming.
Sources of error
- Discretization: the grid and time step limit resolvable features.
- Convergence: an iterative solver stops at a residual, not exactly.
- Model: the equations omit or approximate physics.
- Input uncertainty: parameters and boundary conditions are imperfectly known.
Making error visible
Techniques include plotting the residual or convergence history, showing a grid-refinement comparison so the reader sees whether the answer is resolution-independent, overlaying error estimates as a field, and rendering an ensemble across parameters rather than a single run (see Uncertainty Visualization). Marking under-resolved regions warns where the result is weakest.
Verification and validation
Verification asks whether the equations are solved correctly (a numerics question); validation asks whether the equations match reality (a physics question). Visualizations should make clear which is being shown and never let a verified-but-unvalidated result stand in for measured behavior.
Kronos use
Kronos presents simulation output with convergence and resolution context and honest gates, and does not present a modeled result as a hardware measurement; construction begins Q2 2027 and no hardware net-gain claim is made before first tritium.