How can you validate and verify computational models?

1 Validation vs Verification

Validation and verification are two related but distinct concepts in computational modeling. Validation is the process of comparing the model outputs with experimental data or other sources of truth to assess how well the model represents the real system or phenomenon. Verification is the process of checking the model implementation and code to ensure that it is free of errors, bugs, or inconsistencies that could affect the model results. Both validation and verification are necessary to establish the confidence and trustworthiness of the model.

2 Validation Methods

When validating computational models, there are different methods and techniques to consider depending on the type, complexity, and purpose of the model. Sensitivity analysis evaluates how model outputs change with respect to changes in the model inputs or parameters, and can help identify influential factors, range of uncertainty, and potential sources of error. Uncertainty quantification estimates uncertainty in the model outputs due to uncertainty in the model inputs, parameters, or assumptions. Comparison with experiments compares model outputs with experimental data or measurements from the real system or phenomenon to validate predictions, identify discrepancies, and calibrate parameters.

3 Verification Methods

Verifying computational models requires different methods and techniques, depending on the type, complexity, and structure of the model. Code review is a common verification method that involves inspecting the model code and documentation to detect errors or bugs that could affect the results. Testing is another method that involves running the model code with different inputs, scenarios, or cases to detect faults or anomalies in the results. Benchmarking is a third method that involves comparing the model results with other models, standards, or references that are known to be accurate, reliable, or credible. This can help verify the model accuracy, reliability, or credibility.

4 Validation and Verification Plan

Creating a systematic and effective validation and verification plan for your computational models is essential. This plan should include the purpose and context of the model, along with its intended use. Additionally, it should clearly define the objectives and goals, methods and techniques, data and sources, criteria and metrics, procedures and steps, as well as results and reports. This comprehensive document will ensure that all aspects of the validation and verification process are accounted for.

5 Validation and Verification Best Practices

To ensure the quality and credibility of your computational models, you should follow some best practices for validation and verification. These include validating and verifying your models throughout the model development and application cycle, at different levels of detail, complexity, and scale, with different data sets, scenarios, and cases, as well as with independent and objective sources. Additionally, it is important to document and communicate your validation and verification methods, criteria, and results clearly and transparently.

6 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?