Numerical Validation Methods

Modelling resource heterogeneities in cloud simulations and quantifying their accuracy

“…The reason is that a simulator only abstracts a model but does not always reflect the real world scenarios. Therefore, the solution of any research problem through numerical simulation (experimentation) is not only to find a result, but also to guarantee the quality, precision and accuracy of the results [24].…”

“…To answer the above two questions, it is essential to establish verification and validation 7 criteria that "allow objective quantification of the difference between the results and the reality" [24]. Thacker et al [25] describe that numerical simulations never completely match "real" experimental results due to the model abstraction.…”

“…The authors, of the manuscript in footnote [ 8 ], have described different methods including anti-bugging, simplified models, and tracing, that can be used to verify simulation models. Furthermore, Jauregui et al [24] have discussed different kinds of validation that can be performed using:…”

“…These approaches are: (a) real system measurements; (b) expert intuition; and (c) theoretical results/analysis. Out of these three, "comparison with a real system is the most efficient and reliable method of validating a simulation model" [24]. However, this is often infeasible because: (i) the real system does not exist; or (ii) the measurements might be either impossible, or too difficult and expensive to carry out.…”

“…Nevertheless, it is mandatory to introduce a measuring instrument whenever we perform a measurement and consequently this directly or indirectly affects the system being measured/observed. Therefore, it is very important to have the highest similarity between simulation configurations and measurements [24].…”