1Blurb:: Stopping criterion based on relative error 2 3Description:: Multilevel sampling seeks an error balance between the 4estimator variance and the remaining bias error at the highest level, 5the two contributors to mean squared error (MSE). Since the remaining 6bias error is generally unknown, the convergence_tolerance is used to 7provide a error target relative to the Multifidelity Monte Carlo estimator variance 8resulting from the pilot sample. If the pilot samples are not shaped for the 9low-fidelity model, i.e. the number of low-fidelity evaluations is equal to the 10number of high-fidelity evaluations for each level, the Multifidelity estimator 11falls back to a Multilevel Monte Carlo estimator which is used to assess the 12estimator pilot samples variance. 13The samples allocated at each level are proportional to \f$\frac{1}{\epsilon^2}\f$, 14so each order of magnitude reduction in convergence_tolerance will tend to increase the 15sample allocation by two orders of magnitude. Therefore, this control 16should be used with care to avoid allocation of huge sample sets that 17could overrun available memory. 18 19<b> Default Behavior </b> 20 21The default value for convergence_tolerance is currently .0001, which 22may be too resolved for expensive simulations or high variance QoI. 23 24 25Topics:: 26Examples:: 27Theory:: 28Faq:: 29See_Also:: 30