, A Class of Statistics with Asymptotically Normal Distribution, The Annals of Mathematical Statistics, vol.19, issue.3, pp.293-325, 1948.
DOI : 10.1214/aoms/1177730196
, Importance measures in global sensitivity analysis of nonlinear models Reliability Engineering & System Safety, pp.1-17, 1996.
, Global sensitivity analysis, the primer Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates, Mathematics and Computers in Simulation, vol.55, pp.271-280, 2001.
, Theorems and examples on high dimensional model representation Reliability Engineering & System Safety, pp.187-193, 2003.
, The exponential covariance function does not have a true range in the sense that there is no spatial dependence beyond that distance However in the geostatistical literature it is common to refer to the " effective range " , i.e. the distance at which the value of covariance function has decreased to If a is the parameter in the functional form of the covariance function then it is not the effective range, however it is what is called the correlation length. The authors need to clarify which it is, " effective range " or " correlation length/integral range, any case it appears that the authors have chosen a range that is quite small in comparison to the scale of the " block " used later, pp.24-26
, We misused the term "range", denoted by 'a', to refer to the parameter in the functional form of the covariance function C(.) In our illustrative example, the "effective range" is thus close to 3*a (as we use an exponential covariance function: C(h) = C(0) . [ ? ?(|h|) + (1-?)exp
, Manuscript was modified according to this recommendation: we replaced the term "influence" by more appropriate terms throughout the manuscript. We also replaced "lower" by "lesser
, Manuscript was modified according to this recommendation: we replaced the term "uncertainty" by "variance" when it was more appropriate
, line 55 Change " drives " to " determines " --> Thank you. Manuscript was modified according to this recommendation, ? Page, vol.13, issue.14
, Manuscript was modified according to this recommendation
, The real reason that EZ[Y(x] can be represented as an infinite series of Hermite polynomials is that EZ[Y(x] is square integrable with respect to the standard normal density function. Note this is " mean-square convergence, pp.49-54
, Manuscript was modified according to this recommendation. ? Page 15, line 50 Change " lower " to " smaller " --> Thank you
, See comment with respect to page 29 concerning the confidence intervals --> Thank you. Confidence intervals were computed using bootstraping, with n=100 replicas. A clarification was added on this point in the revised manuscriptMean values with a 95% confidence interval were then computed for each estimate using bootstrapping (100 replicas) ? Page 17, line 12 " uncertainty " again --> Thank you. Manuscript was modified according to this recommendation: we replaced the term "uncertainty" by "variance
, line 14 Change " illustrates " to " shows " --> Thank you
, line 25. The R software system incorporates a great many " packages " , some are part of the standard download but others are not. The authors should be more explicit about how they used R, i.e., which " packages " and which functions within those packages
, Manuscript was modified according to this recommendation: we replaced the term "influence" by more appropriate terms throughout the manuscript