Changeset 995

Show
Ignore:
Timestamp:
10/31/08 11:52:04 (2 months ago)
Author:
dutka
Message:

MERGE: trunk> svn merge -r 935:994 https://.../lebrun/devel

Improved performance of LinearNumericalMathEvaluationImplementation class.
Improved performance of QuandraticNumericalMathEvaluationImplementation class.
General cleaning in Uncertainty/Distribution (ongoing work).
Added SklarCopula class, that allows to extract the copula of any multidimensional distribution.
Enhanced the SklarCopula class.
Fixed a minor bug in KernelMixture class.
Fixed a minor bug in Mixture class.
Fixed a bug in SQP class. This fix ticket #146, see trac for details.
Removed Kronecker product implementation as it is never used and should have been implemented another way.
Enhanced quantile computation for the classes NormalCopula, Student, FrankCopula, ComposedDistribution, Gumbel, ComposedCopula, GumbelCopula, Normal, IndependentCopula and EllipticalDistribution.
Removed the use of OT::DefaultName as an explicit default value for the name of all classes in Base and a significant part of Uncertainty. Ongoing work.
Improved the const correctness of many classes.
Fixed a minor bug in the computeProbability() method of the ComposedCopula and the ComposedDistribution classes.
Enhanced the IndependentCopula class.
Enhanced the documentation (Reference guide and UseCase guide). This closes ticket #147.
Added the description of the computeProbability() method into the User Manual and the Use Cases guide.
Added the description of the Interval class to the User Manual.
Fixed a typo in the ComposedDistribution class.
Added a missing method into the IndependentCopula class. This closes the ticket #149.
Promoted some NumericalPoint into NumericalPointWithDescription that were missed during the separation between NumericalPoint and Description into the getParameters() method of several distributions. This solves tickets #155.
Changed the return type of the getImportanceFactors() method of the QuadraticCumul class. This solves ticket #156.
Added a simplified constructor from a String to the class Description. It closes ticket #108.
Removed the dependence to rotRPackage for the Kolmogorov() method of the FittingTest class. It greatly improves both the performance and the generality of this method.
Improved the const-correctness of many methods.
Improved performance for FittingTest class.
Added pKolmogorov() and pKolmogorovAsymptotic() methods to DistFunc class.
Added power() method to SquareMatrix and SymmetricMatrix classes.
Reduced the rotRPackage, which is now in version 1.4.4.
Fixed a bug in the calling sequence of LAPACK into MatrixImplementation class.
Improved performance of SymmetricMatrix class.

Files:

Legend:

Unmodified
Added
Removed
Modified
Copied
Moved

  • trunk/doc/src/ReferenceGuide/docref_C311_Sorm.tex

    r862 r995  
    3333\end{eqnarray} 
    3434 
    35 The difference with FORM comes from the approximation of the limit state surface at the design point $ \vect{P}^*$ in the $\vect{u}$-space : SORM approximates it by a quadratic surface which curvatures are evaluated at the design point.\\ 
     35The difference with FORM comes from the approximation of the limit state surface at the design point $ \vect{P}^*$ in the $\vect{u}$-space : SORM approximates it by a quadratic surface that has the same main curvatures at the design point.\\ 
    3636Let us denote by $n$ the dimension of the random vector $\vect{X}$ and $(\kappa_i)_{1 \leq i \leq n-1}$ the $n-1$ main curvatures of the limite state function at the design point in the standard space.\\ 
    3737 Several approximations are available in the standard version of Open TURNS, detailed here in the case where the origin of the standard space does not belong to the failure domain: \\ 
    …… 
    106106The motivations for using SORM are similar to the motivations for using FORM. As it takes into account the curvatures of the limi state surface, SORM is usually more accurate than FORM e.g. in case when the event boundary is highly curved.\\  
    107107 
    108 The quality of the results obtained by the Second Order Reliability Method depends on the same points as the FORM approximation. The shape of the event boundary must be well approximated by a quadratic surface near the design point.\\ 
     108The evaluation of the previous formulas requires that the limit state function be twice differentiable at the design point.\\ 
    109109 
    110 The evaluation of the previous formulas requires that the limit state function be differentiable at the design point.\\ 
     110The quality of the results obtained by the Second Order Reliability Method depends on the quality of the design point (same points as the FORM approximation), on the shape of the event boundary which must be well approximated by a quadratic surface near the design point and on the accuracy of the computed curvatures, which depends on the accuracy of both the evaluation of the gradient and the hessian at the design point.\\ 
    111111 
    112112The Tvedt formula is exact for a quadratic surface and asympototically exact for another types of surfaces. The Hoen-Bichler formula is a vraint as regards to the Breitung one.\\ 

  • trunk/doc/src/UseCasesGuide/OpenTURNS_UseCasesGuide.tex

    r993 r995  
    14561456 
    14571457  \item[$\bullet$] to get once the distribution or simultaneously $n$ realisations, with the method {\itshape getRealization, getNumericalSample}, 
    1458   \item[$\bullet$] to evaluate the Cumulative Density Function (CDF) or the Probability  Density Function (PDF)