git-svn-id: http://root.cern.ch/svn/root/trunk@22920 27541ba8-7e3a-0410-8455-c3a389f83636

   mathcore
   mathmore
   fftw
   foam
   fumili
   genvector
   matrix
   minuit
   minuit2
   mlp
   physics
   smatrix
   splot
   unuran
   quadp


git-svn-id: http://root.cern.ch/svn/root/trunk@22885 27541ba8-7e3a-0410-8455-c3a389f83636

- port latest changes (forgot in previous commit) in mathcore 
- update doc for groups and smatrix page
- changes in MathMore : 
     - add GSL MC integration
     - add GSL Multi-dim minimization, non linear fitting and simulated annealing
     - remove cdf (are in mathcore) now 
     - update doc for groups and mathmore page


git-svn-id: http://root.cern.ch/svn/root/trunk@20232 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@19826 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@19825 27541ba8-7e3a-0410-8455-c3a389f83636

Add protection when including header files.


git-svn-id: http://root.cern.ch/svn/root/trunk@15372 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@15311 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@15013 27541ba8-7e3a-0410-8455-c3a389f83636

- add in MatrixInversion.icc inversion for general squared matric and for symmetric matrices from CLHEP with N >= 7. The have better performances than original SMatrix version.
- add in MatrixFunction TensorProd for performing the Tensor product (request from LHCb)
- add in SVector a constructor from iterators. It is only for the SMatrix iterators (T*) and not a generic iterator to avoid a conflict with the (double,double)constructor
- remove from SMatrix the Sinvert and Sdet functions
- fix a bug in testKalman and stressKalman for CLHEP matrices (Vector index starts from 0, while matrix from 1).


git-svn-id: http://root.cern.ch/svn/root/trunk@15006 27541ba8-7e3a-0410-8455-c3a389f83636

add IsInUse function in SMAtrix and Expression to create automatically a temporary object in operations like A = B * A


git-svn-id: http://root.cern.ch/svn/root/trunk@14834 27541ba8-7e3a-0410-8455-c3a389f83636

Add two BinaryOp expression for the case of constant, where the objects are stored by value. In the default BinaryOp object are then stored by reference, giving a substuntial improvement in performances for operations like M3 = M1 + M2.

Add AssignSym::Evaluate to force an efficient assignment to a symmetric matrix from a general matrix. This is used now in the Similarity product for symmetric matrices

Add a dedicated Assign::Evaluate from a symmetric expression to a symmetric matrix. Need to loop only on the independent element

Capitalize Assign::Evaluate method

Add copy constructor in SMatrix from a symmetric matric to a standard matrix

Add tests for all constrant operations for vector, normal and symmetric matrices in testSMatrix

use AssignSym::Evaluate in testKalman for SMatrix_sym


git-svn-id: http://root.cern.ch/svn/root/trunk@14321 27541ba8-7e3a-0410-8455-c3a389f83636

- rename apply(i,j) to operator(i,j) in expression and function classes due to some bizarre conflicts with apply(i) found on Windows.
- rename linking warning in testSMatrix on Windows


git-svn-id: http://root.cern.ch/svn/root/trunk@14209 27541ba8-7e3a-0410-8455-c3a389f83636

- modify matrix expression to use both i and j index. This improves substuntially multiplication and transpose operations where there is no need to calculate the right offset for each element
- use a static const function in MatRepSym for the offset as suggested by Vincenzo. This gives some small improvements on Windows and Solaris
- generate dictionary also for the MatRepStd classes


git-svn-id: http://root.cern.ch/svn/root/trunk@14197 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@14119 27541ba8-7e3a-0410-8455-c3a389f83636

- add constructor for identity matrix
- rename Produc Similarity
- add Similarity for symmetric matrices
- change syntax for Sub methods as requested by LHCb
- add flag to retur status in Inverse methods


git-svn-id: http://root.cern.ch/svn/root/trunk@14114 27541ba8-7e3a-0410-8455-c3a389f83636

add support for symmetric matrices by introducing a matrix storage class
from J. Palacios.
Use a storage with pre-calculated offset for matrices of sizes <=6.
Create also a different expression class for vectors, this solves the problem on Windows.
Add Cramer inversion for symmetric matrices for sizes up to 6x6
Adapt Dinv,Dsinv and determinant calculation routines in the case of symmetric storage (but are not yet optimized)


git-svn-id: http://root.cern.ch/svn/root/trunk@14008 27541ba8-7e3a-0410-8455-c3a389f83636

add methods for retrieve slice and blocks of matrices. Capitaliza name of transpose function. Add as well tests for slices


git-svn-id: http://root.cern.ch/svn/root/trunk@13525 27541ba8-7e3a-0410-8455-c3a389f83636

add stl iterator interface in matrices. Cannoot add constructor Vector(begin,end) since it conflicts with Vector( val1, val2). Add also const methods for inverse and determinant. Rename enumeration variables to be consistent with ROOT naming conventions


git-svn-id: http://root.cern.ch/svn/root/trunk@13491 27541ba8-7e3a-0410-8455-c3a389f83636

Documentation coming


git-svn-id: http://root.cern.ch/svn/root/trunk@13343 27541ba8-7e3a-0410-8455-c3a389f83636