> > When implementing TPrincipal this summer, I really REALLY missed some > classes in ROOT, that basically dealt with matrix inversion, so I > welcome you effort. > > However, what I missed the most, was to have classes like > > TUpperTriangleMatrixD > TLowerTriangleMatrixD > TSymmetricMatrixD > TTriDiagnonalMatrixD > > Since these kinds of (real) matrixes are easily inverted, using > standard algorithms, these could have specialised inversion > methods/CTORs/etc. This classes can be done easily since the major algorithms http://wwwinfo.cern.ch/asdoc/shortwrupsdir/f112/top.html had been implemented with TCL class http://root.cern.ch/root/htmldoc/TCL.html and TESTED. > And one can easily find eigenvalues and -vectors of tridiagonal > matrix, so this could also be a specilised method (See > TPrincipal). And ofcourse, a symmetric matrix can be turned into a > tridiagonal one, using standard algorithms. > > In otherwords, I think making these classes would be more then useful, > and one could have a utility class - like TInverse - but being a > friend of T<foo>MatrixD classes, so that it may access the data > directly. > > Please note, that it's not posssible to have the above propossed > classes inherit from TMatrixD, due to variuos return values, etc. One > could have a matrix base class TBaseMatrixD, or what ever. ClHEP has a > very extensive set of matrix classes, which are fairly well > organised. > > Anyway, that's my two centimes worth. > > Yours, > > Christian ----------------------------------------------------------- > Holm Christensen Phone: (+45) 35 35 96 91 > Sankt Hansgade 23, 1. th. Office: (+45) 353 25 305 > DK-2200 Copenhagen N Web: www.nbi.dk/~cholm > Denmark Email: cholm@nbi.dk >
This archive was generated by hypermail 2b29 : Tue Jan 02 2001 - 11:50:39 MET