From vollinga at thep.physik.uni-mainz.de Mon Jul 11 12:25:39 2005 From: vollinga at thep.physik.uni-mainz.de (Jens Vollinga) Date: Mon, 11 Jul 2005 12:25:39 +0200 Subject: [GiNaC-devel] GiNaC 1.3.2 release Message-ID: <20050711102539.GA19009@thep.physik.uni-mainz.de> Hi, GiNaC 1.3.2 is out and available. The changes are: - GCD avoids to produce expanded expressions. - Fixed bug in expanding expressions containing dummy indices. - Fixed static initialization order bug. - collect_common_factors now works with powers. - Modernized configuration scripts. Thanks to all the contributers, especially C.Dams, V.Kisil and A.Sheplyakov. As always, this release can be downloaded from ftp://ftpthep.physik.uni-mainz.de/pub/GiNaC/ Enjoy, Jens Vollinga From kisilv at maths.leeds.ac.uk Tue Jul 12 12:10:26 2005 From: kisilv at maths.leeds.ac.uk (Vladimir Kisil) Date: Tue, 12 Jul 2005 11:10:26 +0100 Subject: [GiNaC-devel] Another clifford patch Message-ID: Dear All, Here is another patch for Clifford part of GiNaC which contains mainly four components: 1. Some housekeeping alteration of code. 2. Previously GiNaC method is_equal() did not take in account representation labels, i.e. a call dirac_gamma(mu).is_equal(dirac_gamma(mu, 1)) returned true. This leads to miscalculations since for example (dirac_gamma(mu) *(dirac_gamma(mu, 1)).to_rational(L) returned something like pow(symbol7, 2) instead of symbol7*symbol8. I fixed it through an "inclusion" of representation_label as op(2) for dirac_gammas. A fake alteration of let_op(2) is also made. This may not look somewhat trick but I cannot find a better solutions. By the way, does the same problem affects color.cpp? 3. Since metric member variable of the clifford class can contains symbolic entries I add susb() methods which can access metric as well. 4. I add a new member variable commutator sign to clifford class. It is used in the transformations for two clifford instances X and Y as follows: X*Y = commutator_sign*Y*X + 2*metric(X, Y) commutator_sign is an int type and for Clifford algebras is equal to -1. Such an addition with only two lines altered in canonicalize_clifford() allows to derive from clifford class subclasses for Lie algebras (commutator_sign = 1, an demo example is attached) and general algebras defined through commutation identities (e.g. q-deformed algebras can be defined for commutator_sign =0 and a proper construction of the metric). The bubble-sorting realisation of canonicalize_clifford() was initially good for simply anticommuting Dirac gammas. However since the clifford class was very generalised it becomes rather inefficient for high powers of elements and a proper modification is in the "to-do" list. Will it be worth to develop the included Lie algebra subclass into the proper part of GiNaC? Such an addition do break a binary compatibility however this only happens for low level constructors which was recommended to avoid in the user programs. Best wishes, Vladimir -- Vladimir V. Kisil email: kisilv at maths.leeds.ac.uk -- www: http://maths.leeds.ac.uk/~kisilv/ -------------- next part -------------- A non-text attachment was scrubbed... Name: ginac-clifford.diff Type: text/x-c Size: 9319 bytes Desc: A patch for Clifford URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: lie-alg-demo.cpp Type: text/x-c Size: 9289 bytes Desc: A demo of Lie algebra reaisation in GiNaC URL: From vollinga at thep.physik.uni-mainz.de Tue Jul 12 20:06:12 2005 From: vollinga at thep.physik.uni-mainz.de (Jens Vollinga) Date: Tue, 12 Jul 2005 20:06:12 +0200 Subject: [GiNaC-devel] Another clifford patch In-Reply-To: References: Message-ID: <20050712180612.GA907@thep.physik.uni-mainz.de> Hi Vladimir, On Tue, Jul 12, 2005 at 11:10:26AM +0100, Vladimir Kisil wrote: > Here is another patch for Clifford part of GiNaC which contains > mainly four components: Thanks! It has been applied to CVS. > Will it be worth to develop the included Lie algebra subclass into the > proper part of GiNaC? I don't know. But as soon as I have some more time (maybe in October?), I'd like to add to the GiNaC package a sub-directory containing lots of examples. There could be included the examples from the tutorial, responses to questions on the mailing list and things like your Lie algebra class. With some nice documentation and an index to all the examples, I think this would be a good thing for GiNaC. Maybe this could be a reason not to make this class a permanent part of GinaC itself. Regards, Jens From kisilv at maths.leeds.ac.uk Wed Jul 13 12:37:12 2005 From: kisilv at maths.leeds.ac.uk (Vladimir Kisil) Date: Wed, 13 Jul 2005 11:37:12 +0100 Subject: [GiNaC-devel] Another clifford patch In-Reply-To: Your message of "Tue, 12 Jul 2005 20:06:12 +0200." <20050712180612.GA907@thep.physik.uni-mainz.de> Message-ID: >>>>> "JV" == Jens Vollinga writes: JV> October?), I'd like to add to the GiNaC package a sub-directory JV> containing lots of examples. Thant is really a very good idea! I think many people have something to share from their private collections. Best, Vladimir -- Vladimir V. Kisil email: kisilv at maths.leeds.ac.uk -- www: http://maths.leeds.ac.uk/~kisilv/ From kreckel at ginac.de Sat Jul 23 00:29:43 2005 From: kreckel at ginac.de (Richard B. Kreckel) Date: Sat, 23 Jul 2005 00:29:43 +0200 (CEST) Subject: [GiNaC-devel] Another clifford patch In-Reply-To: <20050712180612.GA907@thep.physik.uni-mainz.de> Message-ID: On Tue, 12 Jul 2005, Jens Vollinga wrote: > On Tue, Jul 12, 2005 at 11:10:26AM +0100, Vladimir Kisil wrote: > > Here is another patch for Clifford part of GiNaC which contains > > mainly four components: > > Thanks! It has been applied to CVS. > > > Will it be worth to develop the included Lie algebra subclass into the > > proper part of GiNaC? > > I don't know. But as soon as I have some more time (maybe in October?), > I'd like to add to the GiNaC package a sub-directory containing lots of > examples. There could be included the examples from the tutorial, > responses to questions on the mailing list and things like your Lie > algebra class. With some nice documentation and an index to all the > examples, I think this would be a good thing for GiNaC. Maybe this could > be a reason not to make this class a permanent part of GinaC itself. This sounds like a helluvagood idea. The only thing I'm worried about is about the stability of the examples' quality. One useful step to assess the quality is to automatically exercise the examples, maybe each in it's own testsuite. Recently, it was discovered that one of the examples of CLN segfaulted when executed [0]. This has been broken since GCC 3 came out in June 2001. The problem has gone unnoticed for 4 years. :-( I've only recently fixed it in CVS. Maybe there should've been a way to force execution of the examples. Maybe the testsuite should've been better. -richy. [0] -- Richard B. Kreckel From kisilv at maths.leeds.ac.uk Wed Jul 27 20:00:25 2005 From: kisilv at maths.leeds.ac.uk (Vladimir V. Kisil) Date: Wed, 27 Jul 2005 19:00:25 +0100 Subject: TINFO policy (WAS: [GiNaC-devel] Another clifford patch) In-Reply-To: Message from "Richard B. Kreckel" of "Sat, 23 Jul 2005 00:29:43 +0200." Message-ID: >>>>> "RK" == Richard B Kreckel writes: RK> This sounds like a helluvagood idea. The only thing I'm worried RK> about is I am also thinking about some kind of policy to choose a right TINFO for new homemade GiNaC classes. It will be good to avoid possible clashes in the case of all such examples collected in one place. Best, -- Vladimir V. Kisil email: kisilv at maths.leeds.ac.uk -- www: http://maths.leeds.ac.uk/~kisilv/