Re: [eigen] about eebce0fdbf01 |

[ Thread Index | Date Index | More lists.tuxfamily.org/eigen Archives ]

*To*: eigen@xxxxxxxxxxxxxxxxxxx*Subject*: Re: [eigen] about eebce0fdbf01*From*: Benoit Jacob <jacob.benoit.1@xxxxxxxxx>*Date*: Fri, 21 May 2010 12:38:59 -0400*Dkim-signature*: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:received:in-reply-to :references:date:message-id:subject:from:to:content-type :content-transfer-encoding; bh=JlCNcYlcEZlT4+oBp+n4aGRbgsZbby+hCgVqhvtY2JY=; b=KswyJdrETmCCPGeu3iCHeUWwQ8aqqHmlDAtqNTgffVf7AvUCw44Ue1YIL6rl6y+F2+ 9NOK1rpnfqDiatc97QJ4xTqTcvl95ho5LU11YeAqTDo5nCrpcUoBR56xA9i6YETPS9EL a24GUDrTuu9uABt5xUYW0JLTBmYYmDjgVshzU=*Domainkey-signature*: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type:content-transfer-encoding; b=ZVdd2wTxq1HzIzGIEMbm1OgS6JE+AESRA0ixY1LYOBE5LJLcCZtA7/rmLaIN2zXyia M5o+fBqvstHhWavPHYAEa/qN5aIRddy87El98d4MVJeSqIUHPqKnM3yjApAsuJhT/128 2UIWBDdz+cvQFk02ybMZwCPFvuFD9m8oKezVQ=

Yes, this stuff needs a cleanup and needs to be rewritten as templates. The difficulty is that we must treat differently the Eigen objects from the numeric types. Not a problem, just needs a little bit of work. For a long time having this non-templated was OK but it's become problematic now that we allow a lot more types (all integer types etc..). Benoit 2010/5/21 Thomas Capricelli <orzel@xxxxxxxxxxxxxxx>: > > > Actually there are lot of things to do there. For example, i dont understand why > inline bool test_ei_isApprox(const float& a, const float& b) > inline bool test_ei_isApprox(const double& a, const double& b) > > and such are 'copy/pasted' and not using template. I've tried to define > > template <typename T> > inline bool test_ei_isApprox(const T& a, const T& b) > { return ei_isApprox(a, b, test_precision<T>()); } > > But the compilation fails because it tries to use it [with T = Eigen::CwiseUnaryOp<Eigen::ei_scalar_real_op<std::complex<double> >, Eigen::Matrix<std::complex<double>, 33331, 33331> >] > > so i gave up.... and i did not dare copy/pasting the "output error" on all those functions.. > > Thomas > -- > Thomas Capricelli <orzel@xxxxxxxxxxxxxxx> > http://www.freehackers.org/thomas > > In data venerdì 21 maggio 2010 13:18:57, Benoit Jacob ha scritto: > : > This commit is a good idea but one must be careful about one thing: >> >> if the test does VERIFY_IS_NOT_APPROX, this is currently defined as >> >> #define VERIFY_IS_NOT_APPROX(a, b) VERIFY(!test_ei_isApprox(a, b)) >> >> so it will output to cerr whenever it succeeds! The fix is probably to >> introduce a separate test_isNotApprox or something like that. >> >> Another thing is that for consistency, VERIFY_IS_EQUAL should do the samel. >> >> Benoit >> >> >> > > >

**References**:**[eigen] about eebce0fdbf01***From:*Benoit Jacob

**Re: [eigen] about eebce0fdbf01***From:*Thomas Capricelli

**Messages sorted by:**[ date | thread ]- Prev by Date:
**Re: [eigen] Blocked QR algorithm - lapack compatible ?** - Next by Date:
**[eigen] Pre-allocating LU objects** - Previous by thread:
**Re: [eigen] about eebce0fdbf01** - Next by thread:
**[eigen] Blocked QR algorithm - lapack compatible ?**

Mail converted by MHonArc 2.6.19+ | http://listengine.tuxfamily.org/ |