CGAL users discussion list

[Andreas Fabri <>]

Joe C wrote:
> Thank you for your analysis, andreas.
> As long as the exception does not break anything, I think I can live 
> with it :) Just curious, if in the catch clause the predicate is run with
> an arbitrary precision number type like Gmpq, why is it still faster 
> than using Gmpq at the first place? 

Hi Joe,

The  try uses something extremely fast.
Only the catch uses Gmpq, and exceptions are the exception,
that is they only happen rarely.

andreas


Or it just chooses some other number
> type which is also faster than Gmpq? In my testing, 
> Exact_predicate_exact_construct
> ions_kernel still has the best performance even with exceptions thrown.
>
>
> Thanks.
>
> Joe
>
> On Thu, May 15, 2008 at 8:45 AM, Andreas Fabri 
> < 
> <mailto:>> wrote:
>
>     Hi Joe,
>
>     The exceptions get catched right, that is it is not up to you to
>     catch them.   They are used in filtered predicates. The idea
>     is to first try to compute the result of a predicate with
>     interval arithmetic on doubles, and if that fails an exception
>     is thrown and in the catch clause the predicate is run with
>     an arbitrary precision number type like Gmpq.
>
>     Now, when you switch to Gmpq you compute all predicates
>     exactly, and that's why there are no exceptions being thrown,
>     but at the price of performance.
>
>     andreas
>
>
>     Joe C wrote:
>
>         Michael and Peter, Thank you for your help!
>
>         As Michael mentioned in his email, I do have some reason to use
>         sqrt kernel. I tried to convert a sphere polyhedron (see
>         attached Sphere.txt) to Nef_polyhedron_3. The generation and the
>         conversion were done by using
>         Exact_predicates_exact_constructions_kernel. Both were
>         successful except that I got some exception:
>         CCGAL::Uncertain_conversion_exception when converting polyhedron
>         to Nef_polyhedron_3. If I ignore all the exceptions, I am still
>         able to do boolean operations (like union) using the converted
>         Nef_polyhedron_3. It seems that the exceptions do not affect the
>         results at all, but just annoying. Then I try using sqrt kernel
>         in Release mode, surprisingly, all the exceptions are gone, but
>         compiling failed when in Debug mode.
>
>         I guess this is more like Kernel Selection problem. I searched
>         the forum and also tried Simple_cartesian<Gmpq>. It is much
>         slower than Exact_predicate_exact_constructions_kernel, but NO
>         exceptions thrown.
>
>         If I still want to have good performance by using
>         Exact_predicates_exact_constructions_kernel, it there any way
>         that I can suppress exceptions thrown (or is it a bad idea to do
>         so)?
>
>         Cheers,
>
>         Joe
>
>
>
>         On Thu, May 15, 2008 at 5:16 AM, Peter Hachenberger
>         
> <
>  
> <mailto:>
>         
> <mailto:
>  
> <mailto:>>>
>         wrote:
>
>            Hi,
>
>            that's a correct analysis of Michael. Try the
>            Exact_predicates_exact_constructions_kernel; it should work.
>         If you
>            really need the kernel with sqrt, you might need to wait for
>         the next
>            release. The Nef_3 package has some functions that I intended as
>            features, but actually limit the number of applicable
>         kernels. I will
>            try to make some of these features optional.
>
>            Peter
>
>            On Thu, 2008-05-15 at 09:55 +0200, Michael Hemmer wrote:
>             > Hi,
>             >
>             > I am not an expert for Nef but AFAIK it does not need the
>         kernel
>            with sqrt.
>             > It seems that the code is trying to decompose the Field
>         type into
>             > Numerator and Denominator (for some reason),
>             > but this is not possible for CORE::Expr.
>             >
>             > Have you tried another kernel, i.e., one that is not using
>            CORE::Expr as
>             > FT.
>             >
>             > best,
>             >  Michael
>             >
>             >
>             > Joe C wrote:
>             >
>             > > Hi, all
>             > >
>             > > I run into some issue when using Nef_polyhedron_3 with
>             > > Exact_predicates_exact_constructions_kernel_with_sqrt
>         Kernel.
>             > >
>             > > To be specific, here is the simple program I was testing:
>             > >
>             > > #include
>            <CGAL/Exact_predicates_exact_constructions_kernel_with_sqrt.h>
>             > > #include <CGAL/Polyhedron_3.h>
>             > > #include <CGAL/Nef_polyhedron_3.h>
>             > >
>             > > typedef
>         CGAL::Exact_predicates_exact_constructions_kernel_with_sqrt
>             > > Kernel;
>             > > typedef Kernel::Point_3                              
>         Point_3;
>             > > typedef CGAL::Polyhedron_3<Kernel>         Polyhedron;
>             > > typedef CGAL::SNC_indexed_items Items;
>             > > typedef CGAL::Nef_polyhedron_3<Kernel, Items>
>          Nef_polyhedron;
>             > >
>             > >
>             > > int main() {
>             > >     Point_3 p( 1.0, 0.0, 0.0);
>             > >     Point_3 q( 0.0, 1.0, 0.0);
>             > >     Point_3 r( 0.0, 0.0, 1.0);
>             > >     Point_3 s( 0.0, 0.0, 0.0);
>             > >
>             > >     Polyhedron P;
>             > >     P.make_tetrahedron( p, q, r, s);
>             > >     Nef_polyhedron N2(P);
>             > >
>             > >     return 0;
>             > > }
>             > >
>             > >
>             > > I am using Visual Studio 2003 and Visual Studio 2005
>         with CGAL
>            3.3.1. <http://3.3.1.> <http://3.3.1.>
>             > > <http://3.3.1.> When runing in Release mode, everything runs
>            fine. But
>             > > when running in Debug mode, I got the following errors:
>             > >
>             > > Is this a bug in CGAL or did I do something wrong? I can
>         not figure
>             > > this out, can anybody help me out?
>             > >
>             > > Joe
>             > >
>             > >
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(300):
>             > > error C2064: term does not evaluate to a function taking 3
>            arguments
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(305):
>             > > error C2064: term does not evaluate to a function taking 3
>            arguments
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(310):
>             > > error C2064: term does not evaluate to a function taking 3
>            arguments
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(315):
>             > > error C2064: term does not evaluate to a function taking 3
>            arguments
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(323):
>             > > error C2440: 'type cast' : cannot convert from
>             > > 'std::allocator<_Ty>::value_type' to
>             > > 'CGAL::Cartesian_base_no_ref_count<FT_,Kernel_>::FT'
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > >         and
>             > >         [
>             > >             FT_=CORE::Expr,
>             > >             Kernel_=CGAL::Simple_cartesian<CORE::Expr>
>             > >         ]
>             > >         No constructor could take the source type, or
>         constructor
>             > > overload resolution was ambiguous
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(323):
>             > > error C2440: 'type cast' : cannot convert from
>             > > 'std::allocator<_Ty>::value_type' to
>             > > 'CGAL::Cartesian_base_no_ref_count<FT_,Kernel_>::FT'
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > >         and
>             > >         [
>             > >             FT_=CORE::Expr,
>             > >             Kernel_=CGAL::Simple_cartesian<CORE::Expr>
>             > >         ]
>             > >         No constructor could take the source type, or
>         constructor
>             > > overload resolution was ambiguous
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(324):
>             > > error C2440: 'type cast' : cannot convert from
>             > > 'std::allocator<_Ty>::value_type' to
>             > > 'CGAL::Cartesian_base_no_ref_count<FT_,Kernel_>::FT'
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > >         and
>             > >         [
>             > >             FT_=CORE::Expr,
>             > >             Kernel_=CGAL::Simple_cartesian<CORE::Expr>
>             > >         ]
>             > >         No constructor could take the source type, or
>         constructor
>             > > overload resolution was ambiguous
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(324):
>             > > error C2440: 'type cast' : cannot convert from
>             > > 'std::allocator<_Ty>::value_type' to
>             > > 'CGAL::Cartesian_base_no_ref_count<FT_,Kernel_>::FT'
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > >         and
>             > >         [
>             > >             FT_=CORE::Expr,
>             > >             Kernel_=CGAL::Simple_cartesian<CORE::Expr>
>             > >         ]
>             > >         No constructor could take the source type, or
>         constructor
>             > > overload resolution was ambiguous
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(306):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(307):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(308):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(309):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(311):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(312):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(313):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(314):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(316):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(317):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(318):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program
>         Files\CGAL-3.3.1\include\CGAL\Nef_S2\Normalizing.h(319):
>             > > error C2676: binary '*=' : 'std::allocator<_Ty>::value_type'
>            does not
>             > > define this operator or a conversion to a type
>         acceptable to the
>             > > predefined operator
>             > >         with
>             > >         [
>             > >             _Ty=CGAL::Fraction_traits<FT>::Numerator_type
>             > >         ]
>             > > C:\Program Files\CGAL-3.3.1\include\CGAL\float.h(133):
>         warning
>            C4244:
>             > > 'return' : conversion from 'double' to
>             > >
>         'CGAL::Algebraic_structure_traits_base<Type,Algebra_type>::Type',
>             > > possible loss of data
>             > >         with
>             > >         [
>             > >             Type=float,
>             > >             Algebra_type=CGAL::Field_with_kth_root_tag
>             > >         ]
>             > > C:\Program Files\CGAL-3.3.1\include\CGAL\Handle.h(85):
>         warning
>            C4311:
>             > > 'reinterpret_cast' : pointer truncation from 'CGAL::Rep
>         *const ' to
>             > > 'unsigned long'
>             > > C:\Program Files\CGAL-3.3.1\include\CGAL\Handle.h(90):
>         warning
>            C4311:
>             > > 'reinterpret_cast' : pointer truncation from 'CGAL::Rep
>         *const ' to
>             > > 'unsigned long'
>             > > C:\Program Files\CGAL-3.3.1\include\CGAL\Handle.h(91):
>         warning
>            C4311:
>             > > 'reinterpret_cast' : pointer truncation from 'CGAL::Rep
>         *const ' to
>             > > 'unsigned long'
>             > > C:\Program Files\Microsoft Visual Studio .NET
>             > > 2003\Vc7\include\xtree(484): warning C4800: 'int' : forcing
>            value to
>             > > bool 'true' or 'false' (performance warning)
>             > > C:\Program Files\Microsoft Visual Studio .NET
>             > > 2003\Vc7\include\xtree(484): warning C4800: 'int' : forcing
>            value to
>             > > bool 'true' or 'false' (performance warning)
>             > >
>             > >
>             > >
>             > >
>             >
>
>            --
>            You are currently subscribed to cgal-discuss.
>            To unsubscribe or access the archives, go to
>            https://lists-sop.inria.fr/wws/info/cgal-discuss
>
>
>
>     -- 
>     You are currently subscribed to cgal-discuss.
>     To unsubscribe or access the archives, go to
>     https://lists-sop.inria.fr/wws/info/cgal-discuss