From drichie at hpti.com  Tue Apr 27 15:02:15 2004
From: drichie at hpti.com (david richie)
Date: Tue Apr 27 16:03:38 2004
Subject: [Ohmms-dev] AntiSymTensor
Message-ID: <003601c42c9a$ec8c4d20$0f01a8c0@yourw92p4bhlzg>

Should AntiSymTensor and SymTensor behave
like Tensor when used in ParticleAttirb<>?

Specifically,  consider code fragment:

   typedef Tensor<double,3> Tensor_t;
   typedef AntiSymTensor<double,3> AntiSymTensor_t;

   ParticleAttrib<Tensor_t> tensor;
   ParticleAttrib<AntiSymTensor_t> antisymtensor;

   t = 0.0e0; 
   antisymtensor = 0.0e0;

The last assignment triggers error.  Above is a mock version
of problem.  I can forward exact source if necessary.  Can
you verify that above should work?

Compile-time error is particularly nasty, but maybe it will
shed some light.  This is real error, not from toy problem above.
In real example I am using AntiSymTensor for torque.  When I
switch Torque_t to a simple Tensor<> compiles fine (but torque
is not a tensor, so not a solution).

g++ -DHAVE_CONFIG_H -I. -I../../src -I. -I../../src/OOMPI    -ftemplate-depth-60 -DADD_  -I/usr/include/libxml2 -c -o FFHydroSDPotential.o `test -f 'StokesianDynamics/FFHydroSDPotential.cpp' || echo '../../src/'`StokesianDynamics/FFHydroSDPotential.cpp
../../src/PETE/OperatorTags.h: In member function `BinaryReturn<T1, T2,
   OpAssign>::Type_t OpAssign::operator()(const T1&, const T2&) const [with T1
   = AntiSymTensor<Scalar_t, 3>, T2 = double]':
../../src/ParticleBase/ParticleAttrib.cpp:1135:   instantiated from `void evaluate(ParticleAttrib<T1>&, const Op&, const Expression<T2>&) [with T = AntiSymTensor<Scalar_t, 3>, Op = OpAssign, RHS = Scalar<double>]'
../../src/ParticleBase/ParticleAttrib.cpp:974:   instantiated from `ParticleAttrib<T1>& assign(ParticleAttrib<T1>&, const RHS&) [with T1 = AntiSymTensor<Scalar_t, 3>, RHS = double]'
../../src/ParticleBase/ParticleAttrib.h:139:   instantiated from `ParticleAttrib<T>& ParticleAttrib<T>::operator=(const RHS&) [with RHS = double, T = AntiSymTensor<Scalar_t, 3>]'
../../src/StokesianDynamics/FFHydroSDPotential.cpp:57:   instantiated from here
../../src/PETE/OperatorTags.h:757: could not convert `(++a)->AntiSymTensor<T,
   D>::operator=(const T&) [with T = Scalar_t, unsigned int D = 3]((+b))' to `
   AntiSymTensor<Scalar_t, 3>&'



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mcc0.mcc.uiuc.edu/pipermail/ohmms-dev/attachments/20040427/fb815894/attachment.html
From jnkim at ncsa.uiuc.edu  Tue Apr 27 15:15:08 2004
From: jnkim at ncsa.uiuc.edu (Jeongnim Kim)
Date: Tue Apr 27 16:15:11 2004
Subject: [Ohmms-dev] AntiSymTensor
In-Reply-To: <003601c42c9a$ec8c4d20$0f01a8c0@yourw92p4bhlzg>
Message-ID: <5.1.0.14.2.20040427161044.023b8a78@pop.ncsa.uiuc.edu>

Let me know if you can compile

antisymtesor[0] = 0.0;

At 05:02 PM 4/27/2004 -0400, david richie wrote:
>Should AntiSymTensor and SymTensor behave
>like Tensor when used in ParticleAttirb<>?
>
>Specifically,  consider code fragment:
>
>    typedef Tensor<double,3> Tensor_t;
>    typedef AntiSymTensor<double,3> AntiSymTensor_t;
>
>    ParticleAttrib<Tensor_t> tensor;
>    ParticleAttrib<AntiSymTensor_t> antisymtensor;
>
>    t = 0.0e0;
>    antisymtensor = 0.0e0;
>
>The last assignment triggers error.  Above is a mock version
>of problem.  I can forward exact source if necessary.  Can
>you verify that above should work?
>
>Compile-time error is particularly nasty, but maybe it will
>shed some light.  This is real error, not from toy problem above.
>In real example I am using AntiSymTensor for torque.  When I
>switch Torque_t to a simple Tensor<> compiles fine (but torque
>is not a tensor, so not a solution).
>
>g++ -DHAVE_CONFIG_H -I. -I../../src -I. 
>-I../../src/OOMPI    -ftemplate-depth-60 -DADD_  -I/usr/include/libxml2 -c 
>-o FFHydroSDPotential.o `test -f 
>'StokesianDynamics/FFHydroSDPotential.cpp' || echo 
>'../../src/'`StokesianDynamics/FFHydroSDPotential.cpp
>../../src/PETE/OperatorTags.h: In member function `BinaryReturn<T1, T2,
>    OpAssign>::Type_t OpAssign::operator()(const T1&, const T2&) const 
> [with T1
>    = AntiSymTensor<Scalar_t, 3>, T2 = double]':
>../../src/ParticleBase/ParticleAttrib.cpp:1135:   instantiated from `void 
>evaluate(ParticleAttrib<T1>&, const Op&, const Expression<T2>&) [with T = 
>AntiSymTensor<Scalar_t, 3>, Op = OpAssign, RHS = Scalar<double>]'
>../../src/ParticleBase/ParticleAttrib.cpp:974:   instantiated from 
>`ParticleAttrib<T1>& assign(ParticleAttrib<T1>&, const RHS&) [with T1 = 
>AntiSymTensor<Scalar_t, 3>, RHS = double]'
>../../src/ParticleBase/ParticleAttrib.h:139:   instantiated from 
>`ParticleAttrib<T>& ParticleAttrib<T>::operator=(const RHS&) [with RHS = 
>double, T = AntiSymTensor<Scalar_t, 3>]'
>../../src/StokesianDynamics/FFHydroSDPotential.cpp:57:   instantiated from 
>here
>../../src/PETE/OperatorTags.h:757: could not convert `(++a)->AntiSymTensor<T,
>    D>::operator=(const T&) [with T = Scalar_t, unsigned int D = 3]((+b))' 
> to `
>    AntiSymTensor<Scalar_t, 3>&'
>
>
>_______________________________________________
>Ohmms-dev mailing list
>Ohmms-dev@mcc.uiuc.edu
>http://mcc0.mcc.uiuc.edu/mailman/listinfo/ohmms-dev

Best wishes,
Jeongnim Kim

---------------------------------------------------------------
Jeongnim Kim                           e-mail: jnkim@uiuc.edu
NCSA                                   Tel: 217-244-6319 (NCSA)
Beckman Institute                           217-244-4971 (MCC)
405 N Mathews Av.                      Fax: 217-244-2909
Urbana, IL 61801
---------------------------------------------------------------

From drichie at hpti.com  Wed Apr 28 14:45:56 2004
From: drichie at hpti.com (david richie)
Date: Wed Apr 28 15:46:10 2004
Subject: [Ohmms-dev] exp() and ParticleAttrib<TinyVector<double,3> >
Message-ID: <006001c42d61$da4a5a40$0f01a8c0@yourw92p4bhlzg>

does there exist support for using exp() function on ParticleAttirb<> container?
If not, can you suggest closest function to use as starting point in OhmmsPETE/

Example:

   typedef TinyVector<double,3> Position_t;
   typedef ParticleAttrib<Position_t> ParticlePosition_t;
   double alpha = 0.9;
   ParticlePosition_t& R1,R2;

   R1 = something
   R2 = exp(-R1/alpha);

???

-David


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mcc0.mcc.uiuc.edu/pipermail/ohmms-dev/attachments/20040428/853a575a/attachment.html