From rguenth at tat.physik.uni-tuebingen.de  Fri Oct 18 13:00:23 2002
From: rguenth at tat.physik.uni-tuebingen.de (Richard Guenther)
Date: Fri, 18 Oct 2002 15:00:23 +0200 (CEST)
Subject: [BUG] FieldEngine::makeOwnCopy() does not compile
Message-ID: <Pine.LNX.4.33.0210181458420.28219-100000@bellatrix.tat.physik.uni-tuebingen.de>

Hi!

FieldEngine::makeOwnCopy() does not compile and I cant figure out how the
current "structure" of it should work. Can anybody enlighten me or fix the
problem?

Thanks, Richard.

--
Richard Guenther <richard.guenther at uni-tuebingen.de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
The GLAME Project: http://www.glame.de/



From rguenth at tat.physik.uni-tuebingen.de  Mon Oct 21 10:21:31 2002
From: rguenth at tat.physik.uni-tuebingen.de (Richard Guenther)
Date: Mon, 21 Oct 2002 12:21:31 +0200 (CEST)
Subject: [PATCH] Re: [pooma-dev] [BUG] FieldEngine::makeOwnCopy() does not
 compile
In-Reply-To: <Pine.LNX.4.33.0210181458420.28219-100000@bellatrix.tat.physik.uni-tuebingen.de>
Message-ID: <Pine.LNX.4.33.0210211220190.15835-100000@bellatrix.tat.physik.uni-tuebingen.de>

On Fri, 18 Oct 2002, Richard Guenther wrote:

> FieldEngine::makeOwnCopy() does not compile and I cant figure out how the
> current "structure" of it should work. Can anybody enlighten me or fix the
> problem?

Ok, I needed this so I sat down and at least fixed it (probably not in an
optimal way, but it seems to work). Apply the following

Index: src/Field/FieldEngine/FieldEngine.h
===================================================================
RCS file: /home/pooma/Repository/r2/src/Field/FieldEngine/FieldEngine.h,v
retrieving revision 1.3
diff -u -r1.3 FieldEngine.h
--- src/Field/FieldEngine/FieldEngine.h	1 Jul 2002 22:25:53 -0000	1.3
+++ src/Field/FieldEngine/FieldEngine.h	21 Oct 2002 10:20:11 -0000
@@ -587,6 +587,10 @@
   {
     PAssert(data_m.isValid());

+    // Remember data_m as model
+
+    RefCountedBlockPtr<Data_t> model = data_m;
+
     // Create a blank slate of engines:

     data_m = RefCountedBlockPtr<Data_t>();
@@ -600,7 +604,7 @@
     {
       for (int c = 0; c < centering_m.size(); ++ c)
       {
-        data(m, c) = model.data(m, c);
+        data(m, c) = model[m*stride_m + c];
         data(m, c).engine().makeOwnCopy();
         data(m, c).relations().makeOwnCopy(s);
       }



From rguenth at tat.physik.uni-tuebingen.de  Thu Oct 24 11:51:32 2002
From: rguenth at tat.physik.uni-tuebingen.de (Richard Guenther)
Date: Thu, 24 Oct 2002 13:51:32 +0200 (CEST)
Subject: Missing ::layout() const in certain Engines
Message-ID: <Pine.LNX.4.33.0210241348060.10183-100000@bellatrix.tat.physik.uni-tuebingen.de>

Hi!

Is there something why f.i. Engine< .. IndexFunction .. > or
Engine< .. ApplyFieldStencil<> .. > dont have a ::layout() const
method? This prevents me from f.e. writing

  a(I) = stencil1(stencil2(b))(I);

(doesnt compile, complains about missing ::layout()). If I just add

  inline DomainLayout<Dim> layout() const { return DomainLayout<Dim>(domain_m); }

to src/Field/DiffOps/FieldStencil.h::Engine<Dim, T, ApplyFieldStencil<Functor, Expression> >
this compiles and seems to work as expected (need to do a formal testcase,
though).

Any hints to something I'm missing?

Richard.

--
Richard Guenther <richard.guenther at uni-tuebingen.de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
The GLAME Project: http://www.glame.de/



From rguenth at tat.physik.uni-tuebingen.de  Thu Oct 24 11:57:55 2002
From: rguenth at tat.physik.uni-tuebingen.de (Richard Guenther)
Date: Thu, 24 Oct 2002 13:57:55 +0200 (CEST)
Subject: [pooma-dev] Missing ::layout() const in certain Engines
In-Reply-To: <Pine.LNX.4.33.0210241348060.10183-100000@bellatrix.tat.physik.uni-tuebingen.de>
Message-ID: <Pine.LNX.4.33.0210241355580.10183-100000@bellatrix.tat.physik.uni-tuebingen.de>

On the same thing, missing ::guardLayers() for the expression engine
prevents

  a(I) = stencil(5.0*a)(I);

Are these just oversights? Is there a design document specifying methods
all engines are required to implement? The same for layout and domain
objects? If not, this seems to be badly needed, as well as tests to check
the availability of those.

Richard.

--
Richard Guenther <richard.guenther at uni-tuebingen.de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
The GLAME Project: http://www.glame.de/



From jcrotinger at proximation.com  Thu Oct 24 17:00:33 2002
From: jcrotinger at proximation.com (James Crotinger)
Date: Thu, 24 Oct 2002 11:00:33 -0600
Subject: [pooma-dev] Missing ::layout() const in certain Engines
Message-ID: <F871BB04B0EA224A821A5274C31DF31509456E@proxim-sf001.santafe.proximation.com>

Hi Richard.

I suspect that some of the things you've mentioned would work fine with
Bricks, but would be difficult to make work with multi-patch engines and
Fields. Also, the team had various notes on things like Engine requirements,
but these were never turned into formal documents and are probably lost.

	Jim

-----Original Message-----
From: Richard Guenther [mailto:rguenth at tat.physik.uni-tuebingen.de] 
Sent: Thursday, October 24, 2002 5:58 AM
To: pooma-dev at pooma.codesourcery.com
Subject: Re: [pooma-dev] Missing ::layout() const in certain Engines

On the same thing, missing ::guardLayers() for the expression engine
prevents

  a(I) = stencil(5.0*a)(I);

Are these just oversights? Is there a design document specifying methods
all engines are required to implement? The same for layout and domain
objects? If not, this seems to be badly needed, as well as tests to check
the availability of those.

Richard.

--
Richard Guenther <richard.guenther at uni-tuebingen.de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
The GLAME Project: http://www.glame.de/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sourcerytools.com/pipermail/pooma-dev/attachments/20021024/a49e0bbc/attachment.html>

From rguenth at tat.physik.uni-tuebingen.de  Fri Oct 25 08:00:38 2002
From: rguenth at tat.physik.uni-tuebingen.de (Richard Guenther)
Date: Fri, 25 Oct 2002 10:00:38 +0200 (CEST)
Subject: POOMA Copyright assignment
In-Reply-To: <75870000.1029315693@warlock.codesourcery.com>
Message-ID: <Pine.LNX.4.33.0210250949300.10183-100000@bellatrix.tat.physik.uni-tuebingen.de>

Hi!

I finally got answers from all involved parties regarding to the copyright
assignment to CodeSourcery and to distributing parts of my work in
general. I'm sorry to say that neither of the involved parties likes to
fill out the assignment - policy is that work done by me can be
distributed (for free) only for scientific/educational/research purposes.

I'll continue to submit bugfixing and small enhancement patches to the
development list as I can just do those in my spare time. For those
(usually one liners) I'm not going to sign anything, too. People at
CodeSourcery can take them at their own risk or ignore them, but obviously
they could be of value to other users. (*)

I'll stop posting major enhancement patches as I'm not legally allowed to
distribute them. I continue announcing the enhancements so if you're a
scientist you may contact me privately and I'll send along patches.

Oh well - politics, Richard.

(*) I'll state for every such small patch that I'll put it into the public
domain. As you usually can't re-invent a one-line bugfix (like a typo) I
hope by posting such thing I dont prevent CodeSourcery from fixing such a
bug "themselves" ;)

--
Richard Guenther <richard.guenther at uni-tuebingen.de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
The GLAME Project: http://www.glame.de/



From rguenth at tat.physik.uni-tuebingen.de  Fri Oct 25 08:10:12 2002
From: rguenth at tat.physik.uni-tuebingen.de (Richard Guenther)
Date: Fri, 25 Oct 2002 10:10:12 +0200 (CEST)
Subject: [pooma-dev] Missing ::layout() const in certain Engines
In-Reply-To: <F871BB04B0EA224A821A5274C31DF31509456E@proxim-sf001.santafe.proximation.com>
Message-ID: <Pine.LNX.4.33.0210251009110.10183-100000@bellatrix.tat.physik.uni-tuebingen.de>

On Thu, 24 Oct 2002, James Crotinger wrote:

> Hi Richard.
>
> I suspect that some of the things you've mentioned would work fine with
> Bricks, but would be difficult to make work with multi-patch engines and
> Fields.

Duh - you are right, my simple solution doesnt work for MultiPatch
engines.

Richard.

--
Richard Guenther <richard.guenther at uni-tuebingen.de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
The GLAME Project: http://www.glame.de/



From rguenth at tat.physik.uni-tuebingen.de  Mon Oct 28 12:59:44 2002
From: rguenth at tat.physik.uni-tuebingen.de (Richard Guenther)
Date: Mon, 28 Oct 2002 13:59:44 +0100 (CET)
Subject: How to apply a functor to ET leafs
Message-ID: <Pine.LNX.4.33.0210281353280.10183-100000@bellatrix.tat.physik.uni-tuebingen.de>

Hi!

I'd like to apply a functor to the parts of an expression and view
the result as an expression. I suspect I can do this somehow using
forEach(), but before I dig too far, I just ask if somebody knows
the answer already. Take

  Field_t A, B, C;
  Interval I;

  C(I) = forEach(A*B, evalAt(0.5), ?)(I);

which should "evaluate" A and B at 0.5 and then just apply the usual
expression for interval I.

Can this be done?

Richard.

--
Richard Guenther <richard.guenther at uni-tuebingen.de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
The GLAME Project: http://www.glame.de/



From rguenth at tat.physik.uni-tuebingen.de  Mon Oct 28 13:39:34 2002
From: rguenth at tat.physik.uni-tuebingen.de (Richard Guenther)
Date: Mon, 28 Oct 2002 14:39:34 +0100 (CET)
Subject: [pooma-dev] How to apply a functor to ET leafs
In-Reply-To: <Pine.LNX.4.33.0210281353280.10183-100000@bellatrix.tat.physik.uni-tuebingen.de>
Message-ID: <Pine.LNX.4.33.0210281425210.10183-100000@bellatrix.tat.physik.uni-tuebingen.de>

On Mon, 28 Oct 2002, Richard Guenther wrote:

> Hi!
>
> I'd like to apply a functor to the parts of an expression and view
> the result as an expression. I suspect I can do this somehow using
> forEach(), but before I dig too far, I just ask if somebody knows
> the answer already. Take
>
>   Field_t A, B, C;
>   Interval I;
>
>   C(I) = forEach(A*B, evalAt(0.5), ?)(I);
>
> which should "evaluate" A and B at 0.5 and then just apply the usual
> expression for interval I.

To follow up myself I created

struct evalAt {
  evalAt(double x) : x_m(x) {}
  const double x_m;
};

template <class T>
struct LeafFunctor<T, evalAt >
{
  typedef double Type_t;
  inline static Type_t apply(const T &p, const evalAt &x)
  {
    return peval(p, x.x_m);
  }
};

and use now

  C(I) = forEach((A*B).engine().expression(), evalAt(0.5), TreeCombine())(I);

but of course I cannot take a view of an expression. So how do I
re-construct a Field<,, ExpressionTag<>> from the forEach result? It needs
to match the A*B expression obviously. Also just using

  C(I) = forEach((A(I)*B(I)).engine().expression(), evalAt(0.5), TreeCombine());

doesnt work, as evalAt gets called with Field<> as T, not with the actual
elements it seems.

I'm lost here, Richard.

--
Richard Guenther <richard.guenther at uni-tuebingen.de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
The GLAME Project: http://www.glame.de/



From rguenth at tat.physik.uni-tuebingen.de  Wed Oct 30 09:37:18 2002
From: rguenth at tat.physik.uni-tuebingen.de (Richard Guenther)
Date: Wed, 30 Oct 2002 10:37:18 +0100 (CET)
Subject: Revised Q: How to apply a Stencil to all leafs in an expression
Message-ID: <Pine.LNX.4.33.0210301030560.10183-100000@bellatrix.tat.physik.uni-tuebingen.de>

Hi!

To make my point in my previous messages more clear (just thought about
what I'm actually trying to do), I like to apply a (Field)Stencil to all
leafs (of Engine type, but just ignore this for now) in an expression.
I.e. instead of writing

  A(I) = (stencil(B)*stencil(C))(I);

I'd like to write something like

  A(I) = applyStencil(B*C)(I);

for brevity (I really have huge expressions, they get completely
unreadable otherwise).

I know this has to be possible somehow, but I'm stuck at
 a) Creating the Field type resulting from stencil(leaf)
 b) how to use forEach() in conjunction with TreeCombine to stick
    the result from a) in place of the leafs.

Can anybody point me to the right traits classes that create the
necessary types for me? I tried to find out where the Field type
is constructed for a simple expression like A = B*C, but I failed
here -- need some 'interactive C++ code explorer' that handles
templates...

Thanks, Richard.

--
Richard Guenther <richard.guenther at uni-tuebingen.de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
The GLAME Project: http://www.glame.de/