dolfin team mailing list archive
-
dolfin team
-
Mailing list archive
-
Message #16699
Re: New Expression syntax
Johan Hake wrote:
> On Friday 27 November 2009 14:44:12 Garth N. Wells wrote:
>> Johan Hake wrote:
>>> On Friday 27 November 2009 11:53:00 Garth N. Wells wrote:
>>>> Johan Hake wrote:
>>>>> On Friday 27 November 2009 03:02:08 Anders Logg wrote:
>>>>>> On Fri, Nov 27, 2009 at 09:41:03AM +0000, Garth N. Wells wrote:
>>>>>>> Anders Logg wrote:
>>>>>>>> On Fri, Nov 27, 2009 at 08:37:08AM +0000, Garth N. Wells wrote:
>>>>>>>>> Is it still possible with the new Expression syntax to do something
>>>>>>>>> like:
>>>>>>>>>
>>>>>>>>> f2, f3 = Expressions("sin(3.0*x[0])*sin(3.0*x[1])*sin(3.0*x[2])",
>>>>>>>>> "1.0 + 3.0*x[0] + 4.0*x[1] + 0.5*x[2]", V=V)
>>>>>>>>>
>>>>>>>>> The above is used in the unit tests. Or to keep things simple
>>>>>>>>> should we remove this kind of fancy feature?
>>>>>>>> I guess it's a nice feature to have but it's nothing I care that
>>>>>>>> much for.
>>>>>>> All tests except the unit test for function are working. Expression
>>>>>>> needs to be updated for some of the fancier syntax.
>>>>>>>
>>>>>>> Garth
>>>>>> I made a very simple fix in expression.py to get Expressions()
>>>>>> working.
>>>>>>
>>>>>> It currently createas a separate Expression class for each argument so
>>>>>> it will call compile_expressions() for each one, which is less
>>>>>> efficient than compiling a bunch of expressions at once.
>>>> Doesn't seem to work when I tried.
>>>>
>>>>> The whole point of using Expressions, is to avoid double compilations
>>>>> :)
>>>>>
>>>>>> Johan, could you have a look at this function?
>>>>>>
>>>>>> I also commented out the unit test testSameBases since I couldn't
>>>>>> figure out what it's supposed to do or if it is still valid after the
>>>>>> interface changes.
>>>>>>
>>>>>> Please have a look.
>>>>> Will do!
>>>> I had a look, but haven't committed anything yet.
>>> Commit anything if you like. I have time to look at it this evening
>>> (night for you ;))
>> I'll leave it to you because expression.py is leaving me confused,
>> namely why does Expression involve the metaclass ExpressionMetaClass,
>> but Expressions doesn't? As a result why do we want the __init__
>> function defined inside create_expression_class to be called for an
>> Expression but not the expressions in an Expressions?
>
> Ok, I'll look at it.
>
> create_expression_class is called in Expressions. It is used to create the
> full fledged Expression class that inherits, dolfin.Expression, ufl.Function
> and the newly compiled class that inherits cpp.Expression.
>
It would help me if someone could tell when when the __init__ function
defined inside function 'create_expression_class' is called. When I do
f1 = Expression("2", degree=2)
the __init__ function is called, but when I do
f2,f3 = Expressions("2", "1", degree=2)
the __init__ function is not called.
Garth
> Whenever a class is created that either is or inherits dolfin.Expression, the
> ExpressionMetaClass.__new__ is employed. However we do not want to make any
> costumizations in classes returned by create_expression_class, which means we
> just want to return the class as it is passed to the
> ExpressionMetaClass.__new__ method. Therefore I have the lines 213-217, which
> just do that.
>
> Put
>
> print name, bases
>
> first in the ExpressionMetaClass.__new__ method and try to create some user
> defined Expressions, and you might get a clue of what is happening.
>
> Johan
>
>> There is a lot of magic going on which makes it hard to follow.
>>
>> Garth
>>
>>> Johan
>>>
>>>> Garth
>>>>
>>>>> Johan
>>>>>
>>>>>> --
>>>>>> Anders
>> _______________________________________________
>> Mailing list: https://launchpad.net/~dolfin
>> Post to : dolfin@xxxxxxxxxxxxxxxxxxx
>> Unsubscribe : https://launchpad.net/~dolfin
>> More help : https://help.launchpad.net/ListHelp
>>
Follow ups
References