On Tue, May 27, 2008 at 5:04 PM, Friedrich Hagedorn <[EMAIL PROTECTED]> wrote:
>
> On Tue, May 27, 2008 at 04:28:22PM +0200, Ondrej Certik wrote:
>>
>> On Tue, May 27, 2008 at 3:56 PM, Friedrich Hagedorn <[EMAIL PROTECTED]>
>> wrote:
>> >
>> > On Tue, May 27, 2008 at 05:22:54AM -0700, astozzia wrote:
>> >>
>> >> evec=dvecM/sqrt(s.dot(dvec,dvec))
>> >>
>> >> Note that I had to use the 'Matrix' and divide my the norm of the
>> >> 'list'. Kinda weird...
>> >
>> > The problem is that we dont have a true vector support in sympy yet.
>> > Specialy you cant define a function of a matrix/vector, thats bad.
>> >
>> > The derivation of
>> >
>> > In [4]: vec = s.array([cos(phi(t)),x(t),z(t)])
>> > In [5]: dvec = diff(vec,t)
>> >
>> > is simply done by a 'elementwise' derivation of the single elements
>> > of the vec array. From the mathematical point of view I think this
>> > isnt ideal. The 'diff(vec, t)' should better wrapped to something
>> > like the jacobion matrix.
>> >
>> > We had a long discussion an sympy-patches
>> >
>> >
>> > http://groups.google.com/group/sympy-patches/browse_thread/thread/78bf61123b3d7405
>> >
>> > where I had sum up my opinion. But you know the time is raw ... :-)
>>
>> I think simply one has to write a patch for that and if it's good it will go
>> in.
>>
>> > And I am not sure if its so good to do this (vector extension) in sympy
>> > because I think the sympycore project had a better
>> > (mathematical) structure for that.
>>
>> Unfortunately, sympycore is a different project not directly usable by
>> sympy (we may include it in thirdparty, but that's still not the
>> solution imho, but at least we can then easily use one or the other,
>> before someone will merge things completely, which unfortunately, will
>> take much longer).
>
> I guessed thats this would be the real hard work :-(
>
>> However, I have always been all for merging our
>> efforts. Could you please summarize which structure in sympycore is
>> better? Let's use the same way in sympy.
>
> Ok, just a short try do explain my fuzzy imagination about this. Its more
> an emotion to this :-)
>
> I like the (abstract) algebra idea. You have
> expressions (operations) of symbols. And when you eval an expression
> the operations must defined for the symbol types like
>
> Numbers,
> Functions,
> Vector, Matrices and
> of curse (sub-)expressions.
>
> And for all the things there are exists different algebras to define
> the operations.
>
> I have no idea how to implement this in sympy (ok I am relativ new here)
>
>>>> M = Matrix(1,3,[x,y,z])
>>>> f = Function('f')
>>>> f(M)
>
> f(M)
>
> without isinstance(...) because the argument should be a symbol and not
> a Matrix ?!? But may be I am a little bit confused and the the things
> are not so different in sympy/sympycore as I thought ;-)
How would you do it differently? Currently this blows up here:
109 if isinstance(a, Matrix):
--> 110 raise NotImplementedError('matrix support')
Because nonone has implemented unknown functions on matrices yet, but
I don't see any other problem here. Or is here some other pitfall?
>
>> I read that discussion in
>> sympy-patches ---- I am not sure what exactly you propose, and which
>> things you agreed with Basti and which not. So let's get things
>> moving.
>
> I meant the least message (number 28 from 7.5. 12:53). I didnt know
> how to link to a single post in a thread.
More options -> Individual message:
http://groups.google.com/group/sympy-patches/msg/24beaf720dd4ae03
>
>> Basically, a good measure how to determine if something is
>> useful/worthy is just to write a patch, try if all tests pass and if
>> they do, and the patch is maintainable, it will go in. You don't have
>> to write a real patch, just think about things in this way.
>
> I slowly get an idea of your develoment strategy :-)
If you have a better strategy, please share it with us. :)
Ondrej
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"sympy" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/sympy?hl=en
-~----------~----~----~----~------~----~------~--~---
