On 12 November 2011 20:43, Aaron Meurer <[email protected]> wrote:
> On Fri, Nov 11, 2011 at 4:25 PM, [email protected]
> <[email protected]> wrote:
> >
> >
> > On 11 November 2011 23:31, Aaron Meurer <[email protected]> wrote:
> >>
> >> On Fri, Nov 11, 2011 at 2:13 PM, [email protected]
> >> <[email protected]> wrote:
> >> > Hi
> >> >
> >> > I'm a bit ashamed that Aaron gave better introduction than me for my
> own
> >> > code :)
> >> > Anyway, I think the documentation of Plot and the module is quite
> >> > detailed
> >> > so you can look also at help(Plot) anhelp(newplot).
> >> >
> >> > About the warnings - my idea was to structure the base backend class
> in
> >> > such
> >> > a way that any missing functionality in the backend subclass will just
> >> > raise
> >> > a warning but not an error (unless it is some essential
> functionality).
> >> > So
> >> > those will be addressed later.
> >> >
> >> > It can plot Integrals (due to an old addition done to lambdify thanks
> to
> >> > Certik) but not Sums or anything fancy like product of Kets and Bras.
> To
> >> > do
> >> > those a more in depth refactoring of lambda will be needed as Certik's
> >> > method for adding Integral to lambdify does not scale well. I think
> this
> >> > is
> >> > an important problem.
> >>
> >> I agree. Is there an issue for this?
> >
> > No issue for the moment. I take responsibility for creating one when I
> > gather enough understanding of the code and the problem.
> >>
> >> >
> >> > Discontinuous functions may pose problems for the moment (none seen
> for
> >> > the
> >> > moment, but this is mostly by chance).
> >>
> >> As far as I can tell, this is a nontrivial problem to solve, as many
> >> very good plotting systems choke on discontinuous functions. Perhaps
> >> there can be some kind of symbolic heuristic applied to find
> >> discontinuities.
> >
> > I think Maple has some option in its plotting module about detecting
> > asymptotes (numerically?) but I can not check it at the moment. Anyhow,
> this
> > is not the most important feature at the moment.
> >>
> >> >
> >> > About the bug in matplotlib - Aaron, you said that you will make a
> pull
> >> > request for them. Should I do something or you have already taken care
> >> > of
> >> > this.
> >>
> >> Sorry for the misunderstanding, but this is not what I said. I was
> >> suggesting to you that you do it. I haven't even been able to
> >> reproduce the bug you are seeing, so I can't even tell with certainty
> >> what the correct fix is.
> >
> > I'm sorry :D (my mistake). I'll do it.
> >>
> >> >
> >> > I'll start writing tests for the module in the near future. Then the
> >> > core
> >> > devs should tell me if this code is going in and how.
> >>
> >> I definitely think it should go in. I guess the question is how to
> >> properly replace/merge it with the old module, which isn't compatible
> >> (it uses different syntax, keywords, etc.)
> >>
> >> Actually, to what degree is the new Plot() function incompatible with
> >> the old one? Would it be possible to merge the two without breaking
> >> compatibility with the old module?
> >
> > There are some things about the api of the old module that I don't like,
> but
> > I'll check if there is a compromise that can be made. I understand the
> > importance of not breaking the api.
>
> What are these specifically. Is it possible to support both APIs at
> the same time, while deprecating the old one, r would they have to be
> completely different functions to prevent ambiguous input?
>
About the parts that I don't like: Using a big string for all the options
and the possibility to use an arbitrary big index when adding new plot.
That's all, I think. Those two (especially the first one) would be
difficult to support.
About supporting both apis - I think it's possible. For example: when there
is an ambiguity it will default to the old module, and when it is using the
old module it will raise a deprecation warning. All this will be done from
a proxy class Plot that is in sympy.plotting. The old module will be in
sympy.plotting.pyglet. The deprecation warning will read:
"To use directly the pyglet module do: from sympy.plotting.pyglet import
PygletPlot. To use exclusively the new module (that has pyglet as one of
the possible backends) do: from sympy.plotting.future import Plot".
If this is ok, I'll do it in my pull request.
And the old module needs only to be moved, not removed (and a backend for
it needs to be written, but that would take under 2 hours (under a day or
two for a CGI student)).
One big advantage (the only one) of the old module over matplotlib is that
it does all in OpenGL so you can turn the graph in realtime (imagine a
teacher showing a hydrogen orbital in front of the class, matplotlib is not
fast enough).
> Also, even if you don't plan to have this ready by the next release,
> we can start deprecating the old module now.
>
I suppose that a warning wont hurt.
> The differences that I've noticed are that with the new module, you
> have to specify a variable and range (but assumedly, this will be
> relaxed at some point?), and that you have to call p.show() to
> actually see the plot. One idea regarding the last point would be to
> create a function plot() (lowercase) which automatically shows the
> plot, which would be used for quick plotting, and then leave Plot()
> for more advanced plots.
>
> About the variable with range: I was not thinking about relaxing the
requirement, but I suppose it would be a wanted feature. I'll write some
code for it. Maybe only in the plot() function (lower case). And yes, I
agree that a plot() function will be a good idea.
> Aaron Meurer
>
> >
> > But I was thinking it would be best to have both modules and then a
> backend
> > for the old module in the new module. And maybe move the old module to
> > plotting.oldplot or leave the new module in plotting.newplot.
> >>
> >> >
> >> > About the '3d' string - you are right it's a bad default. It's just
> that
> >> > contour was written first, but I'll change this now.
> >>
> >> Cool. By the way, will it be possible in the future to call something
> >> like p.change_plot_type('contour') (only with a better name than that
> >> :) and it will change it from 3d to a contour?
> >
> > At the moment 3dsurface (not parametric) and contour are represented by
> > different classes. It makes sense to merge them and add an option
> (actually
> > there is a repetition of code at the moment). It would be something like:
> > p = Plot(blah blah)
> > p[0].visualisation_method = '3d' or 'contour'
> > the name of the attribute may be other ('plot_type', 'projection', ...)
> >>
> >> >
> >> > Finally - I was squashing and rebasing this pull request quite a bit.
> >> > Now as
> >> > it's getting more attention I'll stop doing it, so you are free to
> make
> >> > changes if you are interested.
> >>
> >> Great. This makes it easier to follow your progress.
> >>
> >> Aaron Meurer
> >>
> >> >
> >> > On 11 November 2011 21:41, Aaron Meurer <[email protected]> wrote:
> >> >>
> >> >> Oh, I forgot to mention that I got the following warnings:
> >> >>
> >> >> In [6]: p = Plot(Heaviside(x)*(1 - x)*sin(y), (x, -1, 1), (y, -pi,
> pi))
> >> >>
> >> >> In [7]: p.show()
> >> >> /sw/lib/python2.7/site-packages/matplotlib/axes.py:4368: UserWarning:
> >> >> No labeled objects found. Use label='...' kwarg on individual plots.
> >> >> warnings.warn("No labeled objects found. "
> >> >>
> >> >> In [8]: p = Plot(Heaviside(x)*(1 - x)*sin(y), (x, -1, 1), (y, -pi,
> pi),
> >> >> '3d')
> >> >>
> >> >> In [9]: p.show()
> >> >>
> >> >>
> >> >>
> /Users/aaronmeurer/Documents/python/sympy/sympy/sympy/plotting/newplot.py:901:
> >> >> UserWarning: xscale is not supported in 3D matplotlib backend.
> >> >> warnings.warn('xscale is not supported in 3D matplotlib backend.')
> >> >>
> >> >>
> >> >>
> /Users/aaronmeurer/Documents/python/sympy/sympy/sympy/plotting/newplot.py:879:
> >> >> UserWarning: axis_center is not supported in 3D matplotlib backend.
> >> >> warnings.warn('axis_center is not supported in 3D matplotlib
> >> >> backend.')
> >> >>
> >> >>
> >> >>
> /Users/aaronmeurer/Documents/python/sympy/sympy/sympy/plotting/newplot.py:895:
> >> >> UserWarning: xscale is not supported in 3D matplotlib backend.
> >> >> warnings.warn('xscale is not supported in 3D matplotlib backend.')
> >> >>
> >> >> Aaron Meurer
> >> >>
> >> >> On Fri, Nov 11, 2011 at 1:39 PM, Aaron Meurer <[email protected]>
> >> >> wrote:
> >> >> > Hi.
> >> >> >
> >> >> > This looks great. For others, to run the examples, download the
> >> >> > examples script and put in the sympy directory. Then, checkout
> >> >> > Krastanov's branch (from the pull request). Then, run IPython, and
> >> >> > type %run examples.py. And then type p0.show(), p1.show(), etc.
> (up
> >> >> > to p4).
> >> >> >
> >> >> > And if you just want to test the plotting of your own functions in
> >> >> > isympy, you have to run "from sympy.plotting.newplot import *", or
> >> >> > else it will use the old plotting. The syntax is
> >> >> >
> >> >> > In [8]: p = Plot(Heaviside(x)*(1 - x)*sin(y), (x, -1, 1), (y, -pi,
> >> >> > pi),
> >> >> > '3d')
> >> >> >
> >> >> > In [9]: p.show()
> >> >> >
> >> >> > (if you don't add '3d' in this case, it will default to a contour
> >> >> > plot, which btw is maybe not the best default)
> >> >> >
> >> >> > Aaron Meurer
> >> >> >
> >> >> > On Fri, Nov 11, 2011 at 7:46 AM, [email protected]
> >> >> > <[email protected]> wrote:
> >> >> >> The proposal that I made in
> https://github.com/sympy/sympy/pull/673
> >> >> >> may
> >> >> >> or
> >> >> >> may not became part of sympy but I like it and it's already quite
> >> >> >> useful for
> >> >> >> me.
> >> >> >>
> >> >> >> Here are some examples. I would like to know what do you think.
> The
> >> >> >> 3d
> >> >> >> stuff
> >> >> >> runs only on the latest version of matplotlib _after_ fixing a bug
> >> >> >> (mentioned in the commit history, but those will be squashed
> soon).
> >> >> >
> >> >> > I didn't have any problems with it, though you do seem to have
> found
> >> >> > a
> >> >> > bug in matplotlib. I would submit a pull request to them fixing
> it.
> >> >> >
> >> >> > Aaron Meurer
> >> >> >
> >> >> >>
> >> >> >> The script to produce them is also attached (as the api is
> probably
> >> >> >> more
> >> >> >> important than the visuals (the _series[index] stuff is just a
> >> >> >> workaround
> >> >> >> until getters are written)).
> >> >> >>
> >> >> >> Regards
> >> >> >> Stefan
> >> >> >>
> >> >> >> --
> >> >> >> 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.
> >> >> >>
> >> >> >
> >> >>
> >> >> --
> >> >> 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.
> >> >>
> >> >
> >> > --
> >> > 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.
> >> >
> >>
> >> --
> >> 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.
> >>
> >
> > --
> > 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.
> >
>
> --
> 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.
>
>
--
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.
