Thank you! This works!
/Johan


On Tue, Sep 10, 2013 at 1:08 AM, Rasmus <rasmus@gmx.us> wrote:
Johan Ekh <ekh.johan@gmail.com> writes:

> Hi all,
> I would like to create a plot with matplotlib and have it exported to a
> beamer presentation without storing the plot in a file. Is that possible?

I guess you'd want to plot is as a pgf file, whether real of
'virtual'.  You could send the result to STDOUT but it may take a bit
more effort.  Also, a simple test with sys.stdout says the pgf backend
doesn't support stdout. . .

If *printing* to a pgf file everything works out of the box in recent
versions of Org.

> Can someone point me to an example or a good starting point?

  http://matplotlib.org/users/pgf.html

Here's an example of a simple plot.

#+TITLE: =matplotlib= and =pgf=
#+LATEX_HEADER: \usepackage{pgf}
#+NAME:spectrum
#+BEGIN_SRC python :var OUT="test.pgf" :exports results :results value file
  import matplotlib as mpl
  pgf_with_pdflatex = {
      "pgf.texsystem": "pdflatex",
      "text.usetex": True,
      'pgf.rcfonts': False,
      'font.size': 9,
      'fond.family': 'serif',
      "pgf.preamble": [
          r"\usepackage[utf8]{inputenc}",
          r"\usepackage[T1]{fontenc}"]}
  mpl.rcParams.update(pgf_with_pdflatex)
  import matplotlib.pyplot as plt
  from numpy import pi, cos, linspace
  s1, t1, t2 = 1, .8, .2
  s = lambda w: s1 / (2 * pi) * (1 + t1 ** 1 + t2 ** 2 + (1 + t2) * 2 * t1 * cos(w) + 2 * t2 * cos(4 * w))
  x = linspace(0, pi, 1000)
  plt.figure(figsize=(4,1.5))
  plt.plot(x, s(x))
  plt.xlim( 0, pi)
  plt.xlabel("$\\omega$")
  plt.ylabel("Spectrum")
  plt.tight_layout(0)
  plt.savefig(OUT, format = 'pgf')
  return(OUT)
#+END_SRC

#+RESULTS: spectrum
[[file:test.pgf]]




--
. . . The proofs are technical in nature and provides no real understanding.