📄 a8743.htm
字号:
<HTML><HEAD><TITLE>PyQwt: Python Bindings for Qwt</TITLE><METANAME="GENERATOR"CONTENT="Modular DocBook HTML Stylesheet Version 1.72"><LINKREL="HOME"TITLE="GUI Programming with Python: QT Edition"HREF="book1.htm"><LINKREL="UP"TITLE="Appendices"HREF="p8645.htm"><LINKREL="PREVIOUS"TITLE="Reading the Qt Documentation"HREF="a8647.htm"><LINKREL="NEXT"TITLE="PyQwt"HREF="x8818.htm"></HEAD><BODYCLASS="APPENDIX"BGCOLOR="#FFFFFF"TEXT="#000000"LINK="#0000FF"VLINK="#840084"ALINK="#0000FF"><DIVCLASS="NAVHEADER"><TABLESUMMARY="Header navigation table"WIDTH="100%"BORDER="0"CELLPADDING="0"CELLSPACING="0"><TR><THCOLSPAN="3"ALIGN="center">GUI Programming with Python: QT Edition</TH></TR><TR><TDWIDTH="10%"ALIGN="left"VALIGN="bottom"><AHREF="a8647.htm"ACCESSKEY="P">Prev</A></TD><TDWIDTH="80%"ALIGN="center"VALIGN="bottom"></TD><TDWIDTH="10%"ALIGN="right"VALIGN="bottom"><AHREF="x8818.htm"ACCESSKEY="N">Next</A></TD></TR></TABLE><HRALIGN="LEFT"WIDTH="100%"></DIV><DIVCLASS="APPENDIX"><H1><ANAME="PYQWT">Appendix B. PyQwt: Python Bindings for Qwt</A></H1><DIVCLASS="TOC"><DL><DT><B>Table of Contents</B></DT><DT><AHREF="a8743.htm#SECTNUMPY">NumPy</A></DT><DT><AHREF="x8818.htm">PyQwt</A></DT></DL></DIV><FONTCOLOR="RED"> Gerard Vermeulen </FONT><P>Using sip, it is possible to wrap any C++ library. Jim Bublitz, for instance, has wrapped the core libraries of KDE 2, and Gerard Vermeulen has wrapped the Qwt toolkit. This appendix has been written by Gerard Vermeulen to introduce this extension library.</P><P>PyQwt is a set of Python bindings for the Qt Widgets for Technics toolkit, which is freely downloadable at http://qwt.sourceforge.net. PyQwt is equally free, and available from http://gerard.vermeulen.free.fr.</P><P>Behind the innocuous trigram 'Qwt' a complex set of widgets is hiding. This extension library, written by Josef Wilgen with the aid of many others, fills in a noticeable gap in the Qt library: data visualisation. This toolkit features fast plotting of <TTCLASS="LITERAL">Numerical Python arrays</TT> (and Python lists or tuples) of Python floats.</P><P>Remember how we created a rolling chart in <AHREF="c7391.htm">Chapter 21</A> when we investigated <TTCLASS="CLASSNAME">QPainter</TT>s? It was quite an interesting job, but for serious applications you'd need a stronger package.</P><P>Fortunately, Python possesses a very strong array manipulation package: the Numerical Python Extensions (or, affectionately, <SPANCLASS="APPLICATION">numpy</SPAN>, available at http://www.pfdubois.com/numpy), which, when paired with the Qwt extensions, gives you the power to create complex graphing and charting applications.</P><DIVCLASS="SECT1"><H1CLASS="SECT1"><ANAME="SECTNUMPY">NumPy</A></H1><P>The Numerical Python Extensions, also called NumPy or Numeric, turn Python into an ideal tool for experimental numerical and scientific computing (better than specialized programs like MatLab, Octave, RLab or SciLab). NumPy is useful for everybody who analyzes data with the help of a spreadsheet program like Microsoft Excel—it is not just for mathematicians and scientists who crunch lots of data.</P><P>NumPy defines a new data type, <TTCLASS="LITERAL">NumPy array</TT>, and a very complete set of operators and functions to manipulate <TTCLASS="LITERAL">NumPy arrays</TT>. All the functionality of NumPy can be obtained in pure Python, but NumPy gives you speed and elegance.</P><P>In the following, I assume that you have installed NumPy on your system. Doing so is not really difficult. There are binary packages for Windows, or source packages for all platforms. A source package is installed using distutils (see <AHREF="c8349.htm">Chapter 26</A>), by typing</P><PRECLASS="SCREEN">root@calcifer:/home/boud# python setup_all.py install </PRE><P>Once numpy is installed, you can start Python (or open the Interpreter window in BlackAdder) and import the NumPy extension:</P><PRECLASS="PROGRAMLISTING">[packer@slow packer]$ pythonPython 2.1.1 (#1, Aug 20 2001, 08:17:33) [GCC 2.95.3 19991030 (prerelease)] on linux2Type "copyright", "credits" or "license" for more information.>>> from Numeric import *>>> </PRE><P>A <TTCLASS="LITERAL">NumPy array</TT> looks like a list and can be created from a list (in fact, from any sequency type: list, tuple or string).</P><P>Let's create and print a 1-dimensional <TTCLASS="LITERAL">NumPy array</TT> of Python floats:</P><PRECLASS="PROGRAMLISTING">>>> a = array([1.0, 4.0, 9.0, 16.0, 25.0, 36.0, 49.0, 64.0, 81.0, 100.0])>>> print a[ 1. 4. 9. 16. 25. 36. 49. 64. 81. 100.]>>> </PRE><P>This creates a 1-dimensional <TTCLASS="LITERAL">NumPy array</TT>. All elements in the list should have the same data type.</P><P>A 2-dimensional <TTCLASS="LITERAL">NumPy array</TT> is created from a list of sub-lists: </P><PRECLASS="PROGRAMLISTING">>>> b = array([[0.0, 1.0], [2.0, 3.0]])>>> print b[[ 0. 1.] [ 2. 3.]]>>> </PRE><P>The sub-lists should have the same length, and all the elements in all the sub-lists should have the same data type.</P><P>You can show off with <TTCLASS="LITERAL">NumPy array</TT>s of even higher dimensions (up to 40, by default). For example, a 3-dimensional <TTCLASS="LITERAL">NumPy array</TT> is created from a list of sub-lists of sub-sub-lists:</P><PRECLASS="PROGRAMLISTING">>>> c = array([[[0.0, 1.0], [2.0, 3.0]], [[4.0, 5.0], [6.0, 7.0]]])>>> print c[[[ 0. 1.] [ 2. 3.]] [[ 4. 5.] [ 6. 7.]]]>>> </PRE><P>The sub-lists should have the same length, the sub-sub-lists should have the same length, and all elements of all sub-sub-lists should have the same data type.</P><P>In the following, I am going to compare the functionality of <TTCLASS="LITERAL">NumPy arrays</TT> and lists. Here is an easier method to create a <TTCLASS="LITERAL">NumPy array</TT>:</P><PRECLASS="PROGRAMLISTING">>>> ax = arange(0.0, 5.0, 0.5)>>> print ax[ 0. 0.5 1. 1.5 2. 2.5 3. 3.5 4. 4.5]>>> </PRE><P>The function call <TTCLASS="FUNCTION">arange(0.0, 5.0, 0.5)</TT> returns an array with elements ranging from 0.0 to 5.0 (non-inclusive) in steps of 0.5. Here is a similiar function to return a list with the same properties: </P><PRECLASS="PROGRAMLISTING">def lrange(start, stop, step): start, stop, step = float(start), float(stop), float(step) size = int(round((stop-start)/step)) result = [start] * size for i in xrange(size): result[i] += i * step return result </PRE><P>After copying and pasting the function definition in your Python interpreter, do:</P><PRECLASS="PROGRAMLISTING">>>> lx = lrange(0.0, 5.0, 0.5)>>> print lx[0.0, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5] </PRE><P>Why are <TTCLASS="LITERAL">NumPy arrays</TT> better than lists? The full answer is speed and elegance. To compare lists and <TTCLASS="LITERAL">NumPy arrays</TT> with respect to elegance, lets use a simple function:</P><PRECLASS="PROGRAMLISTING">def lorentzian(x): return 1.0/(1.0+(x-2.5)**2) </PRE><P>To calculate a list, ly, containing the function values for each element of ly, we can do:</P><PRECLASS="PROGRAMLISTING">>>> ly = [0.0]*len(lx)>>> for i in range(len(lx)): ly[i] = lorentzian(lx[i])... </PRE><P>Do you know that you can get rid of the loop? The following is more elegant and slightly faster:</P><PRECLASS="PROGRAMLISTING">>>> ly = map(lorentzian, lx) </PRE><P><TTCLASS="LITERAL">NumPy arrays</TT> are even more elegant, and they allow:</P><PRECLASS="PROGRAMLISTING">>>> ay = lorentzian(ax) </PRE><P>Almost magic, isn't it? I wrote the function lorentzian(x) assuming that x is Python float. If you call lorentzian with a <TTCLASS="LITERAL">NumPy array</TT> as argument, it returns a <TTCLASS="LITERAL">NumPy array</TT>. This does not work with lists:</P><PRECLASS="PROGRAMLISTING">>>> ly = lorentzian(lx)Traceback (most recent call last): File "<stdin>", line 1, in ? File "<stdin>", line 2, in lorentzianTypeError: unsupported operand type(s) for ->>> </PRE><P>To compare speed, we create a list, xl, and a <TTCLASS="LITERAL">NumPy array</TT>, xa, with 100000 elements and use the profile module to time the statements <TTCLASS="LITERAL">yl = map(lorentzian, xl)</TT> and <TTCLASS="LITERAL">ya = lorentzian(xa)</TT>:</P><PRECLASS="PROGRAMLISTING">>>> import profile>>> xl = lrange(0, 10, 0.0001)>>> xa = arange(0, 10, 0.0001)>>> profile.run('yl = map(lorentzian, xl)') 100002 function calls in 2.200 CPU seconds Ordered by: standard name ncalls tottime percall cumtime percall filename:lineno(function) 100000 1.000 0.000 1.000 0.000 <stdin>:1(lorentzian) 1 1.200 1.200 2.200 2.200 <string>:1(?) 0 0.000 0.000 profile:0(profiler) 1 0.000 0.000 2.200 2.200 profile:0(yl = map(lorentzian, xl))>>> profile.run('ya = lorentzian(xa)') 3 function calls in 0.090 CPU seconds Ordered by: standard name ncalls tottime percall cumtime percall filename:lineno(function) 1 0.090 0.090 0.090 0.090 <stdin>:1(lorentzian) 1 0.000 0.000 0.090 0.090 <string>:1(?) 0 0.000 0.000 profile:0(profiler) 1 0.000 0.000 0.090 0.090 profile:0(ya = lorentzian(xa))>>> </PRE><P>On my computer, the Numerical Python extensions are almost 25 times faster than pure Python!</P><P>There exists a scientific plotting program, <SPANCLASS="APPLICATION">SciGraphica</SPAN> (http://scigraphica.sourceforge.net), which allows you to manipulate your data in a spreadsheet. The underlying engine is a python interpreter with the NumPy. Each column in the spreadsheet is in reality a <TTCLASS="LITERAL">NumPy array</TT>. This clearly demostrates the power of this extension. If you want to know more about NumPy, you can consult the excellent documentation at http://www.pfdubois.com/numpy, the homepage of NumPy.</P></DIV></DIV><DIVCLASS="NAVFOOTER"><HRALIGN="LEFT"WIDTH="100%"><TABLESUMMARY="Footer navigation table"WIDTH="100%"BORDER="0"CELLPADDING="0"CELLSPACING="0"><TR><TDWIDTH="33%"ALIGN="left"VALIGN="top"><AHREF="a8647.htm"ACCESSKEY="P">Prev</A></TD><TDWIDTH="34%"ALIGN="center"VALIGN="top"><AHREF="book1.htm"ACCESSKEY="H">Home</A></TD><TDWIDTH="33%"ALIGN="right"VALIGN="top"><AHREF="x8818.htm"ACCESSKEY="N">Next</A></TD></TR><TR><TDWIDTH="33%"ALIGN="left"VALIGN="top">Reading the Qt Documentation</TD><TDWIDTH="34%"ALIGN="center"VALIGN="top"><AHREF="p8645.htm"ACCESSKEY="U">Up</A></TD><TDWIDTH="33%"ALIGN="right"VALIGN="top">PyQwt</TD></TR></TABLE></DIV></BODY></HTML>⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -