{ "cells": [ { "cell_type": "markdown", "metadata": { "toc": true }, "source": [ "\n", "# Plotting\n", "\n", "single: MatPlotLib single: plots\n", "\n", "The graphical representation of data---plotting---is one of the most\n", "important tools for evaluating and understanding scientific data and\n", "theoretical predictions. However, plotting is not a part of core Python\n", "but is provided through one of several possible library modules. The\n", "most highly developed and widely used plotting package for Python is\n", "MatPlotLib (). It is a powerful and\n", "flexible program that has become the *de facto* standard for 2-d\n", "plotting with Python.\n", "\n", "Because MatPlotLib is an external library---in fact it's a collection of\n", "libraries---it must be imported into any routine that uses it.\n", "MatPlotLib makes extensive use of NumPy so the two should be imported\n", "together. Therefore, for any program for which you would like to produce\n", "2-d plots, you should include the lines\n", "\n", "``` python\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "```\n", "\n", "There are other MatPlotLib sub-libraries, but the `pyplot` library\n", "provides nearly everything that you need for 2-d plotting. The standard\n", "prefix for it is `plt`. The two statements above must appear before any\n", "calls to NumPy or MatPlotLib routines are made.\n", "\n", "One final word before we get started: We only scratch the surface of\n", "what is possible using MatPlotLib and as you become familiar with it,\n", "you will surely want to do more than this manual describes. In that\n", "case, you need to go the the web to get more information. A good place\n", "to start is . Another\n", "interesting web page is .\n", "\n", "An interactive session with `pyplot`\n", "------------------------------------\n", "\n", "single: plots; interactive\n", "\n", "We begin with an interactive plotting session that illustrates some very\n", "basic features of MatPlotLib. Type in the `plot` command shown below and\n", "press the return key. Take care to follow the exact syntax.\n", "\n", "``` ipython\n", "In [1]: plt.plot([1,2,3,2,3,4,3,4,5])\n", "Out[1]: []\n", "```\n", "\n", "
\n", "\"\"
Interactive plot window
\n", "
\n", "\n", "A window should appear with a plot that looks something like the\n", "`fig-zigzagPlotDemo` shown here. By default, the `plot` function draws a\n", "line between the data points that were entered. You can save this plot\n", "to an image file by clicking on the floppy disk icon at the top of the\n", "plot window. You can also zoom, pan, scroll through the plot, and return\n", "to the original view using the other icons in the plot window.\n", "Experimenting with them reveals their functions.\n", "\n", "When you are finished, be sure to close the plot window.\n", "\n", "Let's take a closer look at the `plot` function. It is used to plot\n", "$x$-$y$ data sets and is written like this\n", "\n", "``` python\n", "plt.plot(x, y)\n", "```\n", "\n", "where `x` and `y` are arrays (or lists) that have the same size. If the\n", "`x` array is missing, that is, if there is only a single array, as in\n", "our example above, the `plot` function uses `0, 1, ..., N-1` for the `x`\n", "array, where `N` is the size of the `y` array. Thus, the `plot` function\n", "provides a quick graphical way of examining a data set.\n", "\n", "More typically, you supply both an $x$ and a $y$ data set to plot.\n", "Taking things a bit further, you may also want to plot several data sets\n", "on the same graph, use symbols as well as lines, label the axes, create\n", "a title and a legend, and control the color of symbols and lines. All of\n", "this is possible but requires calling a number of plotting functions.\n", "For this reason, plotting is usually done using a Python script or\n", "program.\n", "\n", "Basic plotting\n", "--------------\n", "\n", "single: plots; basic\n", "\n", "The quickest way to learn how to plot using the MatPlotLib library is by\n", "example. For our first task, let's plot the sine function over the\n", "interval from 0 to $4\\pi$. The main plotting function `plot` in\n", "MatPlotLib does not plot functions *per se*, it plots $(x,y)$ data\n", "points. As we shall see, we can instruct the function `plot` either to\n", "just draw point---or dots---at each data point, or we can instruct it to\n", "draw straight lines between the data points. To create the illusion of\n", "the smooth function that the sine function is, we need to create enough\n", "$(x,y)$ data points so that when `plot` draws straight lines between the\n", "data points, the function appears to be smooth. The sine function\n", "undergoes two full oscillations with two maxima and two minima between 0\n", "and $4\\pi$. So let's start by creating an array with 33 data points\n", "between 0 and $4\\pi$, and then let MatPlotLib draw a straight line\n", "between them. Our code consists of four parts\n", "\n", "- import the NumPy and MatPlotLib modules (lines 1-2 below)\n", "- create the $(x,y)$ data arrays (lines 3-4 below)\n", "- have `plot` draw straight lines between the $(x,y)$ data points\n", " (line 5 below)\n", "- display the plot in a figure window using the `show` function (line\n", " 6 below)\n", "\n", "Here is our code, which consists of only 6 lines:\n", "\n", "``` python\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "x = np.linspace(0, 4.*np.pi, 33)\n", "y = np.sin(x)\n", "plt.plot(x, y)\n", "plt.show()\n", "```\n", "\n", "
\n", "\"\"
Sine function
\n", "
\n", "\n", "Only 6 lines suffice to create the plot, which consists of the sine\n", "function over the interval from 0 to $4\\pi$, as advertised, as well as\n", "axes annotated with nice whole numbers over the appropriate interval.\n", "It's a pretty nice plot made with very little code.\n", "\n", "One problem, however, is that while the plot oscillates like a sine\n", "wave, it is not smooth. This is because we did not create the $(x,y)$\n", "arrays with enough data points. To correct this, we need more data\n", "points. The plot below was created using the same program shown above\n", "but with 129 $(x,y)$ data points instead of 33. Try it out your self by\n", "copying the above program and replacing 33 in line 3 with 129 so that\n", "the function `linspace` creates an array with 129 data points instead of\n", "33.\n", "\n", "
\n", "\"\"
Sine function plotted using more data points
\n", "
\n", "\n", "The code above illustrates how plots can be made with very little code\n", "using the MatPlotLib module. In making this plot, MatPlotLib has made a\n", "number of choices, such as the size of the figure, the blue color of the\n", "line, even the fact that by default a line is drawn between successive\n", "data points in the $(x,y)$ arrays. All of these choices can be changed\n", "by explicitly instructing MatPlotLib to do so. This involves including\n", "more arguments in the function calls we have used and using new\n", "functions that control other properties of the plot. The next example\n", "illustrates a few of the simpler embellishments that are possible.\n", "\n", "In the `fig-WavyPulse` figure, we plot two $(x,y)$ data sets: a smooth\n", "line curve and some data represented by red circles. In this plot, we\n", "label the $x$ and $y$ axes, create a legend, and draw lines to indicate\n", "where $x$ and $y$ are zero. The code that creates this plot is shown\n", "below.\n", "\n", "``` python\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "\n", "# read data from file\n", "xdata, ydata = np.loadtxt('wavePulseData.txt', unpack=True)\n", "\n", "# create x and y arrays for theory\n", "x = np.linspace(-10., 10., 200)\n", "y = np.sin(x) * np.exp(-(x/5.0)**2)\n", "\n", "# create plot\n", "plt.figure(1, figsize = (6,4) )\n", "plt.plot(x, y, 'b-', label='theory')\n", "plt.plot(xdata, ydata, 'ro', label=\"data\")\n", "plt.xlabel('x')\n", "plt.ylabel('transverse displacement')\n", "plt.legend(loc='upper right')\n", "plt.axhline(color = 'gray', zorder=-1)\n", "plt.axvline(color = 'gray', zorder=-1)\n", "\n", "# save plot to file\n", "plt.savefig('WavyPulse.pdf')\n", "\n", "# display plot on screen\n", "plt.show()\n", "```\n", "\n", "
\n", "\"\"
Wavy pulse
\n", "
\n", "\n", "If you have read the first four chapters, the code in lines 1-9 in the\n", "above script should be familiar to you. Fist, the script loads the NumPy\n", "and MatPlotLib modules, then reads data from a data file into two\n", "arrays, `xdata` and `ydata`, and then creates two more arrays, `x` and\n", "`y`. The first pair or arrays, `xdata` and `ydata`, contain the $x$-$y$\n", "data that are plotted as red circles in the `fig-WavyPulse` figure; the\n", "arrays created in line 8 and 9 contain the $x$-$y$ data that are plotted\n", "as a blue line.\n", "\n", "The functions that do the plotting begin on line 12. Let's go through\n", "them one by one and see what they do. You will notice in several cases\n", "that *keyword arguments* (`kwargs`) are used in several cases. Keyword\n", "arguments are *optional* arguments that have the form `kwarg=` *data*,\n", "where *data* might be a number, a string, a tuple, or some other form of\n", "data.\n", "\n", "> `figure()` \n", "> creates a blank figure window. If it has no arguments, it creates a\n", "> window that is 8 inches wide and 6 inches high by default, although\n", "> the size that appears on your computer depends on your screen's\n", "> resolution. For most computers, it will be much smaller. You can\n", "> create a window whose size differs from the default using the optional\n", "> keyword argument `figsize`, as we have done here. If you use\n", "> `figsize`, set it equal to a 2-element tuple where the elements are\n", "> the width and height, respectively, of the plot. Multiple calls to\n", "> `figure()` opens multiple windows: `figure(1)` opens up one window for\n", "> plotting, `figure(2)` another, and `figure(3)` yet another.\n", ">\n", "> `plot(x, y,` *optional arguments* `)` \n", "> graphs the $x$-$y$ data in the arrays `x` and `y`. The third argument\n", "> is a format string that specifies the color and the type of line or\n", "> symbol that is used to plot the data. The string `'ro'` specifies a\n", "> red (`r`) circle (`o`). The string `'b-'` specifies a blue (`b`) solid\n", "> line (`-`). The keyword argument `label` is set equal to a string that\n", "> labels the data if the `legend` function is called subsequently.\n", ">\n", "> `xlabel(` *string* `)` \n", "> takes a string argument that specifies the label for the graph's\n", "> $x$-axis.\n", ">\n", "> `ylabel(` *string* `)` \n", "> takes a string argument that specifies the label for the graph's\n", "> $y$-axis.\n", ">\n", "> `legend()` \n", "> makes a legend for the data plotted. Each $x$-$y$ data set is labeled\n", "> using the string that was supplied by the `label` keyword in the\n", "> `plot` function that graphed the data set. The `loc` keyword argument\n", "> specifies the location of the legend.\n", ">\n", "> `axhline()` \n", "> draws a horizontal line across the width of the plot at `y=0`. The\n", "> optional keyword argument `color` is a string that specifies the color\n", "> of the line. The default color is black. The optional keyword argument\n", "> `zorder` is an integer that specifies which plotting elements are in\n", "> front of or behind others. By default, new plotting elements appear\n", "> *on top of* previously plotted elements and have a value of\n", "> `zorder=0`. By specifying `zorder=-1`, the horizontal line is plotted\n", "> *behind* all existing plot elements that have not be assigned an\n", "> explicit `zorder` less than -1.\n", ">\n", "> `axvline()` \n", "> draws a vertical line from the top to the bottom of the plot at `x=0`.\n", "> See `axhline()` for explanation of the arguments.\n", ">\n", "> `savefig(` *string* `)` \n", "> saves the figure to data data file with a name specified by the string\n", "> argument. The string argument can also contain path information if you\n", "> want to save the file so some place other than the default directory.\n", ">\n", "> `show()` \n", "> displays the plot on the computer screen. No screen output is produced\n", "> before this function is called.\n", "\n", "single: MatPlotLib functions; figure single: MatPlotLib functions; plot\n", "single: MatPlotLib functions; xlabel, ylabel single: MatPlotLib\n", "functions; legend single: MatPlotLib functions; ayhline, axhline single:\n", "MatPlotLib functions; savefig single: MatPlotLib functions; show\n", "\n", "To plot the solid blue line, the code uses the `'b-'` format specifier\n", "in the `plot` function call. It is important to understand that\n", "MatPlotLib draws *straight lines* between data points. Therefore, the\n", "curve will appear smooth only if the data in the NumPy arrays are\n", "sufficiently dense. If the space between data points is too large, the\n", "straight lines the `plot` function draws between data points will be\n", "visible. For plotting a typical function, something on the order of\n", "100-200 data points usually produces a smooth curve, depending on just\n", "how curvy the function is. On the other hand, only two points are\n", "required to draw a smooth straight line.\n", "\n", "Detailed information about the MatPlotLib plotting functions are\n", "available online, starting with the site\n", ". The main MatPlotLib\n", "site is .\n", "\n", "### Specifying line and symbol types and colors\n", "\n", "In the above example, we illustrated how to draw one line type (solid),\n", "one symbol type (circle), and two colors (blue and red). There are many\n", "more possibilities, which are specified in the tables below. The way it\n", "works is to specify a string consisting of one or more plotting format\n", "specifiers. There are two types of format specifiers, one for the line\n", "or symbol type and another for the color. It does not matter in which\n", "order the format specifiers are listed in the string. Examples are given\n", "following the two tables. Try them out to make sure you understand how\n", "these plotting format specifiers work.\n", "\n", "single: plots; line and symbol specifiers\n", "\n", "The first table below shows the characters used to specify the line or\n", "symbol type that is used. If a line type is chosen, the lines are drawn\n", "between the data points. If a marker type is chosen, the a marker is\n", "plotted at each data point.\n", "\n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", ">
characterdescriptioncharacterdescription
-solid line style3tri_left marker
--dashed line style4tri_right marker
-.dash-dot line stylessquare marker
:dotted line styleppentagon marker
.point marker*star marker
,pixel markerhhexagon1 marker
ocircle markerHhexagon2 marker
vtriangle_down marker+plus marker
^triangle_up markerxx marker
<triangle_left markerDdiamond marker
>triangle_right markerdthin_diamond marker
1tri_down marker|vline marker
2tri_up marker_hline marker
\n", ">\n", "This second table gives the character codes for eight different colors.\n", "Many more are possible but the color specification becomes more complex.\n", "You can consult the web-based MatPlotLib documentation for further\n", "details.\n", "\n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", "> \n", ">
charactercolor
bblue
ggreen
rred
ccyan
mmagenta
yyellow
kblack
wwhite
\n", ">\n", "Here are some examples of how these format specifiers can be used:\n", "\n", "``` python\n", "plot(x, y, 'ro') # plots red circles\n", "plot(x, y, 'ks-') # plot black squares connected by black lines\n", "plot(x, y, 'g^') # plots green triangles that point up\n", "\n", "plot(x, y, 'k-') # plots a black line between the points\n", "plot(x, y, 'ms') # plots magenta squares \n", "```\n", "\n", "You can also make two calls sequentially for added versatility. For\n", "example, by sequentially calling the last two plot calls, the plot\n", "produces magenta squares on top of black lines connecting the data\n", "points.\n", "\n", "These format specifiers give rudimentary control of the plotting symbols\n", "and lines. MatPlotLib provides much more precise and detailed control of\n", "the plotting symbol size, line types, and colors using optional keyword\n", "arguments instead of the plotting format strings introduced above. For\n", "example, the following command creates a plot of large yellow diamond\n", "symbols with blue edges connected by a green dashed line:\n", "\n", "``` python\n", "plt.plot(x, y, color='green', linestyle='dashed', marker='d', \n", " markerfacecolor='yellow', markersize=12, \n", " markeredgecolor='blue')\n", "```\n", "\n", "Try it out! The online MatPlotLib documentation provides all the\n", "plotting format keyword arguments and their possible values.\n", "\n", "### Error bars\n", "\n", "single: plots; error bars\n", "\n", "When plotting experimental data it is customary to include error bars\n", "that indicate graphically the degree of uncertainty that exists in the\n", "measurement of each data point. The MatPlotLib function `errorbar` plots\n", "data with error bars attached. It can be used in a way that either\n", "replaces or augments the `plot` function. Both vertical and horizontal\n", "error bars can be displayed. The figure below illustrates the use of\n", "error bars.\n", "\n", "
\n", "\"\"
Error Bars
\n", "
\n", "\n", "When error bars are desired, you typically replace the `plot` function\n", "with the `errorbar` function. The first two arguments of the `errorbar`\n", "function are the `x` and `y` arrays to be plotted, just as for the\n", "`plot` function. The keyword `fmt` *must be used* to specify the format\n", "of the points to be plotted; the format specifiers are the same as for\n", "`plot`. The keywords `xerr` and `yerr` are used to specify the $x$ and\n", "$y$ error bars. Setting one or both of them to a constant specifies one\n", "size for all the error bars. Alternatively, setting one or both of them\n", "equal to an array that has the same length as the `x` and `y` arrays\n", "allows you to give each data point an error bar with a different value.\n", "If you only want $y$ error bars, then you should only specify the `yerr`\n", "keyword and omit the `xerr` keyword. The color of the error bars is set\n", "with the keyword `ecolor`.\n", "\n", "The code and plot below illustrates how to make error bars and was used\n", "to make the above plot. Lines 14 and 15 contain the call to the\n", "`errorbar` function. The $x$ error bars are all set to a constant value\n", "of 0.75, meaning that the error bars extend 0.75 to the left and 0.75 to\n", "the right of each data point. The $y$ error bars are set equal to an\n", "array, which was read in from the data file containing the data to be\n", "plotted, so each data point has a different $y$ error bar. By the way,\n", "leaving out the `xerr` keyword argument in the `errorbar` function call\n", "below would mean that only the $y$ error bars would be plotted.\n", "\n", "``` python\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "\n", "# read data from file\n", "xdata, ydata, yerror = np.loadtxt('expDecayData.txt', unpack=True)\n", "\n", "# create theoretical fitting curve\n", "x = np.linspace(0, 45, 128)\n", "y = 1.1+ 3.0*x*np.exp(-(x/10.0)**2)\n", "\n", "# create plot\n", "plt.figure(1, figsize = (6,4) )\n", "plt.plot(x, y, 'b-', label=\"theory\")\n", "plt.errorbar(xdata, ydata, fmt='ro', label=\"data\", \n", " xerr=0.75, yerr=yerror, ecolor='black')\n", "plt.xlabel('x')\n", "plt.ylabel('transverse displacement')\n", "plt.legend(loc='upper right')\n", "\n", "# save plot to file\n", "plt.savefig('ExpDecay.pdf')\n", "\n", "# display plot on screen\n", "plt.show()\n", "```\n", "\n", "We have more to say about the `errorbar` function in the sections on\n", "logarithmic plots. But the brief introduction given here should suffice\n", "for making most plots not involving logarithmic axes.\n", "\n", "### Setting plotting limits and excluding data\n", "\n", "It turns out that you often want to restrict the range of numerical\n", "values over which you plot data or functions. In these cases you may\n", "need to manually specify the plotting window or, alternatively, you may\n", "wish to exclude data points that are outside some set of limits. Here we\n", "demonstrate methods for doing this.\n", "\n", "#### Setting plotting limits\n", "\n", "single: plots; setting axis limits\n", "\n", "Suppose you want to plot the tangent function over the interval from 0\n", "to 10. The following script offers an straightforward first attempt.\n", "\n", "``` python\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "\n", "theta = np.arange(0.01, 10., 0.04)\n", "ytan = np.tan(theta)\n", "\n", "plt.figure()\n", "plt.plot(theta, ytan)\n", "plt.show()\n", "```\n", "\n", "
\n", "\"\"\n", "
\n", "\n", "The resulting plot, shown above, doesn't quite look like what you might\n", "have expected for $\\tan\\theta$ *vs* $\\theta$. The problem is that\n", "$\\tan\\theta$ diverges at $\\theta = \\pi/2, 3\\pi/2, 5\\pi/2, ...$, which\n", "leads to large spikes in the plots as values in the `theta` array come\n", "near those values. Of course, we don't want the plot to extend all the\n", "way out to $\\pm\\infty$ in the $y$ direction, nor can it. Instead, we\n", "would like the plot to extend far enough that we get the idea of what is\n", "going on as $y\\rightarrow\\pm\\infty$, but we would still like to see the\n", "behavior of the graph near $y=0$. We can restrict the range of `ytan`\n", "values that are plotted using the MatPlotLib function `ylim`, as we\n", "demonstrate in the script below.\n", "\n", "``` python\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "\n", "theta = np.arange(0.01, 10., 0.04)\n", "ytan = np.tan(theta)\n", "\n", "plt.figure()\n", "plt.plot(theta, ytan)\n", "plt.ylim(-8, 8) # restricts range of y axis from -8 to +8\n", "plt.axhline(color=\"gray\", zorder=-1)\n", "\n", "plt.show()\n", "```\n", "\n", "The figure produced by this script is shown below. The plot now looks\n", "much more like the familiar $\\tan\\theta$ function we know. We have also\n", "include a call to the `axline` function to create an $x$ axis.\n", "\n", "
\n", "\"\"
Tangent function (with spurious lines)
\n", "
\n", "\n", "The vertical blue lines at $\\theta = \\pi/2, 3\\pi/2, 5\\pi/2$ should not\n", "appear in a plot of $\\tan\\theta$ *vs* $\\theta$. However, they do appear\n", "because the `plot` function simply draws lines between the data points\n", "in the `x`-`y` arrays provided in its arguments. Thus, `plot` draws a\n", "line between the very large positive and negative `ytan` values\n", "corresponding to the `theta` values on either side of $\\pi/2$ where\n", "$\\tan\\theta$ diverges to $\\pm\\infty$. It would be nice to exclude that\n", "line.\n", "\n", "#### Masked arrays\n", "\n", "single: plots; masked arrays single: masked arrays\n", "\n", "We can exclude the data points near $\\theta = \\pi/2, 3\\pi/2, 5\\pi/2$ in\n", "the above plot, and thus avoid drawing the nearly vertical lines at\n", "those points, using NumPy's *masked array* feature. The code below shows\n", "how this is done and produces the graph below. The masked array feature\n", "is implemented in line 6 with a call to NumPy's `masked_where` function\n", "in the sub-module `ma` (masked array). Therefore, it is called by\n", "writing `np.ma.masked_where`. The `masked_where` function works as\n", "follows. The first argument sets the condition for masking elements of\n", "the array, which is specified by the second argument. In this case, the\n", "function says to mask all elements of the array `ytan` (the second\n", "argument) where the absolute value of `ytan` is greater than 20. The\n", "result is set equal to `ytanM`. When `ytanM` is plotted, MatPlotLib's\n", "`plot` function omits all masked points from the plot. You can think of\n", "it as the `plot` function lifting the pen that is drawing the line in\n", "the plot when it comes to the masked points in the array `ytanM`.\n", "\n", "
\n", "\"\"
Tangent function
\n", "
\n", "\n", "``` python\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "\n", "theta = np.arange(0.01, 10., 0.04)\n", "ytan = np.tan(theta)\n", "ytanM = np.ma.masked_where(np.abs(ytan)>20., ytan)\n", "\n", "plt.figure()\n", "plt.plot(theta, ytanM)\n", "plt.ylim(-8, 8)\n", "plt.axhline(color=\"gray\", zorder=-1)\n", "\n", "plt.show()\n", "```\n", "\n", "### Subplots\n", "\n", "single: plots; subplots\n", "\n", "Often you want to create two or more graphs and place them next to one\n", "another, generally because they are related to each other in some way.\n", "The plot below shows an example of such a plot. In the top graph,\n", "$\\tan\\theta$ and $\\sqrt{(8/\\theta)^2-1}$ *vs* $\\theta$ are plotted. The\n", "two curves cross each other at the points where\n", "$\\tan\\theta=\\sqrt{(8/\\theta)^2-1}$. In the bottom $\\cot\\theta$ and\n", "$-\\sqrt{(8/\\theta)^2-1}$ *vs* $\\theta$ are plotted. These two curves\n", "cross each other at the points where\n", "$\\cot\\theta=-\\sqrt{(8/\\theta)^2-1}$.\n", "\n", "
\n", "\"\"
Crossing functions
\n", "
\n", "\n", "The code that produces this plot is provided below.\n", "\n", "``` python\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "\n", "theta = np.arange(0.01, 8., 0.04)\n", "y = np.sqrt((8./theta)**2-1.)\n", "ytan = np.tan(theta)\n", "ytan = np.ma.masked_where(np.abs(ytan)>20., ytan)\n", "ycot = 1./np.tan(theta)\n", "ycot = np.ma.masked_where(np.abs(ycot)>20., ycot)\n", "\n", "plt.figure(1)\n", "\n", "plt.subplot(2, 1, 1)\n", "plt.plot(theta, y)\n", "plt.plot(theta, ytan)\n", "plt.ylim(-8, 8)\n", "plt.axhline(color=\"gray\", zorder=-1)\n", "plt.axvline(x=np.pi/2., color=\"gray\", linestyle='--', zorder=-1)\n", "plt.axvline(x=3.*np.pi/2., color=\"gray\", linestyle='--', zorder=-1)\n", "plt.axvline(x=5.*np.pi/2., color=\"gray\", linestyle='--', zorder=-1)\n", "plt.xlabel(\"theta\")\n", "plt.ylabel(\"tan(theta)\")\n", "\n", "plt.subplot(2, 1, 2)\n", "plt.plot(theta, -y)\n", "plt.plot(theta, ycot)\n", "plt.ylim(-8, 8)\n", "plt.axhline(color=\"gray\", zorder=-1)\n", "plt.axvline(x=np.pi, color=\"gray\", linestyle='--', zorder=-1)\n", "plt.axvline(x=2.*np.pi, color=\"gray\", linestyle='--', zorder=-1)\n", "plt.xlabel(\"theta\")\n", "plt.ylabel(\"cot(theta)\")\n", "\n", "plt.show()\n", "```\n", "\n", "The function `subplot`, called on lines 13 and 24, creates the two\n", "subplots in the above figure. `subplot` has three arguments. The first\n", "specifies the number of rows that the figure space is to be divided\n", "into; on line 13, it's two. The second specifies the number of columns\n", "that the figure space is to be divided into; on line 13, it's one. The\n", "third argument specifies which rectangle the will contain the plot\n", "specified by the following function calls. Line 13 specifies that the\n", "plotting commands that follow will be act on the first box. Line 24\n", "specifies that the plotting commands that follow will be act on the\n", "second box.\n", "\n", "We have also labeled the axes and included dashed vertical lines at the\n", "values of $\\theta$ where $\\tan\\theta$ and $\\cot\\theta$ diverge.\n", "\n", "Logarithmic plots\n", "-----------------\n", "\n", "single: plots; logarithmic axes\n", "\n", "Data sets can span many orders of magnitude from fractional quantities\n", "much smaller than unity to values much larger than unity. In such cases\n", "it is often useful to plot the data on logarithmic axes.\n", "\n", "### Semi-log plots\n", "\n", "single: plots; semi-log\n", "\n", "For data sets that vary exponentially in the independent variable, it is\n", "often useful to use one or more logarithmic axes. Radioactive decay of\n", "unstable nuclei, for example, exhibits an exponential decrease in the\n", "number of particles emitted from the nuclei as a function of time. In\n", "the plot below, for example, we show the decay of the radioactive\n", "isotope Phosphorus-32 over a period of 6 months, where the radioactivity\n", "is measured once each week. Starting at a decay rate of nearly $10^4$\n", "electrons (counts) per second, the decay rate diminishes to only about 1\n", "count per second after about 6 months or 180 days. If we plot counts per\n", "second as a function of time on a normal plot, as we have done in the\n", "plot on the left below, then the count rate is indistinguishable from\n", "zero after about 100 days. On the other hand, if we use a logarithmic\n", "axis for the count rate, as we have done in the plot on the right below,\n", "then we can follow the count rate well past 100 days and can readily\n", "distinguish it from zero. Moreover, if the data vary exponentially in\n", "time, then the data will fall along a straight line, as they do for the\n", "case of radioactive decay.\n", "\n", "
\n", "\"\"
Semi-log plotting
\n", "
\n", "\n", "MatPlotLib provides two functions for making semi-logarithmic plots,\n", "`semilogx` and `semilogy`, for creating plots with logarithmic $x$ and\n", "$y$ axes, with linear $y$ and $x$ axes, respectively. We illustrate\n", "their use in the program below, which made the above plots.\n", "\n", "``` python\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "\n", "# read data from file\n", "time, counts, unc = np.loadtxt('SemilogDemo.txt', unpack=True)\n", "\n", "# create theoretical fitting curve\n", "tau = 20.2 # Phosphorus-32 half life = 14 days; tau = t_half/ln(2)\n", "N0 = 8200. # Initial count rate (per second)\n", "t = np.linspace(0, 180, 128)\n", "N = N0 * np.exp(-t/tau)\n", "\n", "# create plot\n", "plt.figure(1, figsize = (10,4) )\n", "\n", "plt.subplot(1, 2, 1)\n", "plt.plot(t, N, 'b-', label=\"theory\")\n", "plt.plot(time, counts, 'ro', label=\"data\")\n", "plt.xlabel('time (days)')\n", "plt.ylabel('counts per second')\n", "plt.legend(loc='upper right')\n", "\n", "plt.subplot(1, 2, 2)\n", "plt.semilogy(t, N, 'b-', label=\"theory\")\n", "plt.semilogy(time, counts, 'ro', label=\"data\")\n", "plt.xlabel('time (days)')\n", "plt.ylabel('counts per second')\n", "plt.legend(loc='upper right')\n", "\n", "plt.tight_layout()\n", "\n", "# display plot on screen\n", "plt.show()\n", "```\n", "\n", "The `semilogx` and `semilogy` functions work the same way as the `plot`\n", "function. You just use one or the other depending on which axis you want\n", "to be logarithmic.\n", "\n", "#### The `tight_layout()` function\n", "\n", "single: MatPlotLib functions; tight\\_layout\n", "\n", "You may have noticed the `tight_layout()` function, called without\n", "arguments on line 30 of the program. This is a convenience function that\n", "adjusts the sizes of the plots to make room for the axes labels. If it\n", "is not called, the $y$-axis label of the right plot runs into the left\n", "plot. The `tight_layout()` function can also be useful in graphics\n", "windows with only one plot sometimes.\n", "\n", "### Log-log plots\n", "\n", "single: plots; log-log\n", "\n", "MatPlotLib can also make log-log or double-logarithmic plots using the\n", "function `loglog`. It is useful when both the $x$ and $y$ data span many\n", "orders of magnitude. Data that are described by a power law $y=Ax^b$,\n", "where $A$ and $b$ are constants, appear as straight lines when plotted\n", "on a log-log plot. Again, the `loglog` function works just like the\n", "`plot` function but with logarithmic axes.\n", "\n", "More advanced graphical output\n", "------------------------------\n", "\n", "The plotting methods introduced in the previous sections are perfectly\n", "adequate for basic plotting and are therefore recommended for simple\n", "graphical output. Here, we introduce an alternative syntax that\n", "harnesses the full power of MatPlotLib. It gives the user more options\n", "and greater control. Perhaps the most efficient way to learn this\n", "alternative syntax is to look at an example. The figure below\n", "illustrating `MultPlotDemo` is produced by the following code:\n", "\n", "
\n", "\"\"
Mulitple plots in the same window
\n", "
\n", "\n", " # Demonstrates the following:\n", " # plotting logarithmic axes\n", " # user-defined functions\n", " # \"where\" function, NumPy array conditional\n", "\n", " import numpy as np\n", " import matplotlib.pyplot as plt\n", "\n", " # Define the sinc function, with output for x=0 defined\n", " # as a special case to avoid division by zero. The code\n", " # below defining the sinc function is developed and\n", " # explained in Chapter 7, Section 1.\n", " def s(x):\n", " a = np.where(x==0., 1., np.sin(x)/x)\n", " return a\n", "\n", " # create arrays for plotting\n", " x = np.arange(0., 10., 0.1)\n", " y = np.exp(x)\n", "\n", " t = np.linspace(-10., 10., 100)\n", " z = s(t)\n", "\n", " # create a figure window\n", " fig = plt.figure(1, figsize=(9,8))\n", "\n", " # subplot: linear plot of exponential\n", " ax1 = fig.add_subplot(2,2,1)\n", " ax1.plot(x, y)\n", " ax1.set_xlabel('time (ms)')\n", " ax1.set_ylabel('distance (mm)')\n", " ax1.set_title('exponential')\n", "\n", " # subplot: semi-log plot of exponential\n", " ax2 = fig.add_subplot(2,2,2)\n", " ax2.plot(x, y)\n", " ax2.set_yscale('log')\n", " ax2.set_xlabel('time (ms)')\n", " ax2.set_ylabel('distance (mm)')\n", " ax2.set_title('exponential')\n", "\n", " # subplot: wide subplot of sinc function\n", " ax3 = fig.add_subplot(2,1,2)\n", " ax3.plot(t, z, 'r')\n", " ax3.axhline(color='gray')\n", " ax3.axvline(color='gray')\n", " ax3.set_xlabel('angle (deg)')\n", " ax3.set_ylabel('electric field')\n", " ax3.set_title('sinc function')\n", "\n", " # Adjusts white space to avoid collisions between subplots\n", " fig.tight_layout()\n", " plt.show()\n", "\n", "After defining several arrays for plotting, the above program opens a\n", "figure window in line 23 with the statement :\n", "\n", " fig = plt.figure(figsize=(9,8))\n", "\n", "The MatPlotLib statement above creates a **Figure** object, assigns it\n", "the name `fig`, and opens a blank figure window. Thus, just as we give\n", "lists, arrays, and numbers variable names (*e.g.* `a = [1, 2, 5, 7]`,\n", "`dd = np.array([2.3, 5.1, 3.9])`, or `st = 4.3`), we can give a figure\n", "object and the window in creates a name: here it is `fig`. In fact we\n", "can use the `figure` function to open up multiple figure objects with\n", "different figure windows. The statements :\n", "\n", " fig1 = plt.figure()\n", " fig2 = plt.figure()\n", "\n", "open up two separate windows, one named `fig1` and the other `fig2`. We\n", "can then use the names `fig1` and `fig2` to plot things in either\n", "window. The `figure` function need not take any arguments if you are\n", "satisfied with the default settings such as the figure size and the\n", "background color. On the other hane, by supplying one or more keyword\n", "arguments, you can customize the figure size, the background color, and\n", "a few other properties. For example, in the program listing (line 23),\n", "the keyword argument `figsize` sets the width and height of the figure\n", "window; the default size is `(8, 6)`; in our program we set it to\n", "`(9, 8)`, which is a bit wider and higher than the default size. In the\n", "example above, we also choose to open only a single window, hence the\n", "single `figure` call.\n", "\n", "The `fig.add_subplot(2,2,1)` in line 30 is a MatPlotLib function that\n", "divides the figure window into 2 rows (the first argument) and 2 columns\n", "(the second argument). The third argument creates a subplot in the first\n", "of the 4 subregions (*i.e.* of the 2 rows $\\times$ 2 columns) created by\n", "the `fig.add_subplot(2,2,1)` call. To see how this works, type the\n", "following code into a Python module and run it:\n", "\n", " import numpy as np\n", " import matplotlib.pyplot as plt\n", "\n", " fig = plt.figure(figsize=(9,8))\n", " ax1 = fig.add_subplot(2,2,1)\n", "\n", " plt.show()\n", "\n", "You should get a figure window with axes drawn in the upper left\n", "quadrant. The `fig.` prefix used with the `add_subplot(2,2,1)` function\n", "directs Python to draw these axes in the figure window named `fig`. If\n", "we had opened two figure windows, changing the prefix to correspond to\n", "the name of one or the other of the figure windows would direct the axes\n", "to be drawn in the appropriate window. Writing\n", "`ax1 = fig.add_subplot(2,2,1)` assigns the name ax1 to the axes in the\n", "upper left quadrant of the figure window.\n", "\n", "The `ax1.plot(x, y)` in line 27 directs Python to plot the\n", "previously-defined `x` and `y` arrays onto the axes named `ax1`. The\n", "`ax2 = fig.add_subplot(2,2,2)` draws axes in the second, or upper right,\n", "quadrant of the figure window. The `ax3 = fig.add_subplot(2,1,2)`\n", "divides the figure window into 2 rows (first argument) and 1 column\n", "(second argument), creates axes in the second or these two sections, and\n", "assigns those axes (*i.e.* that subplot) the name `ax3`. That is, it\n", "divides the figure window into 2 halves, top and bottom, and then draws\n", "axes in the half number 2 (the third argument), or lower half of the\n", "figure window.\n", "\n", "You may have noticed in above code that some of the function calls are a\n", "bit different from those used before: `xlabel(’time (ms)’)` becomes\n", "`set_xlabel(’time (ms)’)`, `title(’exponential’)` becomes\n", "`set_title(’exponential’)`, *etc.*\n", "\n", "The call `ax2.set_yscale('log')` sets the $y$-axes in the second plot to\n", "be logarithmic, thus creating a semi-log plot. Creating properly-labeled\n", "logarthmic axes like this is more straightforward with the advanced\n", "syntax illustrated in the above example.\n", "\n", "Using the prefixes `ax1`, `ax2`, or `ax3`, direct graphical instructions\n", "to their respective subplots. By creating and specifying names for the\n", "different figure windows and subplots within them, you access the\n", "different plot windows more efficiently. For example, the following code\n", "makes four identical subplots in a single figure window using a `for`\n", "loop.\n", "\n", "``` ipython\n", "In [1]: fig = figure()\n", "\n", "In [2]: ax1 = fig.add_subplot(221)\n", "\n", "In [3]: ax2 = fig.add_subplot(222)\n", "\n", "In [4]: ax3 = fig.add_subplot(223)\n", "\n", "In [5]: ax4 = fig.add_subplot(224)\n", "\n", "In [6]: for ax in [ax1, ax2, ax3, ax4]:\n", " ...: ax.plot([3,5,8],[6,3,1])\n", "\n", "In [7]: show()\n", "```\n", "\n", "Exercises\n", "---------\n", "\n", "1. Plot the function $y=3x^2$ for $-1 \\le x \\le 3$ as a continuous\n", " line. Include enough points so that the curve you plot appears\n", " smooth. Label the axes $x$ and $y$.\n", "\n", "2. Plot the following function for $-15 \\le x \\le 15$:\n", "\n", " $$y = \\frac{\\cos x}{1+\\frac{1}{5}x^2}$$\n", "\n", " Include enough points so that the curve you plot appears smooth.\n", " Label the axes $x$ and $y$.\n", "\n", "3. Plot the functions $\\sin x$ and $\\cos x$ *vs* $x$ on the same plot\n", " with $x$ going from $-\\pi$ to $\\pi$. Make sure the limits of\n", " $x$-axis do not extend beyond the limits of the data. Plot $\\sin x$\n", " in the color green and $\\cos x$ in the color black and include a\n", " legend to label the two curves. Place the legend within the plot,\n", " but such that it does not cover either of the sine or cosine traces.\n", "\n", "4. Create a data file with the data shown below.\n", "\n", " 1. Read the data into Python program and plot $t$ *vs* $y$ using\n", " circles for data points with error bars. Use the data in the\n", " `dy` column as the error estimates for the $y$ data. Label the\n", " horizontal and vertical axes \"time (s)\" and \"position (cm)\".\n", "\n", " 2. On the same graph, plot the function below as a smooth line.\n", " Make the line pass *behind* the data points.\n", "\n", " $$y(t) = \\left[3 + \n", " \\frac{1}{2}\\sin\\frac{\\pi t}{5}\\right]\n", " t\\, e^{-t/10}$$\n", "\n", " Data for Exercise 4\n", " Date: 16-Aug-2013\n", " Data taken by Lauren and John\n", "\n", " t d dy\n", " 1.0 2.94 0.7\n", " 4.5 8.29 1.2\n", " 8.0 9.36 1.2\n", " 11.5 11.60 1.4\n", " 15.0 9.32 1.3\n", " 18.5 7.75 1.1\n", " 22.0 8.06 1.2\n", " 25.5 5.60 1.0\n", " 29.0 4.50 0.8\n", " 32.5 4.01 0.8\n", " 36.0 2.62 0.7\n", " 39.5 1.70 0.6\n", " 43.0 2.03 0.6\n", "\n", "5. Use MatPlotLib's function `hist` along with NumPy's function's\n", " `random.rand` and `random.randn` to create the histogram graphs\n", " shown in Fig. `fig-randhistos`\n", "\n", "6. Plot force *vs* distance with error bars using the following data:\n", "\n", " d=np.array([0.38, 0.64, 0.91, 1.26, 1.41, 1.66, 1.90, 2.18])\n", " f=np.array([1.4, 1.65, 3.0, 3.95, 4.3, 5.20, 6.85, 7.4])\n", " df=np.array([ 0.4, 0.5, 0.4, 0.5, 0.6, 0.5, 0.5, 0.4])\n", "\n", " Your plot should also include a visual straight \"best fit\" to the\n", " data as well as visual \"fits\" that give the smallest and largest\n", " slopes consistent with the data. Note, you only need two points to\n", " define a straight line so the straight lines you draw on the plot\n", " should be arrays of length 2 and no longer. All of your fitted lines\n", " should lie *behind* the data. Try to make your plot look like the\n", " one below. *In addition*, add a legend to your plot the gives the\n", " slope with its uncertainty obtained from your visual fits to the\n", " data.\n", "\n", "
\n", " \"\"\n", "
\n", "\n", " The web page gives a\n", " summary of the main plotting commands available in MatPlotLib. The\n", " two important ones here are `plot` and `errorbar`, which make\n", " regular plots and plots with error bars, respectively. You will find\n", " the following keyword arguments useful: `yerr`, `ls`, `marker`,\n", " `mfc`, `mec`, `ms`, and `ecolor`, which you can find described by\n", " clicking on the `errorbar` function link on the web page cited\n", " above.\n", "\n", "7. The data file below shows data obtained for the displacement\n", " (position) *vs* time of a falling object, together with the\n", " estimated uncertainty in the displacement.\n", "\n", " > Measurements of fall velocity vs time\n", " > Taken by A.P. Crawford and S.M. Torres\n", " > 19-Sep-13 \n", " > time (s) position (m) uncertainty (m)\n", " > 0.0 0.0 0.04\n", " > 0.5 1.3 0.12\n", " > 1.0 5.1 0.2\n", " > 1.5 10.9 0.3\n", " > 2.0 18.9 0.4\n", " > 2.5 28.7 0.4\n", " > 3.0 40.3 0.5\n", " > 3.5 53.1 0.6\n", " > 4.0 67.5 0.6\n", " > 4.5 82.3 0.6\n", " > 5.0 97.6 0.7\n", " > 5.5 113.8 0.7\n", " > 6.0 131.2 0.7\n", " > 6.5 148.5 0.7\n", " > 7.0 166.2 0.7\n", " > 7.5 184.2 0.7\n", " > 8.0 201.6 0.7\n", " > 8.5 220.1 0.7\n", " > 9.0 238.3 0.7\n", " > 9.5 256.5 0.7\n", " > 10.0 275.6 0.8\n", "\n", " 1. Use these data to calculate the velocity and acceleration (in a\n", " Python program `.py` file), together with their uncertainties\n", " propagated from the displacement *vs* time uncertainties. Be\n", " sure to calculate time arrays corresponding the midpoint in time\n", " between the two displacements or velocities for the velocity and\n", " acceleration arrays, respectively.\n", " 2. In a single window frame, make three vertically stacked plots of\n", " the displacement, velocity, and acceleration *vs* time. Show the\n", " error bars on the different plots. Make sure that the time axes\n", " of all three plots cover the same range of times. Why do the\n", " relative sizes of the error bars grow progressively greater as\n", " one progresses from displacement to velocity to acceleration?" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "jupytext": { "cell_metadata_filter": "all", "encoding": "# -*- coding: utf-8 -*-", "formats": "ipynb,py:percent", "notebook_metadata_filter": "all,-language_info,-toc,-latex_envs" }, "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.7.6" } }, "nbformat": 4, "nbformat_minor": 2 }