- get a first peek at the very useful Python packages called
**NumPy**and**matplotlib**

In the last lecture we learned how to create modules. These are files that contain one or more functions and variables. They can be imported and used in other programs or notebooks, saving us a lot of time and headache.

A Python *package* contains a collection of modules that are related to each other. Our first *package* is one of the most useful ones for us science types: **NumPy**.

O.K. First of all - how do you pronounce "NumPy"'? It should be pronounced "Num" as in "Number" and "Pie" as in, well, pie, or Python. It is way more fun to say Numpee! I try to suppress this urge.

Now with that out of the way, what can we do with **NumPy**? Turns out, a whole heck of a lot! But for now, we will just scratch the surface. For starters, **NumPy** can give us the value of the square root of a number with the function **numpy.sqrt( )**. Note how the package name comes first, then the function we wish to use (just as in our example from the last lecture).

To use **NumPy** functions, we must first **import** the package with the command **import**. It may take a while the first time you use **import** after installing Python, but after that it should load quickly.

We encountered **import** very briefly in the last lecture. Now it is time to go deeper. There are many different ways you can use the **import** command. Each way allows your program to access the functions and variables defined in the imported package, but differs in how you call the function after importing:

In [1]:

```
import numpy
#This makes all the functions in NumPy available to you,
#but you have to call them with the numpy.FUNC() syntax
numpy.sqrt(2)
```

Out[1]:

In [2]:

```
# Here is another way to import a module:
import numpy as np # or any other variable e.g.: N
# This does the same as the first, but allows you to set NumPy with a nickname
# In this case, you substitute "np" for numpy:
np.sqrt(2) # or N.pi in the second case.
# Note: Some folks in the NumPy community use N; I use np.
# That seems to be the most common way now.
```

Out[2]:

To import all the functions from NumPy:

In [3]:

```
from numpy import *
# now all the functions are available directly, without the initial module name:
sqrt(2)
```

Out[3]:

The '*' imports all the functions into the local namespace, which is a heavy load on your computer's memory. Alternatively, you can import the few, specific functions you'll use, for example, **sqrt**:

In [4]:

```
from numpy import sqrt # square root
sqrt(4)
```

Out[4]:

Did you notice how "sqrt(4)", where 4 was an integer, returned a floating point variable (2.0)?

**TIP**: I tend to import the **NumPy** package using the **np** option above. That way I know where the functions I'm using come from. This is useful, becuase we don't use or know ALL of the functions available in any given package. AND the same function name can mean different things in different packages. So, a function defined in the package could conflict with one defined in your program. It is just good programming practice to specify the origin of the function you are using.

Here is a (partial) list of some useful **NumPy** functions:

function | purpose |
---|---|

absolute(x) | absolute value |

arccos(x) | arccosine |

arcsin(x) | arcsine |

arctan(x) | arctangent |

arctan2(y,x) | arctangent of y/x in correct quadrant |

cos(x) | cosine |

cosh(x) | hyperbolic cosine |

exp(x) | exponential |

log(x) | natural logarithm |

log10(x) | base 10 log |

sin(x) | sine |

sinh(x) | hyperbolic sine |

sqrt(x) | square root |

tan(x) | tangent |

tanh(x) | hyperbolic tangent |

**NumPy** has more than just *functions*; it also has *attributes* which are variables stored in the package, for example $\pi$.

In [5]:

```
np.pi
```

Out[5]:

**TIP**: In the trigonometric functions, the argument is in RADIANS!.You can convert between degrees and radians by multiplying by: np.pi/180. OR you can convert using the **NumPy** functions **np.degrees( )** which converts radians to degrees and **np.radians( )** which converts degrees to radians.

Also notice how the functions have parentheses, as opposed to **np.pi** which does not. The difference is that **np.pi** is not a function but an *attribute*. It is a variable defined in **NumPy** that you can access. Every time you call the variable **np.pi**, it returns the value of $\pi$.

As already mentioned, **NumPy** has many math functions. We will use a few to generate some data sets that we can then plot using **matplotlib**, another Python module.

First, let's make a list of angles ($\theta$ or **theta**) around a circle. We begin with the list of angles in degrees, convert them to radians (using **np.radians( )**), then construct a list of sines of those angles.

In [5]:

```
thetas_in_degrees=range(0,360,5) # list (generator) of angles from 0 to 359 at five degree intervals
# uncomment the following line, if you'd like to print the list
#print (list(thetas_in_degrees))
thetas_in_radians=np.radians(thetas_in_degrees) # convert to radians
sines=np.sin(thetas_in_radians) # calculate the sine values for all the thetas
sines
```

Out[5]:

Now that we've generated some data, we can look at them. Yes, we just printed out the values, but it is way more interesting to make a plot. The easiest way to do this is using the package **matplotlib** which has many plotting functions, among them a whole module called **pyplot**. By convention, we **import** the **matplotlib.pyplot** module as **plt**.

We've also included one more line that tells **pyplot** to plot the image within the notebook: The magic command: **%matplotlib inline**. Note that this does not work in other environments, like command line scripts; magic commands are only for Jupyter notebooks (lucky us!).

In [10]:

```
import matplotlib.pyplot as plt # import the plotting module
# call this magic command to show the plots in the notebook
%matplotlib inline
plt.plot(thetas_in_degrees,sines); # plot the sines with the angles
```

Every plot should at least have axis labels and can also have a title, a legend, bounds, etc. We can use **matplotlib.pyplot** to add these features and more.

In [11]:

```
# I want to plot the sine curve as a green line, so I use 'g-' to do that:
plt.plot(thetas_in_degrees,sines,'g-',label='Sine')
# the "label" argument saves this line for annotation in a legend
# let's add X and Y labels
plt.xlabel('Degrees') # make and X label
plt.ylabel('Sine') # label the Y axis
# and now change the x axis limits:
plt.xlim([0,360]) # set the limits
plt.title('Sine curve') # set the title
plt.legend(); # put on a legend!
```

Now let's add the cosine curve and a bit of style! We'll plot the cosine curve as a dashed blue line ('b--'), move the legend to a different position and plot the sine curve as little red dots ('r.'). For a complete list of possible symbols (markers), see: http://matplotlib.org/api/markers_api.html

In [12]:

```
cosines=np.cos(thetas_in_radians)
# plot the sines with the angles as a green line
plt.plot(thetas_in_degrees,sines,'r.',label='Sine')
# plot the cosines with the angles as a dashed blue line
plt.plot(thetas_in_degrees,cosines,'b--',label='Cosine')
plt.xlabel('Degrees')
plt.ylabel('Trig functions')
plt.xlim([0,360]) # set the limits
plt.legend(loc=3); # put the legend in the lower left hand corner this time
```

The function **plt.plot( )** in **matplotlib.pyplot** includes many more styling options. Here's a complete list of arguments and keyword arguments that plot accepts:

In [13]:

```
help(plt.plot)
```

One VERY useful function in **NumPy** is to read data sets into an array. Arrays are a new kind of data container, very much like lists, but with special attributes. Arrays must be all of one data type (e.g., floating point). Arrays can be operated on in one go, unlike lists that must be operated on element by element. I sneakily showed this to you by taking the cosine of the entire array returned by **np.radians( )**. It took a list and quietly turned it into an array, which I could operate on. Also, arrays don't separate the numbers with commas like lists do. We will see more benefits (and drawbacks) of arrays in the coming lectures.

The *built-in* function **range( )** makes a list generator as we have already seen. But the **NumPy** function **np.arange( )** makes and array. Let's compare the two:

In [14]:

```
print (list(range(10)))
print (np.arange(10))
```

They are superficially similar (except for the missing commas), but try a simple addition trick:

In [15]:

```
np.arange(10)+2
```

Out[15]:

versus

In [16]:

```
range(10)+2
```

Oh dear! We would have to go through the list one by one to do this addition using a list.

Time for some SCIENCE!

Let's start with data from an earthquake. We will read in data from an Earthquake available from the IRIS website: http://ds.iris.edu/wilber3/find_event. We can read in the data using the function **np.loadtext( )**.

I chose the Christmas Day, 2016 magnitude 7.6 Earthquake in Chile (latitude=-43.42, longitude=-73.95). It was recorded at a seismic station run by Scripps Institution of Oceanography called "Pinyon Flat Observatory" (PFO, latitude=33.3, longitude=-115.7).

In [17]:

```
EQ=np.loadtxt('Datasets/seismicRecord/earthquake.txt') # read in data
print (EQ)
```

Notice that EQ is NOT a **list** (it would have commas). In fact it is an N-dimensional array (actually only 1 dimensional in this case). You can find out what any object is using the built-in function **type( )**:

In [18]:

```
type(EQ)
```

Out[18]:

We'll learn more about the *ndarray* data structure in the next lecture.

But now, let's plot the earthquake data.

In [19]:

```
plt.plot(EQ); # the semi-colon suppresses some annoying jibberish,
# try taking it out!
```

Here, **plt.plot( )** plots the array **EQ** against the index number for the elements in the array because we didn't pass a second argument.

We can decorate this plot in many ways. For example, we can add axis labels and truncate the data with plt.xlim( ), or change the color of the line to name a few:

In [20]:

```
plt.plot(EQ,'r-') # plots as a red line
plt.xlabel('Arbitrary Time') # puts a label on the X axis
plt.ylabel('Velocity'); # puts a label on the Y axis
```

- Make a notebook and change the name of the notebook to: YourLastNameInitial_HW_02 (for example, CychB_HW_02)
- In a
**markdown**cell, write a description of what the notebook does - Create a
**Numpy**array of numbers from 0 to 100 - Create another list that is empty
- Write a
**for**loop that takes the square root of all the values in your list of numbers (using**np.sqrt**) and appends them to the empty list. - Print out all the numbers that are divisible by 4 (using the modulo operator).
- Plot the square roots against the original list.
Create a dictionary with at least 4 key:value pairs

Write your own module that contains at least four functions and uses a dictionary and a list. Include a doc string in your module and a comment before each function that briefly describes the function. Save it with the magic command %%writefile YOURMODULENAME.py

- Import the module into your notebook and call all of the functions.

Hint: For the purposes of debugging, you will probably want to 'reload' your module as you refine it. To do this

*from importlib import reload*

then

*reload(YOURMODULENAME)*

Your code must be fully commented.

In [ ]:

```
```