python.md

Python

---

Pythonic

• Exchange variables

  Good

  a = 1
  b = 2
  tmp = a
  a = b
  b = tmp

  Better

  a, b = b, a

---

PYTHONPATH

For example, one has a code named my_func.py contains resuable functions (e.g., FUNCTION_1, FUNCTION_2, ...).

For example, if one want to use FUNCTION_1 from my_func.py in any other code,

• old way

  If my_func.py does not exist in current working directory,

  • copy my_func.py to current working directory;

  • in new code, from my_func import FUNCTION_1.

• new way

  • put my_func.py under a directory, e.g., /Users/hchen/code/reuse/python;

  • in ~/.bashrc (or ~/.bash_profile), add

    export PYTHONPATH="${PYTHONPATH}:/Users/hchen/code/reuse/python"

    then

    source ~/.bashrc
    

  • now in new code, from my_func import FUNCTION_1.

---

matplotlib

• Change font to Times New Roman

  • On Mac

    After Microsoft Office has been installed, simply add Times New Roman to font.serif and font.sans-serif in matplotlibrc file.

    It will look something like:

    font.serif      : Times New Roman, DejaVu Serif, Bitstream Vera Serif ...
    font.sans-serif : Times New Roman, DejaVu Sans, Bitstream Vera Sans ...
    

  • On Linux

    dotfiles

• Plotting GeoTIFF

  • pyplot.imshow is the fastest way if the RGB array is large.

• Turn off scientific notation and offset of x/y labels

  ax.get_xaxis().get_major_formatter().set_scientific(False)
  ax.get_yaxis().get_major_formatter().set_scientific(False)
  
  ax.get_xaxis().get_major_formatter().set_useOffset(False)
  ax.get_yaxis().get_major_formatter().set_useOffset(False)
  

---

Cartopy

How to use it?

import cartopy

Draw a map

import cartopy.crs as ccrs

---

multiprocessing

Assume we have the following functions,

def FUNCTION_1(a):
    print(a**2)

def FUNCTION_2(a, b):
    print(a+b)

Use 8 processors to run FUNCTION_1 by passing 0, 1, 2, ..., 7 to a to the function.

import multiprocessing as mp

statements = np.arange(8)

pool = mp.Pool(processes=8)
pool.outputs = pool.map(FUNCTION_1, statements)
pool.close()
pool.join()

Use 8 processors to run FUNCTION_2 by passing 0, 1, 2, ..., 7 to a and 7, 6, 5, ..., 0 to b to the function.

First we need to modify Function_2 to

def FUNCTION_2(args):
    a, b = args
    print(a+b)

Then

import multiprocessing as mp

a = np.arange(8)
b = np.arange(8)[::-1]

statements = zip(a, b)

pool = mp.Pool(processes=8)
pool.outputs = pool.map(FUNCTION_2, statements)
pool.close()
pool.join()

---