Python functional programming

+ =f(x)
Python Functional Programming

Functional Programming by Wikipidia:
“Functional programming is a programming paradigm that treats
computation as the evaluation of mathematical functions and avoids
state and mutable data". In other words, functional programming
promotes code with no side effects, no change of value in
variables. It oposes to imperative programming, which enfatizes
change of state”.

What this means?
● No mutable data (no side effect).
● No state (no implicit, hidden state).
Once assigned (value binding), a variable (a symbol) does not change its value.
All state is bad? No, hidden, implicit state is bad.
Functional programming do not eliminate state, it just make it visible and explicit
(at least when programmers want it to be).
● Functions are pure functions in the mathematical sense: their output depend only
in their inputs, there is not “environment”.
● Same result returned by functions called with the same inputs.

What are the advantages?
● Cleaner code: "variables" are not modified once defined, so we don't have to
follow the change of state to comprehend what a function, a, method, a class, a
whole project works.
● Referential transparency: Expressions can be replaced by its values. If we call a
function with the same parameters, we know for sure the output will be the same
(there is no state anywhere that would change it).
There is a reason for which Einstein defined insanity as "doing the same thing over
and over again and expecting different results".

Advantages enabled by referential transparence
● Memoization
○ Cache results for previous function calls.
● Idempotence
○ Same results regardless how many times you call a function.
● Modularization
○ We have no state that pervades the whole code, so we build our project with
small, black boxes that we tie together, so it promotes bottom-up
programming.
● Ease of debugging
○ Functions are isolated, they only depend on their input and their output, so
they are very easy to debug.

● Parallelization
○ Functions calls are independent.
○ We can parallelize in different process/CPUs/computers/…
We can execute func1 and func2 in paralell because a won’t be modified.
result = func1(a, b) + func2(a, c)

● Concurrence
a. With no shared data, concurrence gets a lot simpler:
i. No semaphores.
ii. No monitors.
iii. No locks.
iv. No race-conditions.
v. No dead-locks.

Python is a multi paradigm programming language. As a Python
programmer why uses functional programming in Python?
Python is not a functional language but have a lot of features that enables us to
applies functional principles in the development, turning our code more elegant,
concise, maintanable, easier to understand and test.

Don’t Update, Create - String
name = 'Geison'
name = '{0} Flores'.format(name)
FIRSTNAME = 'Geison'
LASTNAME = '{0} Flores'.format(FIRSTNAME)
NAME = '{0} {1}'.format(FIRSTNAME, LASTNAME)

Don’t Update, Create - Lists
years = [2001, 2002]
years.append(2003)
years += [2004, 2005]
years # [2001, 2002, 2003, 2004, 2005]
YEARS = [2001, 2001]
ALL_YEARS = YEARS + [2003] + [2004, 2005]

Don’t Update, Create - Dict
ages = {'John': 30}
ages['Mary'] = 28
ages # {'John': 30, 'Mary': 28}
AGES = {'John': 30}
ALL_AGES = dict(AGES.itens() + {'Mary': 28}.itens())

Higher Order Functions
Functions and methods are first-class objects in Python, so if you want to pass a
function to another function, you can just treat it as any other object.
def caller(f):
f()
def say_hello(name):
return 'Hello {0}'.format(name)
caller(say_hello)

Higher Order Functions - Map
map(lambda word: word.upper(), ["milu", "rantanplan"])
# result ["MILU", "RANTANPLAN"]
def add_2(n):
n + 2
map(add_2, [1, 2, 3]) # result [3, 4, 5]

Higher Order Functions - Filter
filter(lambda word: len(word) == 4, ["milu", "rantanplan"]) # result ["MILU"]
def greater_than_10(num):
return num > 10
filter(greater_than_10, range(15)) # result [11, 12, 13, 14, 15]

Higher Order Functions - Reduce
import operator
reduce(operator.add, [1, 2, 3, 4, 5]) # result 15
def acumullator(a, b):
return len(a) + len(b)
reduce(acumullator, ["milu", "rantanplan"]) # result 14
reduce(lambda a,b: len(a) + len(b), ["milu", "rantanplan"]) # result 14

Higher Order Functions - Reduce
from itertools import izip
list1 = [1, 2, 3]
list2 = ["a", "b", "c"]
[list(x) for x in izip(list1, list2)] # result [[1, "a"], [2, "b"], [3, "c"]]

Higher Order Functions - Closure
def add_x(x):
def adder(y):
return x + y
return adder
add_5 = add_x(5)
add_7 = add_x(7)
add_5(10) # result 15
add_7(10) # result 17

Currying and Partial Functions
Higher-order functions enable Currying, which the ability to take a function that accepts n
parameters and turns it into a composition of n functions each of them take 1 parameter. A direct
use of currying is the Partial Functions where if you have a function that accepts n parameters then
you can generate from it one of more functions with some parameter values already filled in.
from functools import partial
plus = lambda a, b: a + b # defining a function that sums 2 numbers
plus(3, 5) # result 8
# curring calling partial function by supplying the first parameters with value 1
plus_one = partial(plus, 1)
# I can use the new function as normal
plus_one(5) # result 6

Eager vs Lazy Evaluation
● Eager evaluation: expressions are calculated at the moment that variables is
assined, function called...
● Lazy evaluation: delays the evaluation of the expression until it is needed.
○ Memory efficient: no memory used to store complete structures.
○ CPU efficient: no need to calculate the complete result before returning.
○ Laziness is not a requisite for FP, but it is a strategy that fits nicely on
the paradigm(Haskell).
Python uses eager evaluation (but short-circuits && or ||).
Python generators are a mechanism for lazy evaluation.
Python arrays are not lazy, use enumarators when necessary.

Recursion
Looping by calling a function from within itself. When you don’t have access to mutable
data, recursion is used to build up and chain data construction. This is because looping is
not a functional concept, as it requires variables to be passed around to store the state of
the loop at a given time.
● Purely functional languages have no imperative for-loops, so they use recursion a lot.
● If every recursion created an stack, it would blow up very soon.
● Tail-call optimization (TCO) avoids creating a new stack when the last call in a
recursion is the function itself.
● TCO is not implemented in Python.
● Unfortunarely following recursion style in Python has it’s own tax: Performance.

Solving Python Lack of TCO(Tail Call Optimization)
# The functional solution have problens with big values
fib = lambda n: n if n < 2 else fib(n-1) + fib(n-2)
# The iterative solution using generators works perfectly with large values
def fibs():
a = 0
b = 1
while True:
yield a
a, b = b, a + b

FP in OOP?
It is possible do FP in OOP? Yes it is!
● OOP is orthogonal to FP.
● Well, at least in theory, because:
○ Typical OOP tends to emphasize change of state in objects.
○ Typical OOP mixes the concepts of identity and state.
○ Mixture of data and code raises both conceptual and practical problems.
● OOP functional languages: Scala, F#, ...

A Pratical Example
Exercise: "What's the sum of the first 10 natural number whose square value is
divisible by 5?"
Imperative:
Functional:
n, num_elements, s = 1, 0, 0
while num_elements < 10:
if n**2 % 5 == 0:
s += n
num_elements += 1
n += 1
n #275
sum(filter(lambda x: x**2 % 5 == 0, range(1, 100))[:10])

The last advice
Learn at least one functional language, it will open your mind to a new paradigm
becoming you a better programmer.
Some Functional Languages:
● Haskell
● ML (Standard ML, Objective Caml, ...)
● Scheme
● Erlang
● Scala
● Closure
● F#

Conclusion
● As you can tell, Python helps you write in functional style but it doesn’t force
you to it.
● Writing in functional style enhances your code and makes it more self documented.
Actually it will make it more thread-safe also.
● The main support for FP in Python comes from the use of list conprehension,
lambdas, closures, iterators and generators, also from the modules functools and
itertools.
● Python still lack an important aspect of FP: Pattern Matching and Tails
Recursion.
● There should be more work on tail recursion optimization, to encourage developers
to use recursion.
● Any other thoughts?

References
● http://en.wikipedia.org/wiki/Functional_programming
● http://www.cse.chalmers.se/~rjmh/Papers/whyfp.pdf
● http://docs.python.org/2/howto/functional.html
● http://www.youtube.com/watch?v=Ta1bAMOMFOI
● http://clojure.org/
● http://www.defmacro.org/ramblings/fp.html

Contact me
● Email:
○ geisonfgf@gmail.com
● Skype
○ geisonfgf
● Facebook
○ http://www.facebook.com/geisonfgf
● Twitter
○ http://www.twitter.com/geisonfgf

Python functional programming

Recommended

More Related Content

What's hot (20)

Similar to Python functional programming (20)

More from Geison Goes (11)

Recently uploaded (20)

Python functional programming