# Flow Control and Logic

Click Here to download this tutorial as an iPython Notebook.

We’ve seen some basic math operations and data structures in Python, but to really tie it altogether, we will need a few more things. The first is how to make comparisons between two objects. The second is how to change the behavior of our programs given certain conditions. And lastly, we often need to do certain tasks repeatedly, so we’ll need a way to repeating or looping some instructions. We will now discuss how to make comparison tests, conditionals, branches, and loops in Python.

## Comparison Testing

Generally, we can compare two objects using one of the comparison operators.

Operator Operation Syntax Description
== Tests Equality a == b True if a is equivalent to b; else False
!= Tests Inequality a != b True if a is inequivalent to b; else False
<> Tests Inequality a <> b True if a is inequivalent to b; else False
> Tests greater-than a > b True if a is strictly greater than b; else False
< Tests less-than a < b True if a is strictly less than b; else False
>= Tests Greater-than-or-equal a >= b True if a is not less than b; else False
<= Tests Less-than-or-equal a <= b True if a is not greater than b; else False

Comparisons are by definition binary operations meaning you need two objects to compare. Usually, the objects need to be of the same (or logically similar) types. The result of a comparison is a boolean, ie, true or false. How this works is pretty straight-forward for numeric objects.

5 < 2
5 == 5
2 != 5


Many objects overload these operators so that they can be used intuitively with objects that are not numerics. With lists, for example:

# Lists are compared element-wise, item-by-item
[1, 2, 3] == [1, 2, 3] # Equal, element-by-element.
[1, 2, 3] == [1, 3, 2] # Same length, but 3 != 2.
[1, 2, 3] == [1, 2]    # Unequal lengths, so not equal.
[1, 2, 3] > [1, 2]     # Excess elements only considered if sublist checks out.
[1, 1, 3] > [1, 2]     # 1 < 2 so False.


Python doesn’t care if you mix floats and integers: they’re both numerics and Python knows how to handle that up to a certain exception:

0.2 == 2/5
1.0 == 1
0.999999999999999 == 1
0.99999999999999999 == 1 # Huh?


This last comparison is an example of floating-point representation error and all computers and programming languages suffer from it. Most of the time, this effect doesn’t matter but when it does (as is sometimes the case in science) there are ways around it. In Python, these issues can be skirted when necessary by a 3rd party library called ‘NumPy’ (For Numeric Python). We’ll talk about NumPy in detail later.

Comparisons can apply to non-numeric objects as well, but things can get tricky. Take strings for example:

# Strings are iterables and again, comparison between
#   iterables are character-by-character.
'Jake' == 'Jake'
'Finn' == 'Finns'

# Here's something neat:
'prismo' > 'Prismo'

# As seen below, Python gives priority to capital letters
# which means capitals are 'smaller' than lower case.
alpha = list('AaBbCc'); alpha.sort(); print(alpha)
numerics = list(range(5)); numerics.sort(); print(numerics)


So the message is this: many objects support comparison out-of-the-box, but you need to be aware of the rules used to make the comparison.

### is vs ==

When we saw the list of keywords, there was a comparison keyword is which can be used to test object identity. It is easy to confuse this with object equality.

1 == 1.0  # Literals that are equal...
1 is 1.0  # ... are not necessarily the same object.


The instruction a is b is shorthand for id(a) == id(b). Remember how assignment to a variable name is ordinarily just an alias? Here we see that again:

a = 5
b = [1, a, 3] # Make a list with 'a' as an element.
b[1] is a     # True!

# This is a bad example because 'a' aliases an immutable object.
# Let's try it again with a mutable object:
c = [1, 2, 3]   # Here we make a list (a mutable object)
c is [1, 2, 3]  # False because the '[1, 2, 3]' is a NEW list.
c == [1, 2, 3]  # True


### Other Comparison Keywords

Between now and the last tutorial, we’ve seen all comparison keywords. The set of them consists of is, not, and, or, and in. We are now in a position to understand more about how they work.

We glossed over in last time but now we can look at it more closely. Keyword in tests for an object’s membership in an iterable by comparing the object by value to all the items in an iterable.

1.0 in [1, 2, 3]  # True, because:
1.0 == 1          # is true by value, even though:
1.0 is not 1      # by identity.


As we see in the last line above, we can use the keyword not to negate keyword comparisons. We talked about and, not, and or in regards to booleans. We can use these keywords with comparisons because comparisons are themselves boolean statements,

1 in c               # Seen it. True!
c is [1, 2, 3]       # Seen it. False!
c is not [1, 2, 3]   # 'not' negates the statement. True! This is equivalent to
not (c is [1, 2, 3]) #   ie 'not (True)'
3 not in c           # False!
5 not in c           # True!


Comparisons can be chained together. Because comparison operators are binary, the statement doesn’t necessarily read like in pen-and-paper math:

4 < 7 == 7.0  # True, same as:
(4 < 7) and (7 == 7.0)

4 < 7 and 3 not in c # False!
4 < 7 or 3 not in c  # True!

(4 < 7) or (3 not in c)  # Same as above, but easier to read with parantheses.
((4 < 7) or 3) not in c  # Parentheses used to change order of comparison.


## Conditionals and Branching

Now we can use comparisons to alter program flow, also called branching. Branching in Python is done with the conditional keywords if, else, and elif. There are no unconditional branches in Python (like the dreaded and icky goto statement). For the record, there are also no case or switch branches in Python. All branching starts with the basic conditional if with the following syntax:

if boolean_statement:
# This code is run if condition is True
do_these_instructions
until_the_indentation
goes_back_to_normal

# This code is run after the branch.
outside_code


A few things to point out here are the colon : which marks the end of the conditional line. Omitting the colon is a syntax error. When boolean_statement evaluates to True, the code in the body of the if statement is executed. The body consists of all lines that are indented further than code outside the block, ie, the conditional statement. The amount of indentation is arbitrary, but common convention is four spaces. Using indentation and whitespace is a key design philosophy of Python: it forces programmers to write code blocks that are visually separated.

When Python reaches the end of the code block, execution continues with outside_code. When boolean_statment evaluates to False, the conditional body is skipped and execution immediately goes on to outside_code. Let’s look at some examples:

max_temperature = 2e-3 # Kelvin
min_temperature = 1e-3 # Kelvin
fridge_temperature = 273  # Kelvin

if fridge_temperature > max_temperature:
# The fridge is too hot.
print('The fridge is not cold enough!')

if fridge_temperature < min_temperature:
# Oops, too cold.
print('Turn on the heaters or abort the experiment.')


This example always makes two checks on a fridge’s temperature. We can optimize this a little bit and skip the second check when the first one is true by using the else statement.

if (max_temperature >= fridge_temperature
>= min_temperature):
# Fridge temperature within allowed range.
print('Temperature OK. Proceed with experiment.')
else:
print('Temperature out-of-range!')


Branches must have one and only one if statement and can optionally have a single else statement. Notice how the conditional line is wrapped? It is good practice to make the indentation of a wrapped line to be different than all lines around it so we don’t confuse it as being part of the branch body. PEP8 recommends indenting wrapped lines by two levels.

Suppose we want to do something specific depending on whether the out-of-range condition is hot or cold. We can use the elif statement to make additional comparisons when the preceeding ones are false. You can have as many additional elif statements as you like.

if (max_temperature >= fridge_temperature
>= min_temperature):
# Fridge temperature within allowed range.
print('Temperature OK. Proceed with experiment.')
elif fridge_temperature > max_temperature:
# Only run when above conditionals are false.
print('Temperature too high.')
elif False:
# This block will never run.
pass
else:
# Only run when all above conditionals are false.
print('Temperature too low.')


The above statement has a branch that will never run. However, an empty block is a syntax error. Comments do not qualify as filling a block because the interpereter ignores them. To fill a block but have it do nothing, we can use the pass keyword. When Python sees pass, it acknowledges the author intended the line to do nothing. Again, explicit is better than implicit.

### Inline if (Ternary Statements)

Often we want to conditionally run a single command or set a variable. Writing a full if-else branch to do this is tiresome. So Python, and many other languages implement a so-called ternary operator to do conditionals inline. The syntax for a ternary statement in Python is:

a if conditional else b


Notice there are no : characters or blocks and everything is in a single line. If the statement is so long and complicated that you need to break it up over multiple lines, then you shouldn’t be using a ternary statement. The output is all inline so this can be used for assignment operations. It is also useful in loops and some advanced topics like comprehensions.

adorables = ['bunny', 'puppy', 'kitty']
seen_object = 'bunny'
response = "d'awww" if seen_object in adorables else 'meh'
print(response)


## Loops

The last fundamental control flow in Python is repeating instructions in a loop. Python is a popular language for many reasons, but the way it handles looping is arguably one its best characteristics. There are two ways to loop in Python: for and while. Loops have some additional flow control actions namely continue, break, and else.

### For Loops

For loops repeat a block of code once for each item in a sequence. The syntax of such a loop uses the for and in keywords:

for item in sequence:
# Do this block of code for each 'item'.
print(item)

# End of loop is the end of the block indentation.


What can be used as a sequence? Lots of things. First, iterables like lists, strings, tuples, and sets are sequences. The output of the range built-in function (technically called a generator) is iterable. So are the keys of a dictionary. Any object upon which we can use the in keyword can also be for-looped over.

for i in (1, 2, 3):
print(i**2)

for _ in range(10):
# We don't need to use the iteration variable in the body.
# A good practice is to use '_' as the iterator name
# when the body doesn't make use of it.
print('Print this statement 10 times.')

quarks = ['up', 'down', 'charm', 'strange', 'top', 'bottom']
for quark in quarks:
# Variable 'quark' is assigned to current item in sequence.
charge = '2/3e' if quark[0] in 'uct' else '-1/3e'
print('Charge of ' + quark + ' is ' + charge)

for quark in quarks:
# Because lists are mutable, we can modify them while
# looping over them.
index = quarks.index(quark)
quarks[index].upper()


There are two things to learn from the last example. First is that we can modify a mutable sequence while looping over it. This can be useful, but is potentially dangerous! If we increase the length of the sequence each iteration, we can run into this:

# DANGER! DO NOT RUN THIS CODE...
# or do: I'm a comment, not a cop.
# If you choose to run this, you will enter an infinte loop that
# will fill your computer's memory. Use Ctrl-C to exit the loop.
quarks = ['up', 'down', 'charm', 'strange', 'top', 'bottom']
for quark in quarks:
# Appending to quarks while looping over it
#   causes a runaway condition.
# Even worse, the appended elements get bigger
#   and bigger until  you run out of memory.
quarks.append('anti-' + quark)


The safe way to do this is by looping over a copy of the object while modifying the original:

# This won't enter a runaway loop or consume all your memory.
quarks = ['up', 'down', 'charm', 'strange', 'top', 'bottom']
for quark in list(quarks):
# We used the list ctor to make a copy and
#   then we loop over the copy:
quarks.append('anti-' + quark)


The second thing to learn from the upper-case example, is that we often want the loop body to make use of both the item in a sequence and it’s position in the sequence. The best way to do this is by using the enumerate built-in function, which returns an iterable of (index, item) pairs from an iterable argument:

colors = ('red', 'green', 'blue')
for index, color in enumerate(colors):
print(index, color)
colors[index].upper()


So Python handles loops in a smart way. Compare the following simple example in several other languages:

// C/C++ needs very specifc iterator setup.
int a[] = { 1, 2, 3, 4, 5 };
for(int i = 0; i < (sizeof(a)/sizeof(*a)); i++)
{
printf("%d\n", i*i);
}

(* Mathematica can be hard to read for complex commands. *)
a = [1, 2, 3, 4, 5]
For[i = 0, i < Length[a], i++, Print[a[i]]]

# Python allows this, but considers it 'unpythonic':
a = [1, 2, 3, 4, 5]
for i in len(a):
print(a[i])

# The pythonic way is to write:
for i in a:
print(i)


The pythonic way is simple, clean, clear, and elegant with no need to check lengths or array sizes. Of course these are trivial examples. The difference really shines once your code starts to get complex.

### While Loops

While loops simply repeat until a test condition evaluates as False. The syntax goes as:

while conditional:
# Do the loop body if conditional is 'True'
some_body_code

# Once conditional is 'False', resume execution here.
outside_the_loop_code


Here are some examples of while loops:

# Sit in this loop until interrupted.
while True:
pass

# Compute the Fibonacci numbers less than 200
last, current = 0, 1
while current < 200:
print(current)
last, current = current, last + current


If the body of while loop doesn’t alter the test condition into a false state, the loop will continue until the the computation is interrupted either by the user (Ctrl-C in a shell) or by the machine (program crash, low memory).

### Loop Flow Control

Loops execution can be modified by some additional keyword statements. We can skip the remaining body of a loop for one iteration with the continue statement. This will advance the smallest enclosing loop immediately to the next iteration.

superheroes = ['spider man', 'wolverine', 'professor x',
'batman', 'jean grey', 'catwoman',
'superman', 'green lantern']

# Print only names that do not have '-man' suffix.
for hero in superheroes:
hero = hero.title()
if 'man' in hero:
continue
print(hero)

numbers = [1, 2, 3, 4, 5, 6]
# Print the odd squares in the list.
for number in numbers:
if not (number % 2):
continue
print(number, 'squared is', number**2)


We can use the break statement to skip all further iterations of the smallest enclosing loop.

for i in range(10):
if i == 5:
# When i == 5, stop the loop
break
print(i)
print('Loop terminated')


Lastly, a loop can take an else clause which is called when a loop finishes normally. If a loop ends due to break, the else clause is skipped.

primes = []
prime_candidate = 2
# Find the first 10 prime numbers.
while len(primes) < 10:
for factor in range(2, prime_candidate):
if prime_candidate % factor == 0:
# Non-prime
break
else:
# Loop didn't hit 'break'
print(prime_candidate, 'is prime')
primes.append(prime_candidate)
prime_candidate += 1


## Summary

We’ve seen how to do comparisons, conditional branches, and two types of loops. Next we will learn how to write our own functions so we can organize our code into reusable logical groups.

Now would also be a good time to take a look at data structure comprehensions.