CS1010S Programming Methodology

Lecture 10
Memoization, Dynamic
Programming & Exception Handling
26 Oct 2022
 Recap: OOP Concepts
• Class vs Instance
- Class is a template, which is used to create instances
pusheen = Cat("Pusheen")
Cat Cat
name name: 'Pusheen'
pusheen
lives lives: 9
say() say()

socCat
Cat
name: 'SoCCat'
lives: 9
soCCat = Cat("SoCCat")
say()
 Recap: Class
• Contains class Cat:
def __init__(self, name):
1. Fields/Properties
self.name = name
2. Methods/Functions self.lives = 9

Cat def say(self):

name print("meow")
lives __init__ method
is called on creation
say()
pusheen = Cat("Pusheen")
pusheen.say()
Meow
 Recap: Inheritance
Mammal
name
say()
Cat
name
is-a lives
Cat say()
lives
say()
scratch()
 Recap: Inheritance
Mammal class Mammal:
def __init__(self, name):
name
self.name = name
say() def say(self):
inherits print(self.name + " say")

override is-a
class Cat(Mammal):
Cat def __init__(self, name):
super().__init__(name)
lives
self.lives = 9
say() def say(self):
scratch() print("meow")
def scratch(self):
print(self.name,
"scratch!")
 Recap: Inheritance
Mammal >>> pusheen = Cat("Pusheen")
>>> pusheen.say()
name
meow
say()
inherits >>> pusheen.scratch()
Pusheen scratch!
override is-a
Cat >>> pusheen.lives
9
lives
say()
scratch()
 Recap: super()
Mammal super() refers to superclass
name self refers to instance
say()
class Cat(Mammal):
is-a super()
…
Cat def say(self):
lives super().say()
print("meow")
say()
scratch()
>>> pusheen.say()
self Pusheen say
meow
 Polymorphism
• Which superclass does super() resolve to?
- Depends on the type of the object/instance
Speaker
say def lecture(self, stuff):
self.say(…)
>>> seth = Lecturer() Lecturer
>>> seth.lecture(…) seth lecture
seth.say(…)
 Polymorphism
• Which superclass does super() resolve to?
- Depends on the type of the object/instance
Speaker
say def lecture(self, stuff):
self.say(…)
>>> seth = Lecturer() Lecturer
>>> seth.lecture(…) lecture
Arrogant
>>> ben = ArrogantLecturer(…)
Lecturer def say(self, stuff):
>>> ben.lecture(…)
ben say super().say(…)
ben.say()

Lecturer.lecture has different behaviour

 Multiple Inheritance
• Which superclass does super() resolve to?
- Order of declaration
class SingingArrogantLecturer(ArrogantLecturer, Singer)
Speaker
say
Lecturer
lecture
Arrogant Singer
def say(self, stuff): Lecturer say
super().say(…)
say sing
Singing
Arrogant
ben.say()
ben Lecturer
 Diamond Hierachy
• Intuition: All subclasses’ method must be called before superclass’
- Allows polymorphism to occur naturally

Speaker
say
Lecturer
lecture
Arrogant Singer def say(self, stuff):
def say(self, stuff): Lecturer say super().say(…)
super().say(…)
say sing
Singing
Arrogant
ben.say()
ben Lecturer
 Today’s Agenda
• Optional Arguments
See recording
• Exception Handling
• Memoization
• Dynamic Programming
Errors and Exceptions
 Errors and Exceptions
• Until now error messages haven’t been more than
mentioned, but you have probably seen some
• Two kinds of errors (in Python):
1. syntax errors
2. exceptions
 Syntax Errors
>>> while True print('Hello world')
SyntaxError: invalid syntax
 Exceptions
• Errors detected during execution are called
exceptions
• Examples:
- ZeroDivisonError,
- NameError,
- TypeError
 ZeroDivisionError
>>> 10 * (1/0)
Traceback (most recent call last):
File "<pyshell#3>", line 1, in <module>
10 * (1/0)
ZeroDivisionError: division by zero
 NameError
>>> 4 + spam*3
Traceback (most recent call last):
File "<pyshell#4>", line 1, in <module>
4 + spam*3
NameError: name 'spam' is not defined
 TypeError
>>> '2' + 2
Traceback (most recent call last):
File "<pyshell#5>", line 1, in <module>
'2' + 2
TypeError: Can't convert 'int' object to str
implicitly
 ValueError
>>> int('one')
Traceback (most recent call last):
File "<pyshell#2>", line 1, in <module>
int('one')
ValueError: invalid literal for int() with base 10:
'one'
 Handling Exceptions
The simplest way to catch and handle exceptions is with a try-
except block:

x, y = 5, 0
try:
z = x/y
except ZeroDivisionError:
print("divide by zero")
 Try-Except (How it works I)
• The try clause is executed
• If an exception occurred, skip the rest of the try
clause, to a matching except clause
• If no exception occurs, the except clause is skipped
(go to the else clause, if it exists)
• The finally clause is always executed before
leaving the try statement, whether an exception has
occurred or not.
 Try-Except
• A try clause may have more than 1 except clause,
to specify handlers for different exception.
• At most one handler will be executed.
• Similar with if-elif-else
• finally will always be executed
 Try-Except
try:
try:
# statements except Error1:
except Error1:
# handle error 1 #handle error 1
except Error2:
# handle error 2 except Error2:
except: # wildcard
# handle generic error else: #handle error 2
else:
# no error raised except:
finally: finally:
#handle
# always executed
generic error
 Try-Except Example
def divide_test(x, y):
try:
result = x / y
except ZeroDivisionError:
print("division by zero!")
else:
print("result is", result)
finally:
print("executing finally clause")
 Try-Except Blocks
>>> divide_test(2, 1) def divide_test(x, y):
try:
result is 2.0
result = x / y
executing finally clause
except ZeroDivisionError:
print("division by zero!")
>>> divide_test(2, 0)
else:
division by zero! print("result is", result)
executing finally clause finally:
print("executing finally
>>> divide_test("2", "1") clause")
executing finally clause
Traceback (most recent call last):
File "<stdin>", line 1, in ?
File "<stdin>", line 3, in divide
TypeError: unsupported operand type(s) for /: 'str' and 'str'
 Raising Exceptions
The raise statement allows the programmer to force a specific
exception to occur:

>>> raise NameError('HiThere')

Traceback (most recent call last):
File "<stdin>", line 1, in ?
NameError: HiThere
 Exception Types
• Built-in Exceptions:
https://docs.python.org/3/library/exceptions.html
• User-defined Exceptions
 User-defined Exceptions I
class MyError(Exception):
def __init__(self, value):
self.value = value
def __str__(self):
return repr(self.value)
 User-defined Exceptions II
try:
raise MyError(2*2)
except MyError as e:
print('Exception value:', e.value)
Exception value: 4

raise MyError('oops!')
Traceback (most recent call last):
File "<stdin>", line 1, in ?
__main__.MyError: 'oops!'
 Why use Exceptions?
In the good old days of C, many procedures
returned special ints for special conditions,
i.e. -1
 Why use Exceptions?
• But Exceptions are better because:
- More natural
- More easily extensible
- Nested Exceptions for flexibility
 Recall: Fibonacci
def fib(n):
if n < 2:
return n
else:
return fib(n-1) + fib(n-2)

Time complexity =
How can we do better? 𝑂(Φ𝑛) (exponential!)
 Computing Fibonacci
fib(5)

fib(4) fib(3)

fib(3) fib(2) fib(2) fib(1)

fib(1) fib(1) fib(0) fib(1) fib(0) 1

fib(2)

1 1 0 1 0
fib(1) fib(0)

1 0 Lots of duplicates
What’s the obvious
way to do better?
 Remember what you
had earlier computed!
Memoization
Notice the spelling,
NOT memorization
 Simple Idea!!
Function records, in a table, values that have previously been
computed.
• A memoized function:
- Maintains a table in which values of previous calls are
stored
- Use the arguments that produced the values as keys
• When the memoized function is called, check table to
see if the value exists:
- If so, return value.
- Otherwise, compute new value in the ordinary way and
store this in the table.
 Memoized Fib
def memo_fib(n):
table = {} Does not work!
def helper(n): New table is created at
if n < 2: each call!
return n
elif n in table: This function is our
return table[n] memoized fib
else:
table[n] = helper(n-1) + helper(n-2)
return table[n]
return helper(n)
 Memoized Fib
def memo_fib():
Change to a “creator” function
table = {}
def helper(n):
if n < 2: Return the memoized fib
return n
elif n in table:
return table[n]
else:
table[n] = helper(n-1) + helper(n-2)
return table[n]
return helper Define fib using the creator
fib = memo_fib()
 Computing Fibonacci
memo_fib(5)

memo_fib(4) memo_fib(3)

memo_fib(3) memo_fib(2)

memo_fib(2) memo_fib(1)

memo_fib(1) memo_fib(0)

1 0
Not “computed” but
by table lookup!
Compare to the Original Version
fib(5)

fib(4) fib(3)

fib(3) fib(2) fib(2) fib(1)

fib(1) fib(1) fib(0) fib(1) fib(0) 1

fib(2)

1 1 0 1 0
fib(1) fib(0)

1 0 Lots of duplicates
 Memoize any function
def memoize(fn): Write a “creator” function
table = {}
Use saved value
def helper(n):
if n in table:
Store value Does not work!
return table[n]
Restricts fn to functions
else:
with only 1 input
table[n] = fn(n)
return table[n]
return helper Define using the creator
fn = memoize(fn)
 The * notation
• Call a function for which you don't know in advance
how many arguments there will be using.

def funky(op, args):

return op(*args)
print(funky(lambda x: x*x, (2,))) → 4
print(funky(lambda x,y: x+y, (2, 1))) → 3
 The * notation
• Can also specify that a function takes in optionally many
arguments
def f(x, y, *z):
<body>

• function f can be called with 2, or more arguments.

• Calling f(1, 2, 3, 4, 5, 6): in the body,
- x → 1,
- y → 2,
- z → (3, 4, 5, 6)
 Suppose we want to nmap
• The function nmap
- takes as inputs a function and a sequence of sequences
- applies the function on the n-tuple
• Example
>>> nmap(lambda x, y, z: x+y+z,
[( 0, 1, 2, 3, 4),
( 5, 6, 7, 8, 9),
(10, 11, 12, 13, 14)])
[15, 18, 21, 24, 27]
 How to implement nmap
def nmap(fn, seqs): # assume all seqs same len
result = []
for i in range(len(seqs[0])):
args = map(lambda x: x[i], seqs)
result.append(fn(args))
return result
Error: fn does not take in a sequence
In fact, what does fn take?
It depends on what is passed in. i.e. beyond our control
 How to implement nmap
def nmap(fn, seqs): # assume all seqs same len
result = []
for i in range(len(seqs[0])):
args = map(lambda x: x[i], seqs)
result.append(fn(*args))
return result
Use * notation

Expand args to individual inputs which fn accepts

 Zip function
• Consider the function zip
- takes as inputs a sequence of n-sequences
- “zips” them up into a sequence of n-tuples
• Example: [(0, 5, 10),
zip([( 0, 1, 2, 3, 4), (1, 6, 11),
( 5, 6, 7, 8, 9), (2, 7, 12),
(10, 11, 12, 13, 14)]) (3, 8, 13),
(4, 9, 14)]
 Improved nmap
def nmap(fn, *seqs):
result = []
for seq in zip(seqs):
result.append(fn(*seq))
return result

>>> nmap(lambda x, y, z: x+y+z,

( 0, 1, 2, 3, 4),
( 5, 6, 7, 8, 9),
(10, 11, 12, 13, 14))
[15, 18, 21, 24, 27]
 Memoize any function (fix)
def memoize(fn): Use * notation
table = {}
def helper(*args):
if args in table:
return table[args]
else:
table[args] = fn(*args)
return table[args]
return helper
Redefine using the creator
fn = memoize(fn) aka Decorator
 𝑛
Another Example: 𝐶𝑘
def choose(n, k):
if k > n:
return 0
elif k == 0 or k == n:
return 1
else:
return choose(n-1, k) + choose(n-1, k-1)

Why is the recursion true?

 Remember Count-Change?
• Consider one of the elements 𝑥. 𝑥 is either chosen or
it is not.
• Then number of ways is sum of:
- Not chosen. Ways to choose 𝑘 elements out of remaining
𝑛 − 1 elements; and
- Chosen. Ways to choose 𝑘 − 1 elements out of remaining
𝑛 − 1 elements
 𝑛
Another Example: 𝐶𝑘
def choose(n, k):
if k > n:
return 0
elif k == 0 or k == n:
return 1
else:
return choose(n-1, k) + choose(n-1, k-1)

What is the order of growth?

How can we speed up the computation?
Memoization!
choose = memoize(choose)
 Chocolate Packing
• Suppose we are at a chocolate candy store and want
to assemble a kilogram box of chocolates.
• Some of the chocolates (such as the caramels) at this
store are absolutely the best, and others are only so-
so.
 Chocolate Packing
• We rate each one on a scale of 1 to 10, with 10 being
the highest.
• Each piece of chocolate has a weight; for example, a
caramel-flavoured chocolate weighs 13 grams.
• How do we put together the best box weighing at
most 1 kilogram?
 Abstract Data Type for Chocolates
def make_chocolate(desc,weight,value):
return (desc,weight,value)

def get_description(choc):
return choc[0]
def get_weight(choc):
return choc[1]
def get_value(choc):
return choc[2]
 Here are the Chocolates
shirks_chocolates =
(make_chocolate('caramel dark', 13, 10),
make_chocolate('caramel milk', 13, 3),
make_chocolate('cherry dark', 21, 3),
make_chocolate('cherry milk', 21, 1),
make_chocolate('mint dark', 7, 3),
make_chocolate('mint milk', 7, 2),
make_chocolate('cashew-cluster dark', 8, 6),
make_chocolate('cashew-cluster milk', 8, 4),
make_chocolate('maple-cream dark', 14, 1),
make_chocolate('maple-cream milk', 14, 1))
 Implementing a Box
def make_box(list_of_choc, weight, value):
return (list_of_choc, weight, value)

def make_empty_box():
return make_box((), 0, 0)

def box_chocolates(box):
return box[0]
def box_weight(box):
return box[1]
def box_value(box):
return box[2]
 Implementing a Box
def add_to_box(choc, box):
return make_box(
box_chocolates(box) + (choc,),
box_weight(box) + get_weight(choc),
box_value(box) + get_value(choc))

def better_box(box1, box2):

if box_value(box1) > box_value(box2):
return box1
else:
return box2
 How to Solve this Problem?
• Enumerate all the possible boxes (constrained by
weight limit)
• Compute value of each packing
• Pick box with highest value
 Simple Solution
def pick(chocs, weight_limit):
if chocs==() or weight_limit==0:
return make_empty_box()
elif get_weight(chocs[0]) > weight_limit: # 1st too heavy
return pick(chocs[1:], weight_limit)
else:
# none of 1st kind
box1 = pick(chocs[1:], weight_limit)
# at least one of 1st kind
box2 = add_to_box(chocs[0],
pick(chocs,
weight_limit -
get_weight(chocs[0])))
return better_box(box1, box2)
 Simple Solution
• What is the order of growth?
- Exponential! 𝑂(2𝑛)

• Again, a lot of repeat computations.

• Think memoization!
 Original Simple Solution
def pick(chocs, weight_limit):

if chocs==() or weight_limit==0:
return make_empty_box()
elif get_weight(chocs[0]) > weight_limit:
return pick(chocs[1:], weight_limit)
else:
box1 = pick(chocs[1:], weight_limit)
box2 = add_to_box(chocs[0],
pick(chocs,
weight_limit -
get_weight(chocs[0])))
return better_box(box1, box2)
 Memoized Version
def memo_pick(chocs, weight_limit):
def helper(chocs, weight_limit):
if chocs==() or weight_limit==0:
return make_empty_box()
elif get_weight(chocs[0]) > weight_limit:
return memo_pick(chocs[1:], weight_limit)
else:
box1 = memo_pick(chocs[1:], weight_limit)
box2 = add_to_box(chocs[0],
memo_pick(chocs,
weight_limit -
get_weight(chocs[0])))
return better_box(box1, box2)
return memoize(helper, "pick")(chocs, weight_limit)
 Recap: Memoization
• Two Steps:
1. Write function to perform required computation
2. Add wrapper that stores the result of the computation
(𝑂(1) lookup table)

• When you wrap your function, just make sure that he

recursive calls go to the wrapped version and not the
raw form.
 Homework
Re-factor the chocolate packing code into OOP format.
 Design Pattern: Wrapper
(also called Decorator)
• Memoization as an idea is simply to
remember the stuff you have done before
so that you don’t do the same thing twice
• However, the method that we used to
implement memoization is also an
important concept
 Design Pattern: Wrapper
(also called Decorator)
• Design Pattern: Wrapper (also known as
Decorator)
• Key idea is that you add an extra layer to
introduce additional functionality and use
the original function to do “old work”
Revisting memo_choose
• Consider 1 #f #f #f #f
memo_choose(8, 4) 1 1 #f #f #f
• The following is the table 1 2 1 #f #f
of values at the end of the 1 3 3 1 #f
computation: 1 4 6 4 1
1 5 10 10 5
#f 6 15 20 15
#f #f 21 35 35
#f #f #f 56 70
Recall Pascal’s Triangle
• If we were to fill up the 1 0 0 0 0
table, we expect 1 1 0 0 0
1 2 1 0 0
1 3 3 1 0
1 4 6 4 1
1 5 10 10 5
1 6 15 20 15
1 7 21 35 35
1 8 28 56 70
 Dynamic Programming
• Idea: why don’t we compute choose by filling up this
table from the bottom?
• Fancy name for this simple idea — Dynamic
Programming :-)
• What is the order of growth then?
𝑂(𝑛) or more accurately 𝑂(𝑛𝑘)
 Dynamic Programming: choose
def dp_choose(n, k):
row = [1] * (k+1) allocate
table = [] space for
for i in range(n+1): table (all 1’s)
table.append(row.copy())

for j in range(1, k+1): Fill first row

table[0][j] = 0 with 0’s

for i in range(1, n+1):

for j in range(1, k+1): fill the
table[i][j] = table[i-1][j-1] rest
+ table[i-1][j]
return table[n][k]
return answer
Let’s check out the table
k+1
row = [1] * (k+1) 1 1 1 1 1
table = []
for i in range(n+1):
table.append(
row.copy())
Let’s check out the table
k+1
𝐶00 0
𝐶𝑘
row = [1] * (k+1) 1 1 1 1 1
table = []
1 1 1 1 1
for i in range(n+1):

n+1
table.append(
1 1 1 1 1
row.copy()) 1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
𝑛1 1 1 1 1 𝑛
𝐶 0 𝐶𝑘
Let’s check out the table
j
for j in range(1,k+1): 1 0 0 0 0
table[0][j] = 0
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
Let’s check out the table
for j in range(1,k+1): 1 0 0 0 0
table[0][j] = 0
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
Let’s check out the table
j
for i in range(1,n+1): 1 0 0 0 0
for j in
i
1 1 1 1 1
range(1,k+1):
table[i][j] =
1 1 1 1 1
table[i-1][j-1] + 1 1 1 1 1
table[i-1][j] 1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
Let’s check out the table
j
for i in range(1,n+1): 1 0 0 0 0
for j in
i
1 1 0 1 1
range(1,k+1):
table[i][j] =
1 1 1 1 1
table[i-1][j-1] + 1 1 1 1 1
table[i-1][j] 1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
Let’s check out the table
j
for i in range(1,n+1): 1 0 0 0 0
for j in
i
1 1 0 0 1
range(1,k+1):
table[i][j] =
1 1 1 1 1
table[i-1][j-1] + 1 1 1 1 1
table[i-1][j] 1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
Let’s check out the table
j
for i in range(1,n+1): 1 0 0 0 0
for j in
i
1 1 0 0 0
range(1,k+1):
table[i][j] =
1 1 1 1 1
table[i-1][j-1] + 1 1 1 1 1
table[i-1][j] 1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
Let’s check out the table
j
for i in range(1,n+1): 1 0 0 0 0
for j in
i
1 1 0 0 0
range(1,k+1):
table[i][j] =
1 1 1 1 1
table[i-1][j-1] 1 1 1 1 1
+ table[i-1][j] 1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
Let’s check out the table
j
for i in range(1,n+1): 1 0 0 0 0
for j in
i
1 1 0 0 0
range(1,k+1):
table[i][j] =
1 2 1 1 1
table[i-1][j-1] + 1 1 1 1 1
table[i-1][j] 1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
Fast Forward
1 0 0 0 0
1 1 0 0 0
1 2 1 0 0
1 3 3 1 0
1 4 6 4 1
1 5 10 10 5
1 6 15 20 15
1 7 21 35 35
 Can we adopt a dynamic
programming approach for solving
the chocolate packing problem?

YES (of course)!

How?
Let’s take another look at the chocolate
packing problem…
Remember the table .... dynamic
programming is about filling up some table
efficiently so that the total time complexity is
𝑂(𝑠𝑖𝑧𝑒 𝑜𝑓 𝑡𝑎𝑏𝑙𝑒)
 What does the chocolate packing table look
like?
• Table of list of chocolates vs weight limit
• Each table entry is the optimal choice for a given list
and weight limit.
 What does the chocolate packing table look
like?
• Observation: If we have no chocolates, answer is, easy:
Empty set ().
• Otherwise, the answer for a list of 𝑥 types of chocolates is
the better between:
1. Additional chocolate + Optimal choice with remaining 𝑥 types of
chocolates and reduced weight limit
2. Optimal choice with remaining 𝑥 − 1 types of chocolates and
current weight limit.
 Building the chocolate packing table
Idea: build a table from the smaller cases to larger
cases
Weight Limit ()
0 ()
1 ()
2 ()
3 ()
4 ()
5 ()
6 ()
7 ()
8 ()
… ()
Building the chocolate packing table
Weight Limit () ((A 4 2))
0 () ()
1 () ()
2 () () +A
3 () ()
4 () (A)
5 () (A)
6 () (A)+A
7 () (A)
8 () (A,A)
…

()

…
Building the chocolate packing table
((A 4 2)
Weight Limit () ((A 4 2))
(B 5 3 ))
0 () () ()
1 () () ()
2 () () () Cannot
3 () () () add B
4 () (A) (A)
5 () (A) (B)
6 () (A)
7 () (A)
8 () (A,A)
()
…

…
Building the chocolate packing table
((A 4 2)
Weight Limit () ((A 4 2))
(B 5 3 ))
0 () () ()
1 () () ()
2 () () ()
3 () () ()
4 () (A) (A) +B
5 () (A) B
6 () (A) B +B
7 () (A) B
8 () (A,A)
…

()
…
Building the chocolate packing table
((A 4 2)
Weight Limit () ((A 4 2))
(B 5 3 ))
0 () () ()
1 () () ()
2 () () ()
3 () () ()
4 () (A) (A)
5 () (A) B
6 () (A) B
7 () (A) B +B
8 () (A,A) (A,A)
…

()
…

…
Challenge of the Day: Write dp_pick_chocolates over the weekend. ☺
 Prime Numbers
• In recitation, we defined a function is_prime to
check whether a number is prime
• But what if we wanted to list ALL of the numbers that
are prime, in the interval [0, … , 𝑛]?
 Prime Numbers: Naïve Solution
def is_prime(n): # O(n0.5)
if n==0 or n==1:
return False
elif n == 2:
return True
for i in range(2, int(sqrt(n))+1):
if n % i == 0:
return False
return True

def naive_prime(n): # O(n1.5)

return [is_prime(i) for i in range(n+1)]
 Prime Numbers
Idea: why don’t we compute the
primes by filling up a table from
the bottom?
 Prime Numbers: DP
def dp_prime(n):
bitmap = [True]*(n+1)
bitmap[0] = False # 0 is not prime
bitmap[1] = False # 1 is not prime
for i in range(2,n):
if bitmap[i] == True:
for j in range(2*i,n+1,i):
bitmap[j] = False
return bitmap
 Prime Numbers: DP
How does it work?
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13
 Summary
• Memoization dramatically reduces computation.
- Once a value is computed, it is remembered in a table (along with
the argument that produced it).
- The next time the procedure is called with the same argument, the
value is simply retrieved from the table.
• memo_fib takes time = 𝑂(𝑛)
• memo_choose takes ?? time?
Memoization vs Dynamic Programming
• Sometimes DP requires more computations
• DP requires the programmer to know
exactly which entries need to be computed
• For smart programmer, DP can however be
made more space efficient for some
problems, i.e. limited history recurrences
like Fibonacci