Notes
Recursion
Slides by Christopher M. Bourke
Instructor: Berthe Y. Choueiry
Spring 2006
Computer Science & Engineering 235
Introduction to Discrete Mathematics
Sections 6.1 - 6.2 of Rosen
cse235@cse.unl.edu
Recursive Algorithms
Notes
A recursive algorithm is one in which objects are defined in terms
of other objects of the same type.
Advantages:
I
Simplicity of code
Easy to understand
Disadvantages:
I
Memory
Speed
Possibly redundant work
Tail recursion offers a solution to the memory problem, but really,
do we need recursion?
Recursive Algorithms
Analysis
Weve already seen how to analyze the running time of algorithms.
However, to analyze recursive algorithms, we require more
sophisticated techniques.
Specifically, we study how to define & solve recurrence relations.
Notes
Motivating Example
Notes
Factorial
Recall the factorial function.
1
if n = 1
n! =
n (n 1)! if n > 1
Consider the following (recursive) algorithm for computing n!:
Algorithm (Factorial)
1
2
3
4
5
6
Input
:nN
Output
: n!
if n = 1 then
return 1
end
else
return Factorial(n 1) n
end
Motivating Example
Notes
Factorial - Analysis?
How many multiplications M (n) does Factorial perform?
I
When n = 1 we dont perform any.
Otherwise we perform 1.
Plus how ever many multiplications we perform in the
recursive call, Factorial(n 1).
This can be expressed as a formula (similar to the definition of
n!.
M (0) = 0
M (n) = 1 + M (n 1)
This is known as a recurrence relation.
Recurrence Relations I
Definition
Definition
A recurrence relation for a sequence {an } is an equation that
expresses an in terms of one or more of the previous terms in the
sequence,
a0 , a1 , . . . , an1
for all integers n n0 where n0 is a nonnegative integer.
A sequence is called a solution of a recurrence relation if its terms
satisfy the recurrence relation.
Notes
Recurrence Relations II
Notes
Definition
Consider the recurrence relation: an = 2an1 an2 .
It has the following sequences an as solutions:
1. an = 3n,
2. an = 2n , and
3. an = 5.
Initial conditions + recurrence relation uniquely determine the
sequence.
Recurrence Relations III
Notes
Definition
Example
The Fibonacci numbers are defined by the recurrence,
F (n) = F (n 1) + F (n 2)
F (1) = 1
F (0) = 1
The solution to the Fibonacci recurrence is
!n
!n
1+ 5
1
1 5
1
fn =
2
2
5
5
(your book derives this solution).
Recurrence Relations IV
Notes
Definition
More generally, recurrences can have the form
T (n) = T (n ) + f (n),
T () = c
or
T (n) = T
n
+ f (n),
T () = c
Note that it may be necessary to define several T (), initial
conditions.
Recurrence Relations V
Notes
Definition
The initial conditions specify the value of the first few necessary
terms in the sequence. In the Fibonacci numbers we needed two
initial conditions, F (0) = F (1) = 1 since F (n) was defined by the
two previous terms in the sequence.
Initial conditions are also known as boundary conditions (as
opposed to the general conditions).
From now on, we will use
numbers are
fn
f1
f0
the subscript notation, so the Fibonacci
= fn1 + fn2
= 1
= 1
Recurrence Relations VI
Notes
Definition
Recurrence relations have two parts: recursive terms and
non-recursive terms.
2
10}
T (n) = 2T (n 2) + n
| {z
| {z }
recursive
non-recrusive
Recursive terms come from when an algorithm calls itself.
Non-recursive terms correspond to the non-recursive cost of the
algorithmwork the algorithm performs within a function.
Well see some examples later. First, we need to know how to
solve recurrences.
Solving Recurrences
There are several methods for solving recurrences.
I
Characteristic Equations
Forward Substitution
Backward Substitution
Recurrence Trees
Maple!
Notes
Linear Homogeneous Recurrences
Notes
Definition
A linear homogeneous recurrence relation of degree k with
constant coefficients is a recurrence relation of the form
an = c1 an1 + c2 an2 + + ck ank
with c1 , . . . , ck R, ck 6= 0.
I
Linear: RHS is a sum of multiples of previous terms of the
sequence (linear combination of previous terms). The
coefficients are all constants (not functions depending on n).
Homogeneous: no terms occur that are not multiples of the
aj s.
Degree k: an is expressed in terms of k terms of the sequence.
Linear Homogeneous Recurrences
Notes
Examples
Examples
The Fibonacci sequence is a linear homogeneous recurrence
relation. As are the following.
an = 4an1 + 5an2 + 7an3
an = 2an2 + 4an4 + 8an8
How many initial conditions do we need to specify for these? As
many as the degree, k = 3, 8 respectively.
So, how do we solve linear homogeneous recurrences?
Solving Linear Homogeneous Recurrences I
We want a solution of the form an = rn where r is some (real)
constant.
We observe that an = rn is a solution to a linear homogeneous
recurrence if and only if
rn = c1 rn1 + c2 rn2 + + ck rnk
We can now divide both sides by rnk , collect terms, and we get a
k-degree polynomial.
rk c1 rk1 c2 rk2 ck1 r ck = 0
Notes
Solving Linear Homogeneous Recurrences II
Notes
rk c1 rk1 c2 rk2 ck1 r ck = 0
This is called the characteristic equation of the recurrence relation.
The roots of this polynomial are called the characteristic roots of
the recurrence relation. They can be used to find solutions (if they
exist) to the recurrence relation. We will consider several cases.
Second Order Linear Homogeneous Recurrences
Notes
A second order linear homogeneous recurrence is a recurrence of
the form
an = c1 an1 + c2 an2
Theorem (Theorem 1, p414)
Let c1 , c2 R and suppose that r2 c1 r c2 = 0 is the
characteristic polynomial of a 2nd order linear homogeneous
recurrence which has two distinct1 roots, r1 , r2 .
Then {an } is a solution if and only if
an = 1 r1n + 2 r2n
for n = 0, 1, 2, . . . where 1 , 2 are constants dependent upon the
initial conditions.
1
we discuss how to handle this situation later.
Second Order Linear Homogeneous Recurrences
Example
Example
Find a solution to
an = 5an1 6an2
with initial conditions a0 = 1, a1 = 4
I
The characteristic polynomial is
r2 5r + 6
Using the quadratic formula (or common sense), the root can
be found;
r2 5r + 6 = (r 2)(r 3)
so r1 = 2, r2 = 3
Notes
Second Order Linear Homogeneous Recurrences
Notes
Example Continued
Using the 2nd-order theorem, we have a solution,
an = 1 (2n ) + 2 (3n )
Now we can plug in the two initial conditions to get a system
of linear equations.
a0 = 1 (2)0 + 2 (3)0
a1 = 1 (2)1 + 2 (3)1
1 = 1 + 2
(1)
4 = 21 + 32
(2)
Second Order Linear Homogeneous Recurrences
Notes
Example Continued
I
Solving for 1 = (1 2 )
second.
4 =
4 =
4 =
2 =
in (1), we can plug it into the
21 + 32
2(1 2 ) + 32
2 22 + 32
2
Substituting back into (1), we get
1 = 1
Putting it all back together, we have
an = 1 (2n ) + 2 (3n )
= 1 2n + 2 3n
Second Order Linear Homogeneous Recurrences
Another Example
Example
Solve the recurrence
an = 2an1 + 15an2
with initial conditions a0 = 0, a1 = 1.
If we did it right, we have
1
1
an = (3)n (5)n
8
8
How can we check ourselves?
Notes
Single Root Case
Notes
Recall that we can only apply the first theorem if the roots are
distinct, i.e. r1 6= r2 .
If the roots are not distinct (r1 = r2 ), we say that one
characteristic root has multiplicity two. In this case we have to
apply a different theorem.
Theorem (Theorem 2, p416)
Let c1 , c2 R with c2 6= 0. Suppose that r2 c1 r c2 = 0 has
only one distinct root, r0 . Then {an } is a solution to
an = c1 an1 + c2 an2 if and only if
an = 1 r0n + 2 nr0n
for n = 0, 1, 2, . . . where 1 , 2 are constants depending upon the
initial conditions.
Single Root Case
Notes
Example
Example
What is the solution to the recurrence relation
an = 8an1 16an2
with initial conditions a0 = 1, a1 = 7?
I
The characteristic polynomial is
r2 8r + 16
Factoring gives us
r2 8r + 16 = (r 4)(r 4)
so r0 = 4
Single Root Case
Notes
Example
I
By Theorem 2, we have that the solution is of the form
an = 1 4n + 2 n4n
Using the initial conditions, we get a system of equations;
a0 = 1 = 1
a1 = 7 = 41 + 42
Solving the second, we get that 2 =
And so the solution is
3
4
3
an = 4n + n4n
4
I
We should check ourselves. . .
General Linear Homogeneous Recurrences
Notes
There is a straightforward generalization of these cases to higher
order linear homogeneous recurrences.
Essentially, we simply define higher degree polynomials.
The roots of these polynomials lead to a general solution.
The general solution contains coefficients that depend only on the
initial conditions.
In the general case, however, the coefficients form a system of
linear inequalities.
General Linear Homogeneous Recurrences I
Notes
Distinct Roots
Theorem (Theorem 3, p417)
Let c1 , . . . , ck R. Suppose that the characteristic equation
rk c1 rk1 ck1 r ck = 0
has k distinct roots, r1 , . . . , rk . Then a sequence {an } is a solution
of the recurrence relation
an = c1 an1 + c2 an2 + + ck ank
if and only if
an = 1 r1n + 2 r2n + + k rkn
for n = 0, 1, 2, . . ., where 1 , 2 , . . . , k are constants.
General Linear Homogeneous Recurrences
Any Multiplicity
Theorem (Theorem 4, p418)
Let c1 , . . . , ck R. Suppose that the characteristic equation
rk c1 rk1 ck1 r ck = 0
has t distinct roots, r1 , . . . , rt with multiplicities m1 , . . . , mt .
Notes
General Linear Homogeneous Recurrences
Notes
Any Multiplicity
Theorem (Continued)
Then a sequence {an } is a solution of the recurrence relation
an = c1 an1 + c2 an2 + + ck ank
if and only if
an = (1,0 + 1,1 n + + 1,m1 1 nm1 1 )r1n +
(2,0 + 2,1 n + + 2,m2 1 nm2 1 )r2n +
..
.
(t,0 + t,1 n + + t,mt 1 nmt 1 )rtn +
for n = 0, 1, 2, . . ., where i,j are constants for 1 i t and
0 j mi 1.
Linear Nonhomogeneous Recurrences
Notes
For recursive algorithms, cost functions are often not homogenous
because there is usually a non-recursive cost depending on the
input size.
Such a recurrence relation is called a linear nonhomogeneous
recurrence relation.
Such functions are of the form
an = c1 an1 + c2 an2 + + ck ank + f (n)
Linear Nonhomogeneous Recurrences
Here, f (n) represents a non-recursive cost. If we chop it off, we
are left with
an = c1 an1 + c2 an2 + + ck ank
which is the associated homogenous recurrence relation.
Every solution of a linear nonhomogeneous recurrence relation is
the sum of a particular solution and a solution to the associated
linear homogeneous recurrence relation.
Notes
Linear Nonhomogeneous Recurrences
Notes
Theorem (Theorem 5, p420)
(p)
If {an } is a particular solution of the nonhomogeneous linear
recurrence relation with constant coefficients
an = c1 an1 + c2 an2 + + ck ank + f (n)
(p)
(h)
(h)
then every solution is of the form {an + an }, where {an } is a
solution of the associated homogenous recurrence relation
an = c1 an1 + c2 an2 + + ck ank
Linear Nonhomogeneous Recurrences
Notes
There is no general method for solving such relations. However, we
can solve them for special cases.
In particular, if f (n) is a polynomial or exponential function (or
more precisely, when f (n) is the product of a polynomial and
exponential function), then there is a general solution.
Linear Nonhomogeneous Recurrences
Theorem (Theorem 6, p421)
Suppose that {an } satisfies the linear nonhomogeneous recurrence
relation
an = c1 an1 + c2 an2 + + ck ank + f (n)
where c1 , . . . , ck R and
f (n) = (bt nt + bt1 nt1 + + b1 n + b0 ) sn
where b0 , . . . , bn , s R.
Notes
Linear Nonhomogeneous Recurrences
Notes
Theorem (Continued)
When s is not a root of the characteristic equation of the
associated linear homogeneous recurrence relation, there is a
particular solution of the form
(pt nt + pt1 nt1 + + p1 n + p0 ) sn
When s is a root of this characteristic equation and its multiplicity
is m, there is a particular solution of the form
nm (pt nt + pt1 nt1 + + p1 n + p0 ) sn
Linear Nonhomogeneous Recurrences
Notes
The examples in the text are quite good (see pp420422) and
illustrate how to solve simple nonhomogeneous relations.
We may go over more examples if you wish.
Also read up on generating functions in section 6.4 (though we
may return to this subject).
However, there are alternate, more intuitive methods.
Other Methods
When analyzing algorithms, linear homogenous recurrences of
order greater than 2 hardly ever arise in practice.
We briefly describe two unfolding methods that work for a lot of
cases.
Backward substitution this works exactly as its name implies:
starting from the equation itself, work backwards, substituting
values of the function for previous ones.
Recurrence Trees just as powerful but perhaps more intuitive,
this method involves mapping out the recurrence tree for an
equation. Starting from the equation, you unfold each recursive
call to the function and calculate the non-recursive cost at each
level of the tree. You then find a general formula for each level and
take a summation over all such levels.
Notes
Backward Substitution
Notes
Example
Example
Give a solution to
T (n) = T (n 1) + 2n
where T (1) = 5.
We begin by unfolding the recursion by a simple substitution of the
function values.
Observe that
T (n 1) = T ((n 1) 1) + 2(n 1) = T (n 2) + 2(n 1)
Substituting this into the original equation gives us
T (n) = T (n 2) + 2(n 1) + 2n
Backward Substitution
Notes
Example Continued
If we continue to do this, we get the following.
T (n) = T (n 2) + 2(n 1) + 2n
= T (n 3) + 2(n 2) + 2(n 1) + 2n
= T (n 4) + 2(n 3) + 2(n 2) + 2(n 1) + 2n
..
.
P
= T (n i) + i1
j=0 2(n j)
I.e. this is the functions value at the i-th iteration. Solving the
sum, we get
T (n) = T (n i) + 2n(i 1) + 2
(i 1)(i 1 + 1)
+ 2n
2
Backward Substitution
Example Continued
We want to get rid of the recursive term. To do this, we need to
know at what iteration we reach our base case; i.e. for what value
of i can we use the initial condition, T (1) = 5?
We can easily see that when i = n 1, we get the base case.
Substituting this into the equation above, we get
T (n) =
=
=
=
T (n i) + 2n(i 1) i2 + i + 2n
T (1) + 2n(n 1 1) (n 1)2 + (n 1) + 2n
5 + 2n(n 2) (n2 2n + 1) + (n 1) + 2n
n2 + n + 3
Notes
Recurrence Trees
Notes
When using recurrence trees, we graphically represent the
recursion.
Each node in the tree is an instance of the function. As we
progress downward, the size of the input decreases.
The contribution of each level to the function is equivalent to the
number of nodes at that level times the non-recursive cost on the
size of the input at that level.
The tree ends at the depth at which we reach the base case.
As an example, we consider a recursive function of the form
n
T (n) = T
+ f (n), T () = c
Recurrence Trees
Notes
Iteration
0
T (n/)
Cost
f (n)
T (n)
T (n/ 2 )
T (n/ 2 )
T (n/)
T (n/ 2 )
2
T (n/ 2 ) f
.
.
.
i
.
.
.
i
.
.
.
n
2
n
i
.
.
.
log n
log n
T ()
Recurrence Trees
Notes
Example
The total value of the function is the summation over all levels of
the tree:
log n
X
n
T (n) =
i f
i
i=0
We consider the following concrete example.
Example
T (n) = 2T
n
2
+ n,
T (1) = 4
Recurrence Trees
Notes
Example Continued
Iteration
0
T (n/2)
T (n/4)
Cost
n
T (n)
T (n/8)
T (n/8)
n + n
2
2
T (n/2)
T (n/4)
T (n/8)
T (n/8)
T (n/4)
T (n/8)
T (n/8)
T (n/4)
T (n/8)
T (n/8)
n
4
n
8
.
.
.
.
.
.
2i
0
@
n
2i
1
A
.
.
.
.
.
.
log2 n
2log2 n T (1)
Recurrence Trees
Notes
Example Continued
The value of the function then, is the summation of the value of
all levels. We treat the last level as a special case since its
non-recursive cost is different.
(log2 n)1
T (n) = 4n +
X
i=0
2i
n
= n(log n) + 4n
2i
Smoothness Rule I
In the previous example we make the following assumption: that n
was a power of two; n = 2k . This was necessary to get a nice
depth of log n and a full tree.
We can restrict consideration to certain powers because of the
smoothness rule.
Definition
A function f : N 7 R is called smooth if it is monotonically
nondecreasing and
f (2n) (f (n))
Most slow growing functions (logarithmic, polylogarithmic,
polynomial) are smooth while exponential functions are not.
Notes
Smoothness Rule II
Notes
Theorem
For a smooth function f (n) and a fixed constant b Z such that
b 2,
f (bn) (f (n))
Thus the order of growth is preserved.
How To Cheat With Maple I
Notes
Maple and other math tools are great resources. However, they are
not substitutes for knowing how to solve recurrences yourself.
As such, you should only use Maple to check your answers.
Recurrence relations can be solved using the rsolve command
and giving Maple the proper parameters.
The arguments are essentially a comma-delimited list of equations:
general and boundary conditions, followed by the name and
variable of the function.
How To Cheat With Maple II
> rsolve({T(n) = T(n-1) + 2*n, T(1) = 5}, T(n));
1
1 + 2(n + 1)
n + 1 2n
2
You can clean up Maples answer a bit by encapsulating it in the
simplify command:
> simplify(rsolve({T(n) = T(n-1) + 2*n, T(1) = 5},
T(n)));
3 + n2 + n
Notes
