( ) 2022

9 00 – 12 30
Mathematical and Computing Science Time 9:00AM – 12:30PM

1. A, B, C 2
2. 1 9 3
3.
4.
5. 1 1
6.

Instruction

1. Solve 2 problems out of Problems A, B, and C.

2. Solve 3 problems out of Problems 1 to 9.
3. Note that if you solved more problems than specified above, problems you solved might
not be scored.
4. Write the problem number and your examinee number in the designated place of each
answer sheet.
5. Use one answer sheet per problem.
6. You may use the other side of the answer sheet, but in that case you should indicate
that, for example, by writing “continue to the other side.”

1
 A
x, y ∈ Rd 1 d d 2 d×d
A

A = xy T − yxT
√
zT z z ∈ Rd z = zT z

(1) d = 2 A5 = A x, y ∈ R2
(2) A2 x y

(3) A2 x, y
(4) A3 = A

2
Problem A
Let x, y ∈ Rd be linearly independent d-dimensional column vectors. The dimension d
is assumed to be greater than or equal to 2. The d × d matrix A is defined by

A = xy T − yxT ,
√
where z T is the transpose of z. The norm of z ∈ Rd is denoted by z = z T z. In the
answer, the Cauchy-Schwarz inequality and the condition of equality can be used without
proof.

(1) For d = 2, find a pair of x, y ∈ R2 such that A5 = A holds.

(2) Show that each eigenvector of non-zero eigenvalues for A2 is expressed by a linear
sum of x and y.
(3) Express non-zero eigenvalues of A2 in terms of x, y and their norms.
(4) Show that A3 = A.

3
 B
A(x) B(y) R C 2- C(x, y) R2 C 2-

(1) f (x, y)
f (x, y) = A(x) + xB(y)
f A B xfxy = fy
2
fy = ∂f /∂y, fxy = ∂ f /∂x∂y
(2) g(x, y)
g(x, y) = A(x)B(y)
gxy g gx gy A B
(3) h(x, y) = C(x + y, x − y) hxx (x, y) − hyy (x, y)
(4) k(x, y)
k(x, y) = A(x + y)B(x − y)
(kxx − kyy )k kx ky A B

4
Problem B
Let A(x) and B(y) be real-valued C 2 -functions on R, and let C(x, y) be a real-valued
C 2 -function on R2 . Answer the following questions:
(1) Show that the function defined by

f (x, y) = A(x) + xB(y)

satisfies the relation xfxy = fy not containing A and B, where fy = ∂f /∂y and
fxy = ∂ 2 f /∂x∂y etc.
(2) Consider the function defined by

g(x, y) = A(x)B(y).

Show that the product gxy g can be expressed in terms of gx and gy without use of A
and B.
(3) By setting h(x, y) = C(x + y, x − y), compute hxx (x, y) − hyy (x, y).
(4) Consider the function defined by

k(x, y) = A(x + y)B(x − y).

Show that the product (kxx − kyy )k can be expressed in terms of kx and ky without
use of A and B.

5
 C
x, y, . . . ∧( ), ∨ ( ), ¬ ( )
ϕ, ψ
0( ) 1( ) f ϕ
f (ϕ) f (x) = 0, f (y) = 1 f (x ∨ ¬(y ∧ ¬x)) = 0
ϕ f (ϕ) = 1 f

(1)

(x ∨ y ∨ z) ∧ ¬(y ∧ z) ∧ ¬x

(2)
ϕ ψ ϕ∧ψ

(3)
ϕ ψ ϕ∨ψ

(4) ϕ1 , ϕ2 , . . . , ϕn (n ≥ 2) ϕ+

(ϕ1 ∨ ¬x1 ) ∧ (x1 ∨ ϕ2 ∨ ¬x2 ) ∧ (x2 ∨ ϕ3 ∨ ¬x3 ) ∧ · · · ∧ (xn−2 ∨ ϕn−1 ∨ ¬xn−1 ) ∧ (xn−1 ∨ ϕn )

x1 , x2 , . . . , xn−1 ϕ 1 , ϕ 2 , . . . , ϕn

(4-1) ϕ1 ∨ ϕ2 ∨ · · · ∨ ϕn ϕ+
(4-2) ϕ+ ϕ1 ∨ ϕ2 ∨ · · · ∨ ϕn

6
Problem C
Logical formulas are constructed from propositional variables x, y, . . . by using the logical
connectives ∧ (conjunction), ∨ (disjunction), and ¬ (negation). We use letters ϕ, ψ, . . .
to denote logical formulas. A truth assignment is a function that assigns a truth value
0 (false) or 1 (true) to each propositional variable. For a truth assignment f , f (ϕ)
denotes the truth value of ϕ under f . For example, if f (x) = 0 and f (y) = 1, then
f (x ∨ ¬(y ∧ ¬x)) = 0. A logical formula ϕ is said to be satisfiable if and only if there
exists a truth assignment f such that f (ϕ) = 1.

(1) Determine and explain whether the following formula is satisfiable or not.

(x ∨ y ∨ z) ∧ ¬(y ∧ z) ∧ ¬x

(2) Determine whether the following statement holds.

If both ϕ and ψ are satisfiable, then ϕ ∧ ψ is satisfiable.
If this statement holds, prove it. If not, give a counterexample.
(3) Determine whether the following statement holds.
If either ϕ or ψ is satisfiable, then ϕ ∨ ψ is satisfiable.
If this statement holds, prove it. If not, give a counterexample.
(4) Given formulas ϕ1 , ϕ2 , . . . , ϕn (n ≥ 2), a formula ϕ+ is defined as

(ϕ1 ∨ ¬x1 ) ∧ (x1 ∨ ϕ2 ∨ ¬x2 ) ∧ (x2 ∨ ϕ3 ∨ ¬x3 ) ∧ · · · ∧ (xn−2 ∨ ϕn−1 ∨ ¬xn−1 ) ∧ (xn−1 ∨ ϕn ),

where the propositional variables x1 , x2 , . . . , xn−1 are mutually distinct and do not
appear in ϕ1 , ϕ2 , . . . , ϕn . Prove the following two statements.
(4-1) If ϕ1 ∨ ϕ2 ∨ · · · ∨ ϕn is satisfiable, then ϕ+ is satisfiable.
(4-2) If ϕ+ is satisfiable, then ϕ1 ∨ ϕ2 ∨ · · · ∨ ϕn is satisfiable.

7
 1
p p Z/pZ Fp GL2 (Fp ) Fp 2×2

SL2 (Fp ) = {M ∈ GL2 (Fp ) ; det(M ) = 1}

 
a b
det( ) = ad − bc 2 × 2
c d

(1) (∗) G H 1

(∗) H G
(2) SL2 (Fp ) GL2 (Fp )
(3) GL2 (F3 ) 2
(4) GL2 (F5 )/SL2 (F5 ) 4

8
Problem 1
For a prime number p, we denote by Fp the finite prime field Z/pZ. We define GL2 (Fp )
to be the group of 2 × 2 invertible matrices whose entries belong to Fp and define the
subgroup SL2 (Fp ) of GL2 (Fp ) by

SL2 (Fp ) = {M ∈ GL2 (Fp ) ; det(M ) = 1},

 
a b
where det( ) = ad − bc stands for the determinant of a 2 × 2 matrix. Answer the
c d
following questions.

(1) Give an example of a group G and its subgroup H that satisfies the condition (∗)
below. When answering, provide an explanation why it is such an example.
(∗) H is not a normal subgroup of G
(2) Prove that SL2 (Fp ) is a normal subgroup of GL2 (Fp ).
(3) Find all the elements of GL2 (F3 ) that have order 2.
(4) Show that the quotient group GL2 (F5 )/SL2 (F5 ) is a cyclic group of order 4.

9
 2
X

(1) X
X
(2) X × X
Δ = {(x, x) ∈ X × X ; x ∈ X}
X
(3) X ∼

[x] = {y ∈ X ; y ∼ x}

x∈X

X/∼ = {[x] ; x ∈ X}

X ∼ p:X x → [x] ∈ X/∼

V = {V ⊂ X/∼ ; p−1 (V ) X }

X/∼
(4) ∼

R = {(x, y) ∈ X × X ; x ∼ y}

X ×X p : X → X/∼
(X/∼, V)

10
Problem 2
Let X be a topological space. Answer the following questions.
(1) Give the definition of Hausdorff space. Suppose that X is a Hausdorff space. Deter-
mine the topology of X under the assumption that X is a finite set.
(2) Consider the subset
Δ = {(x, x) ∈ X × X ; x ∈ X}
of X ×X. Show that X is a Hausdorff space if Δ is closed with respect to the product
topology on X × X.
(3) Let ∼ be an equivalence relation on X, and denote by

[x] = {y ∈ X ; y ∼ x}

the equivalence class of x ∈ X. Moreover, let

X/∼ = {[x] ; x ∈ X}

be the quotient set of X by the equivalence relation ∼ and denote by p : X x→

[x] ∈ X/∼ the canonical projection. Show that

V = {V ⊂ X/∼ ; p−1 (V ) is an open subset of X}

is a topology on the set X/∼.

(4) Let ∼ be the equivalence relation as in (3), and consider the subset

R = {(x, y) ∈ X × X ; x ∼ y}

in X × X. Suppose that R is closed with respect to the product topology on X × X

and the projection p : X → X/∼ is an open map. Then show that (X/∼, V) is a
Hausdorff space.

11
 3

 1
f (y) + y
f (x) = dy (*)
0 ex+y
[0, 1] f
[0, 1] C[0, 1]

f  = sup{|f (x)| ; 0 ≤ x ≤ 1}

X
(1) f  X
(2) X f  1
f (y) + y
(T f )(x) = dy
0 ex+y
T X X
(3) (*) X

12
Problem 3
Consider the integral equation
 1
f (y) + y
f (x) = dy, (*)
0 ex+y
where f is a real-valued continuous function over the interval [0, 1].
Let X be a Banach space of real continuous functions over [0, 1] with the norm

f  = sup{|f (x)| ; 0 ≤ x ≤ 1}.

Answer the following questions.

(1) Show that f  really defines the norm over X.

(2) For f ∈ X, let  1
f (y) + y
(T f )(x) = dy.
0 ex+y
Answer, with a reason, whether T defines a linear map from X to X or not.
(3) Show that (*) has a unique solution in X.

13
 4
(P )

: cT x
(P ) : Ax ≤ b
x ≥ 0.

A ∈ Rm×n , b ∈ Rm , c ∈ Rn x ∈ Rn
0 T
u, v u≤v

(1) b = 0 (P )
(2) b = 0 (P ) 0
(3)
⎛ ⎞
3
   
1 7 1 2 7 ⎜−2⎟
A= , b= , c=⎜
⎝−2⎠
⎟
1 2 −1 2 4
−2

(P )

14
Problem 4
Consider a linear programming problem (P ) as follows:

maximize : cT x
(P ) subject to : Ax ≤ b
x ≥ 0.

Here, A ∈ Rm×n , b ∈ Rm , c ∈ Rn are input data, x ∈ Rn is a variable vector, and 0 is the

zero vector. The superscript T indicates the transpose of a vector, and u ≤ v denotes
the element-wise inequality for vectors u and v.
Answer the following questions.

(1) Suppose b = 0. Show that (P ) has a feasible solution.

(2) Suppose b = 0. Show that, if (P ) has an optimal solution, the optimal value is equal
to 0.
(3) Suppose that the input data are given as follows:
⎛ ⎞
3
   
1 7 1 2 7 ⎜−2⎟
A= , b= , and c = ⎜ ⎟
⎝−2⎠ .
1 2 −1 2 4
−2

Obtain an optimal solution of (P ).

15
 5
R B(R) R f, g R
A ∈ B(R) r∈R

A = {x ∈ R : f (x) ≤ g(x)},
 ∞
r= min f (x), g(x) dx
−∞

P X, Y
f, g

(1) P(X ∈ A) + P(Y ∈ Ac ) = r Ac A

(2) P(X = Y ) ≤ r X Y

(3) |P(X ∈ A) − P(Y ∈ A)| = |P(X ∈ Ac ) − P(Y ∈ Ac )| = 1 − r

(4) B ∈ B(R) |P(X ∈ B) − P(Y ∈ B)| ≤ 1 − r

16
Problem 5
Let R and B(R) denote the set of real numbers and the Borel field on R, respectively.
Let f and g be both probability density functions on R, and define A ∈ B(R) and r ∈ R
as

A = {x ∈ R : f (x) ≤ g(x)},
 ∞
r= min f (x), g(x) dx.
−∞

In the following, P denotes the probability and the marginal probability distributions of
random variables X and Y are determined by f and g, respectively.

(1) Show P(X ∈ A) + P(Y ∈ Ac ) = r, where Ac denotes the complement of A.

(2) Show P(X = Y ) ≤ r, where X and Y are not independent of each other.
(3) Show |P(X ∈ A) − P(Y ∈ A)| = |P(X ∈ Ac ) − P(Y ∈ Ac )| = 1 − r.
(4) Show |P(X ∈ B) − P(Y ∈ B)| ≤ 1 − r for any B ∈ B(R).

17
 6
x, y n θ |θ| < 1 θ
2
R
 
1
p(x, y; θ) = C(θ) exp − (x2 + y 2 − 2θxy)
2
N (θ) C(θ) θ (x, y)
2
(X1 , Y1 ), (X2 , Y2 ), ..., (Xn , Yn ) R N (θ)
θ f (θ)
θ
f (θ) =
1 − θ2

 ∞
a>0 exp(−ax2 /2)dx = 2π/a.
−∞

(1) C(θ)
(2) (X1 , Y1 ), (X2 , Y2 ), ..., (Xn , Yn ) θ θ̂
f (θ̂) X1 , X2 , ..., Xn Y1 , Y2 , ..., Yn
(3) (X1 , Y1 ), (X2 , Y2 ), ..., (Xn , Yn ) N (0) f (θ̂)2

18
Problem 6
Let x, y be real numbers, n be a positive integer, and θ be a real number satisfying
|θ| < 1. For a given parameter θ, the probability distribution that has a probability
density function on R2
 
1 2 2
p(x, y; θ) = C(θ) exp − (x + y − 2θxy)
2
is denoted by N (θ). Here C(θ) is a function of θ and does not depend on (x, y). Assume
that (X1 , Y1 ), (X2 , Y2 ), ..., (Xn , Yn ) are R2 -valued random variables that are indepen-
dently distributed from N (θ). A function f (θ) of θ is defined by
θ
f (θ) = .
1 − θ2
Answer the following questions. The following formula can be used without proof.
 ∞
If a > 0, exp(−ax2 /2)dx = 2π/a.
−∞

(1) Find the function C(θ).

(2) For given (X1 , Y1 ), (X2 , Y2 ), ..., (Xn , Yn ), the maximum likelihood estimator of the
parameter θ is denoted by θ̂. Express f (θ̂) using X1 , X2 , ..., Xn and Y1 , Y2 , ..., Yn .
(3) Assume that (X1 , Y1 ), (X2 , Y2 ), ..., (Xn , Yn ) are independently distributed from
N (0). Find the expectation of f (θ̂)2 .

19
 7
Σ Σ2 = {0, 1}

(1) A, B ⊆ Σ∗ A B A\B = {w | w ∈ A w ∈ B}
1
(2) Σ2 C ( 2)

C = {ww | w ∈ Σ∗2 }

(3) Σ2 D

D = {w1 w2 | |w1 | = |w2 | w1 = w2 }

(4) E ⊆ Σ∗ E
F, G ( 2)
X F, G
X E =F ∪G
X F ∩G=∅

1:
5 (Q, Σ, δ, q0 , F )
1. Q ( )
2. Σ ( )
3. δ : Q × Σ → Q
4. q0 ∈ Q
5. F ⊆ Q

2:
L p( ) :
w |w| ≥ p L w
3 w = xyz :
1. i≥0 xy i z ∈ L
2. |y| > 0
3. |xy| ≤ p
yi i y

20
Problem 7
Let Σ be an arbitrary alphabet and Σ2 = {0, 1}. Answer the following questions.
(1) Let A, B ⊆ Σ∗ . Show that if A and B are regular, then A \ B = {w |
w ∈ A and w ∈ B} is also regular by using deterministic finite automata (Note 1).
(2) Show that the language C over Σ2 is not regular by using the pumping lemma
(Note 2).
C = {ww | w ∈ Σ∗2 }

(3) Show that the language D over Σ2 is not regular.

D = {w1 w2 | |w1 | = |w2 | and w1 = w2 }

(4) Let E ⊆ Σ∗ . Show that if E is regular and infinite, then there exist languages F and
G that satisfy the following conditions by using the pumping lemma (Note 2).
X F and G are regular and infinite.
X E = F ∪ G.
X F ∩ G = ∅.

Note 1: Deterministic finite automaton

A deterministic finite automaton is a 5-tuple (Q, Σ, δ, q0 , F ), where
1. Q is a finite set of states,
2. Σ is a finite alphabet,
3. δ : Q × Σ → Q is the transition function,
4. q0 ∈ Q is the start state, and
5. F ⊆ Q is the set of accept states.

Note 2: The pumping lemma

If L is a regular language, then there exists a number p (the pumping length) such that
for any string w in L such that |w| ≥ p, we can divide w into three pieces,
w = xyz, satisfying the following conditions:
1. for each i ≥ 0, xy i z ∈ L,
2. |y| > 0, and
3. |xy| ≤ p,
where y i is the string obtained by concatenating i copies of y.

21
 8
S S (a) (b) NIL
(c) 2 S

CONS(,r) S , r
LEFT(x) x
RIGHT(x) x
CONSP(x) x

CONS(x1 ,CONS(x2 ,. . .,CONS(xn ,NIL). . .)) S

[x1 , x2 , . . . , xn ]

(1) 2 app(x,y) [x1 , . . . , xn ]

[y1 , . . . , ym ] [x1 , . . . , xn , y1 , . . . , ym ] app([1, [2]],[[3, 4], 5]) =
[1, [2], [3, 4], 5]
(2) f(x) S f([[1, 2], [3], [[4], 5]])

def f(x)
if CONSP(x) then
return app(f(LEFT(x)),f(RIGHT(x)))
elsif x == NIL then
return NIL
else
return CONS(x,NIL)
end
end

(3) m(n) m S m(n) sn

f(sn ) CONS

def m(n)
if n == 0 then
return NIL
else
return CONS(m(n-1),0)
end
end

(4) f(x) f(x) n

x O(n) CONS S

22
Problem 8
Answer the following questions about programs that manipulate S-expressions. An S-
expression in this problem is either (a) an integer, (b) NIL or (c) a pair of two S-
expressions. They are manipulated by the following operations.
CONS(,r) returns a pair whose left and right elements are respectively  and r
LEFT(x) returns its left element when x is a pair
RIGHT(x) returns its right element when x is a pair
CONSP(x) returns true when x is a pair, false otherwise
We call S-expressions that are in the form of CONS(x1 ,CONS(x2 ,. . .,CONS(xn ,NIL). . .))
lists, and denote them as [x1 , x2 . . . , xn ].

(1) Define a function app(x,y) that concatenates two lists, where the concate-
nated list of [x1 , . . . , xn ] and [y1 , . . . , ym ] is [x1 , . . . , xn , y1 , . . . , ym ]. For example,
app([1, [2]],[[3, 4], 5]) = [1, [2], [3, 4], 5].
(2) The following f(x) is a function defined on S-expressions. Write the result of
f([[1, 2], [3], [[4], 5]]).

def f(x)
if CONSP(x) then
return app(f(LEFT(x)),f(RIGHT(x)))
elsif x == NIL then
return NIL
else
return CONS(x,NIL)
end
end

(3) The following m(n) is a function definition on non-negative integers. Let sn be the S-
expression created by m(n). Calculate the number of the CONS operations performed
by f(sn ).

def m(n)
if n == 0 then
return NIL
else
return CONS(m(n-1),0)
end
end

(4) Define a function that computes the same result as f(x), yet performs only O(n) CONS
operations for x that contains n integer elements. You may not use data structures
other than S-expressions in the definition.

23
 9

X 213 = 8192
X 25 = 32

X
X

2 C f1 f2

double f1(double A[4096][4096]) { double f2(double A[4096][4096]) {

int i, j; int i, j;
double s = 0.0; double s = 0.0;
for (i = 0; i < 4096; i++) { for (j = 0; j < 4096; j++) {
for (j = 0; j < 4096; j++) { for (i = 0; i < 4096; i++) {
s += A[i][j]; s += A[i][j];
} }
} }
return s; return s;
} }

X double 8
X
X A 32
X i j s A

(1) A[0][0] A
(2) f1 f2
(3) (2) f1 f2

24
(4) A[0][0] A
A f1
f2

(16 )
100000 B[0][0]
100008 B[0][1]
100010 B[0][2]
100018 B[0][3]
100020 B[1][0]
100028 B[1][1]
100030 B[1][2]
100038 B[1][3]
100040 B[2][0]
100048 B[2][1]
100050 B[2][2]
100058 B[2][3]
100060 B[3][0]
100068 B[3][1]
100070 B[3][2]
100078 B[3][3]

double B[4][4]
B 16 100000

25
Problem 9
Consider a computer that has sufficiently large main memory and a single level of cache
memory (cache, hereafter) as follows.

X The cache capacity is 213 = 8192 bytes.

X The size of a cache block (also called a cache line) is 25 = 32 bytes. Note that data
transfer between the main memory and the cache is done in units of cache blocks.
X The direct mapping method is used.
X The cache is a data cache, which is not affected by fetching of executed instructions.

Two functions f1 and f2 below written in C language are compiled and executed on this
computer.
double f1(double A[4096][4096]) { double f2(double A[4096][4096]) {
int i, j; int i, j;
double s = 0.0; double s = 0.0;
for (i = 0; i < 4096; i++) { for (j = 0; j < 4096; j++) {
for (j = 0; j < 4096; j++) { for (i = 0; i < 4096; i++) {
s += A[i][j]; s += A[i][j];
} }
} }
return s; return s;
} }

We assume the followings.

X The size of double data type is 8 bytes.
X Elements in a two dimensional array are stored in the main memory as in the figure
on the next page.
X The start address of the array A is a multiple of 32.
X Local variables i, j, s and the start address of the array A are stored on registers.
Therefore accesses to them do not cause memory accesses.
X When execution of each function is started, the cache is empty (all cache blocks are
invalid).
Answer the following questions.

(1) List the elements of A that are copied to the cache simultaneously, when A[0][0] is
read for the first time.
(2) Calculate cache miss ratios of execution of f1 and f2, respectively.
(Continued on the next page)

26
(3) Explain briefly why cache miss ratios of f1 and f2 are different. The explanation has
to include either of “temporal locality” or “spatial locality”.
(4) When the cache block that includes A[0][0] is on the cache, it will be evicted by sub-
sequent access to another element of A. List such elements for f1 and f2, respectively.
If the cache block is evicted for multiple times, list all the elements.

Address Element
in hexadecimal
100000 B[0][0]
100008 B[0][1]
100010 B[0][2]
100018 B[0][3]
100020 B[1][0]
100028 B[1][1]
100030 B[1][2]
100038 B[1][3]
100040 B[2][0]
100048 B[2][1]
100050 B[2][2]
100058 B[2][3]
100060 B[3][0]
100068 B[3][1]
100070 B[3][2]
100078 B[3][3]

Memory layout of elements of a two-dimensional array declared by double B[4][4].

Here the start address of array B is 100000 in hexadecimal.

27