Note: This

∗ semester was the

Klausur Braindump first time Swarm
Intelligence —
as compared to

Swarm Intelligence
Organic Computing
— was held, yet
there was no time
to discuss the last
Various contributors two chapters (P2P,
CAN). Therefore
this Braindump
Summer Semester ’21 might not
representative.
be

1 Introduction 11 Points

~
a) (3 Points) Recite three reasons given in the lecture as to why systems are becoming more
complex.

v b) (3 Points) Give three self-∗-properties.

- c) (3 Points) Define “emergence”. There might have

been more here
⑳
d) (2 Points) Which research question and algorithm were related to emergence.
Hint: Swarm of webpages

2 Particle Swarm Optimisation 13 Points

~
a) (1 Point) What properties does the objective function have?

i b) (2 Points) Give and explain two similarly but opposing goals.

-
c) (2 Points) List all the attributes of a particle and a swarm. Alternatively
question is what
the

·
d) (4 Points) Draw a particle with its components, and - how it changes in the next time step. new attributes arise
from a particle
e) (4 Points)
② Give the exact formula for operating on a particle. Explain each parameter and being part of a
type (random, dimensions, …). swarm

-
3 Convergence of PSO 9 Points

·
a) (2 Points) When does PSO converge?
Hint: You only need reference two variables

-
b) Let the Matrix A be
A=
!
1−b a
"
,
−b a

~
where the variables a and b may have any value.
i) (4 Points) Compute the eigenvalues of the matrix A.
-
ii) (2 Points) For which values of λ 1 and λ2 (λ1 ≥ λ2 ) does the swarm converge.
iii) (1 Point) Assuming λ is real-valued and λ is a non-real complex number, and the
⑳ 1 2
condition from ii) is given, sketch a diagram of how the simulation would converge.
This problem was
probably broken, as
this combination

triangle
of Eigenvalues can
only occur with
I a complex-valued
matrix.
∗ Asalways: The document is provided “as is”, without warranty of any kind, express or implied, including
but not limited to the warranties of merchantability, fitness for a particular purpose and noninfringement.
Please report issues or improvements via GitLab: https://gitlab.cs.fau.de/oj14ozun/si-ss21.

1
 4 HITS 13 Points

V a) (1.5 Points) Give three kinds of search queries.

b) (1.5 Points) What does “HITS” stand for?

-
c) (2 Points) What do the variables x (p)
and y (p) designate?

-
d) (6 Points) Give the formula for The points were
shared among d)
x(p) = and e).

and

y (p) =

Hint: Use a → b to represent edges

Y e) Mark the right variables being computed:

q1 q1

q2 p p q2

q3 q3

- ! x(p) ! x(p)
! y -! y
(p) (p)

~ f) (2 Points) HITS uses the matrices AA⊤ and A⊤ A. Give two important mathematical
properties these matrices must have.

-
5 Evolutionary Algorithms 16 Points

~ a) (6 Points) Draw the evolutionary cycle for an (µ, λ)-EA algorithm, with the objective function
F.

- b) (2 Points) Let π = (7, 5, 3, 1, 2, 8, 4, 6) be an unsorted sequence. Compute INV(π) and

HAM(π).
π might have al-
ternatively ended

-
in . . . , 8, 2, 4, 6) or
c) (2 Points) Assume π has been sorted. Give the values of INV(π) and HAM(π) that must . . . , 2, 8, 4, 6).
hold in that case.

-
d) (3 Points) Given the following genotypes for the Travelling Salesperson Problem,
1 2 3 4 5 6 7 8

1 2 7 6 5 4 3 8

-
draw the phenotypes before and after applying the inversion operator.

e) (3 Points) Apply the one-point crossover at the indicated line: Parent 1 has a differ-
ent value.

Parent 2 1 2 7 6 5 4 3 8

Parent 1 1 2 3 4 5 6 7 8

2
6 Ant Colony Optimisation 11 Points

ACO is being used to solve TSP on the Graph

1
6 2

5 3
4

where the partial route of the ant a is sa = (1, 3, 2, 5).

It is assumed that the distance between every node is 1. The evaporation rate is ρ = 0.5.
For every pheromone value τij = τji holds.

v a) (1 Point) Give the set of vertices ant a visiting the node 5 is allowed to consider next
N (sa ) =
#

- b) (6 Points) Compute the probabilities that ant a visiting the node 5 will choose these vertexes.
Assume the pheromone values are τ45 = τ25 = 3 and τ56 = τ15 = τ35 = 2.

-
• p({1, 5} | s ) probability that is the ant a will move from node 5 to 1. ②
a

- • p({5, 6} | sa ) probability that is the ant a will move from node 5 to 6.

-
Lor2
• p({4, 5} | s ) probability that is the ant a will move from node 5 to 4.
a

· c) (2 Points) Compute the difference g56 that choosing the edge 5 → 6 would make, using “Ant-
Quantity” function and taking Q to be 2.
>

v d) (2 Points) Compute the updated pheromone values for τ56 and τ45 , assuming there is only
one ant a that chose the edge 5 → 6 in this turn. -
-
-
--

d
236(tt) =

(1
-

9)2st) + 0356 :

( 0s)
+
=>
x2 + 3
&
:
.

/ ++1
2
us
=
(9) Zug (t) + D2us .

+ O
8 3x
= .

I 1 . 5
E
3
24s =
[23 = 3
, 296"2, 5 Egg :
2
.