2022 23 Assignment
2022 23 Assignment
2022 23 Assignment
dx
! = 4x + 2y
dt
dy
! =−x +y
dt
a. Determine the matrix A needed to cast the equations above in matrix form (2 mark):
dx
! = A x , where ! x = (x, y) and d! x /dt = (d x /dt, d y/dt)
dt
b. Determine the eigenvalues of A numerically. Determine the appropriate eigenvectors
also numerically. Check on paper whether your python programme gives the correct
results (add a scan of your calculation to your Jupyter notebook). (3 marks)
c. Solve the linear differential equation system. You may chose to do it on paper or
numerically. If on paper please add a scan of it to your Jupyter notebook. (2 marks)
2) You are given a dataset (see attached excel sheet). These are data from a calibration
run of a temperature sensor connected to a Raspberry Pie. You can find the output
values as read with the Raspberry Pie at a given temperature in the excel sheet. You
do not have the datasheet at hand and you have to test against various models:
a. Linear model (!y = a x + b)
b. Polynomial of the second order (!y = a x 2 + bx + c)
c. Polynomial of the third order (!y = a x 3 + bx 2 + cx + d)
Solve this least-square problem numerically. Indicate your solutions for ! x = (a, b, c, d ).
Argue which of these models you would use for this sensor. Include in your reasoning
also some range discussion and reliability arguments. (5 marks)
3 2
[ ]
A
! = 2 3 .
2 −2
4) Use any none-quadratic picture of dimension larger than 400x400 and apply Singular
Value Decomposition to find an approximation to the original image.
a. Investigate the dependency of the reconstructed image on the number (r) of
singular values that you use for reconstruction. Choose at least 5 values of r
for that investigation. Also produce an error image (difference between original and
from approximated matrix reconstructed image). Indicate the compression ratio for
each r. (10 marks)
b. Which r would you use so that you can still perform face recognition on the
reconstructed image? (2 marks)
c. Which r would you chose if you had to recognize dark and bright spot in the image
only? (2 marks)
d. Which r would you chose if you had to recognize forms (like cars, people, etc.) (2
marks)
e. Can you think of other methods to do image compression without losing too much
information? (2 marks)
Please write a very short report with your answers to the assignments and do not forget to
show the images and argue for your choices of r. Please also check what is meant by
compression ratio in the literature versus what will be calculated by the python library.
Assessment criteria. You will only get full marks if it is clear from your programme/text how
you got to your results. I am looking for statements that are well argumented for.
9
!Grade = 1 + (#marks )
34