Presentation is loading. Please wait.

Presentation is loading. Please wait.

School of Engineering and Computing Sciences

Similar presentations


Presentation on theme: "School of Engineering and Computing Sciences"— Presentation transcript:

1 School of Engineering and Computing Sciences
Studies in Computer Science? Dr. George Mertzios School of Engineering and Computing Sciences Durham University, UK 1

2 Main Questions What is Computer Science (CS) ?
What is Durham University ? Why should I study CS at Durham ? 2

3 What is Computer Science?
users engineers / administrators designers 3

4 What is Computer Science?
Design of architectures for PCs / laptops, micro-processors, mobile devices, storage media (hard disks, CDs), … Verify the correctness of the design are all computations provably correct? Design efficient processes (algorithms) for specific computational tasks must be both fast and correct ! 4

5 Challenges in algorithm design
From very practical applications programming (writing code in Java, C++, …) software engineering To highly theoretical (but applicable!) tasks how to schedule cargo ships to minimize costs? how to find quickly the best route from street A to street B ? 5

6 Challenges in algorithm testing
1993: Intel introduced processor Pentium P5 1994: A mathematics professor (Thomas Nicely) detects some incorrect computations of P5 the design was not properly tested ! Customers stopped trusting Intel 1995: Intel announced $475 million loss defective processors became key rings 6

7 Challenges in algorithm testing
7

8 How does… a CD player cope with scratches? a .zip file compress data?
a space probe send data across millions of miles? online banking keep your money safe? They are all coding problems 8

9 1971 NASA Mariner 9: Error correction
A photo pixel was 6 bits long (64 grey shades) To avoid data loss: data repetition is necessary what if we lose 1 bit ? Power constraints: max. only about 30 bits to represent one pixel 5 repetition code: easy to implement corrects 2 errors Each photo-receptor in the camera measures the brightness of a section of the Martian surface about 4-5 km square, and outputs a grayness value in the range This value is represented as a binary 6-tuple.  The TV image is thus digitalized by the photo-receptor bank and is output as a stream of thousands of binary 6-tuples. Without coding and a failure probability p = 0.05, 26% of the image would be in error ... unacceptably poor quality for the nature of the mission. Any coding will increase the length of the transmitted message. Due to power constraints on board the probe and equipment constraints at the receiving stations on Earth, the coded message could not be much more than 5 times as long as the data. Thus, a 6-tuple of data could be coded as a codeword of about 30 bits in length. A second concern involves the coding procedure. Storage of data requires shielding of the storage media – this is dead weight aboard the probe and economics require that there be little dead weight. Coding should therefore be done “on the fly”, without permanent memory requirements.  Finally, decoding needs to be done rapidly. The Jet Propulsion Laboratory in Pasadena, California will process the signals and reconvert them to picture images for the press which will be gathered at JPL. Besides this NASA priority, rapid decoding is needed so that feedback to the probe becomes viable – redirecting the camera based on what is seen. The 5-repeat code would satisfy the mission specs, but it is only 2-error correcting ... leaving 1% of the image in error. The actual code selected is 7-error correcting and this reduced the probability of error in the image to only 0.01%. Matrix of all 64 codewords represented here. The Hadamard code. If the probe wants to send the pixel = 26 in binary, it will transmit Row 26 of the matrix. This is a 32-bit string (where black = 0 and white = 1). This string might get some errors when it travels through outer space. However, provided 7 or fewer errors occur, the receiver on Earth can still recover the number 26 and the pixel Basic example of code: 3-repetition code. If we want to transmit the pixel a, we actually send the same pixel three times in a row: aaa. This corrects only one error. 9

10 1971 NASA Mariner 9: Error correction
Instead this code was used: If you want to transmit the number = 26 in binary: transmit instead the 32 bits of the 26th line of the code ! it corrects 7 errors! Each photo-receptor in the camera measures the brightness of a section of the Martian surface about 4-5 km square, and outputs a grayness value in the range This value is represented as a binary 6-tuple.  The TV image is thus digitalized by the photo-receptor bank and is output as a stream of thousands of binary 6-tuples. Without coding and a failure probability p = 0.05, 26% of the image would be in error ... unacceptably poor quality for the nature of the mission. Any coding will increase the length of the transmitted message. Due to power constraints on board the probe and equipment constraints at the receiving stations on Earth, the coded message could not be much more than 5 times as long as the data. Thus, a 6-tuple of data could be coded as a codeword of about 30 bits in length. A second concern involves the coding procedure. Storage of data requires shielding of the storage media – this is dead weight aboard the probe and economics require that there be little dead weight. Coding should therefore be done “on the fly”, without permanent memory requirements.  Finally, decoding needs to be done rapidly. The Jet Propulsion Laboratory in Pasadena, California will process the signals and reconvert them to picture images for the press which will be gathered at JPL. Besides this NASA priority, rapid decoding is needed so that feedback to the probe becomes viable – redirecting the camera based on what is seen. The 5-repeat code would satisfy the mission specs, but it is only 2-error correcting ... leaving 1% of the image in error. The actual code selected is 7-error correcting and this reduced the probability of error in the image to only 0.01%. Matrix of all 64 codewords represented here. The Hadamard code. If the probe wants to send the pixel = 26 in binary, it will transmit Row 26 of the matrix. This is a 32-bit string (where black = 0 and white = 1). This string might get some errors when it travels through outer space. However, provided 7 or fewer errors occur, the receiver on Earth can still recover the number 26 and the pixel Basic example of code: 3-repetition code. If we want to transmit the pixel a, we actually send the same pixel three times in a row: aaa. This corrects only one error. 10

11 What is Durham University?
3rd oldest university in England (1832) after Oxford and Cambridge Top ranked university: 5th in The Complete University Guide 2014 6th in The Times & The Sunday Times Good University Guide 2014 World Top 25 for employer reputation: QS World Rankings 2013 regarding the quality & international demand for its graduates 11

12 What is Durham University?
A unique College system: 16 Colleges every student is a member of a College accommodation, sports, socializing & more strongly connected communities, offering a great network after graduation 12

13 What is Durham University?
13

14 What about Durham city? A beautiful ancient town in the North-East: 14

15 What about Durham city? A beautiful ancient town in the North-East:
13 minutes to Newcastle 2 hours to Edinburgh 3 hours to London Mostly a student city: more than students from 150 countries a lot of student cafés / pubs / party places One of the most safe cities in England Very cheap housing (and beer): estimated £ annual living costs (see here) 15

16 Source: DLHE survey of leavers
Why CS at Durham ? Durham CS graduates get great jobs: Average salary for Durham CS graduates is higher than (almost) all other subjects Source: DLHE survey of leavers 16

17 Source: Complete University Guide 2014
Why CS at Durham ? Computer Science Graduate Prospects: Source: Complete University Guide 2014 17

18 Job examples From: 18

19 Job examples From: 19

20 Job examples From: 20

21 Hawk-Eye: a story of success
used officially in many sports (cricket, tennis, …) Inventor: Paul Hawkins CS Durham graduate (2001) 21

22 Durham CS graduates get great jobs
Companies approach us asking for our graduates 22

23 Durham CS graduates get great jobs
We have long industrial relationships offering excellent internship opportunities Performance through technology 23

24 Durham CS graduates get great jobs
DLHE survey of leavers 24

25 Durham CS graduates get great jobs
DLHE survey of leavers 25

26 Durham CS graduates get great jobs
Raindrop Information Systems DLHE survey of leavers 26


Download ppt "School of Engineering and Computing Sciences"

Similar presentations


Ads by Google