System Modelling 2019/2 -- Theoretical Questions 3
Example theoretical questions for practising for the 2nd mid-term exam in System Modelling. (In the mid-term exam, you get +1/4 point for each correct answer, -1/4 point for each incorrect ones, and 0 point for each question left out. You have to reach 50% of the possible points in this system. The Google Form is not able to implement this points system.)
1. Summarizing 1000 measured data points…
1/a … the mode can be less than the 0.1 quantile. *
1 point
1/b … the median can be less than the 0.1 quantile. *
1 point
1/c … the average can be less than the 0.1 quantile. *
1 point
1/d … the first quartile can be less than the 0.1 quantile. *
1 point
2. In case of visual analysis…
2/a … we can use parallel coordinate diagrams for examining the connection between two numerical variables. *
1 point
2/b … if the shape of the boxplots of two variables are exactly the same, then the values of the variables are in a linear relationship. *
1 point
2/c … the number of the occurrences of the concrete values can be explicitely read from a boxplot. *
1 point
2/d … the same information can be read from every histogram of the same continuous variable. *
1 point
3. If the throughput of the system increases when changing its structure or load, …
3/a … while the throughput of a given activity remains the same, then the visitation number of the activity decreases. *
1 point
3/b … but its maximum throughput remains the same, then the utilization decreases. *
1 point
3/c … then (assuming a stable state of the system) the arrival rate increases too. *
1 point
3/d … then, assuming the same response time, the number of requests being under service decreases. *
1 point
4. When designing/dimensioning the performance (capacity) of a system …
4/a … the expected arrival rate in the future can be read from the benchmarks. *
1 point
4/b … the different expected performances for different loads can be found out through simulation. *
1 point
4/c … we can use the Law of Zipf for cache dimensioning, as it can give us the load for serving the most frequent (few) requests. *
1 point
4/d … we can use the Law of Zipf, because it assumes a linear relationship between the response time and the utilization. *
1 point
6. Well-structured process models …
6/a … are always deterministic, as they cannot contain decision-merge structures. *
1 point
6/b … close every fork symmetrically with a corresponding join, through which the branches of the process never have to wait at different joins, stuck in a deadlock. *
1 point
6/c … are always fully specified, thanks to always using well-structured blocks for building the model. *
1 point
6/d … can only contain loops that have only one exit branch. *
1 point
7. Requirements …
7/a … are the expectations for a system to be implemented. *
1 point
7/b … can sometimes be interpreted in several ways, when given informally. *
1 point
7/c … are non-functional statements in their nature, while functional cirteria can only be formulated in lower level specifications. *
1 point
7/d … can be specified iteratively, based on experience in modelling and development. *
1 point
8. During testing, oracle and reference …
8/a … are similar in the sense, that both are used for checking the output of the system being tested. *
1 point
8/c …are connected in the sense, that an oracle can always be prepared from the reference with the same semantics. E.g. that oracle could simply compare the correspondence with the reference. *
1 point
8/d … are different in the sense, that the reference verifies the behaviour without a test input, while the oracle must always have a specified input, too. *
1 point
9. Test coverage …
9/a … is a value between 0 and 1. *
1 point
9/b … is measured with the goal to get an estimation of the quality and completeness of the test suite. *
1 point
9/c … can be increased through making the oracle more permissive. *
1 point
9/d … can be increased through taking additional test cases. *
1 point
10. Model checking …
10/a … has the goal to prove the fulfillment of the specification with mathematical precision. *
1 point
10/b … detects deviation from the specification through observing the system in runtime. *
1 point
10/c … is a kind of statical analysis, as (e.g. compared to testing) there is no need to execute or simulate the model. *
1 point
10/d … analyzes every possible behaviour of a model exhaustively. *
1 point
This content is neither created nor endorsed by Google. Report Abuse - Terms of Service