Computer Science GRE subject exam

A practice exam can be found here:

http://www.ets.org/Media/Tests/GRE/pdf/CompSci.pdf

Instructions:
1. Don't study! Take the exam immediately (or whenever possible).
2. Report your I-took-the-practice-CompSci-GRE-without-studying score on here :)

Me:

Correct: 38
Incorrect: 32
Scaled score: 740

I should have skipped a lot of the questions I knew I didn't know, but I wanted to give an honest effort to each question...so I answered all of them. To be fair, if I had skipped questions that I knowingly didn't know (does that make sense?), I could have pushed myself up in the rankings a bit. Still...not bad for not studying? It kicked my trash...

Do you still go to BYU?

Not sure how you got the answers to the five ** questions. I also answered the ones I would have normally skipped (I know nothing about grammars) just for fun, and apparently I started to fall apart around question 50. Anyway +47 -18 which apparently scales to 820. Not bad for a stinking mathematician. [If you let me skip, it ends up being +44 -9 =12, which gets me the exact same 820. Hooray guessing!]

Ooooh yeah there were 5 questions not graded. Oops I need to go back and rework my score, haha.

ethic: no, i graduated about a year ago. my actual reason for posting this is because i'm looking at grad schools, and i need to get back in the books and study in order to take the GRE.

I got too bored with it after 2 questions, but I got both of them right. So I guess my score was 100% :P.
Yes, the first two *were* painfully easy.

Sigh, CS! Glad I chose Software Engineering :)

Honestly, I don't know why they keep mentioning bubble sort in CS. Seriously. Many of those questions are painfully obvious, many are mindless regurgitation, and some are challenging!

Downloaded. Should be fun to see what a non graduate can do with it.
But I'll need some free time. I'll try this later.

Dude that thing is like 50 pages... I'll just assume I'm somewhere around infinity.

Quote:

---

Originally Posted by EVOEx

I got too bored with it after 2 questions, but I got both of them right. So I guess my score was 100% :P.
Yes, the first two *were* painfully easy.

---

Quote:

---

Originally Posted by zacs7

Sigh, CS! Glad I chose Software Engineering :)

Honestly, I don't know why they keep mentioning bubble sort in CS. Seriously. Many of those questions are painfully obvious, many are mindless regurgitation, and some are challenging!

---

As I took the exam I noticed that the questions get progressively harder. The first 35 questions were definitely much easier for me than the last 35 questions. I don't know if it's a function of me getting tired because of the length of the exam, or if the last 35 questions were legitimately more challenging.

Regardless, there were some concepts on the exam that were never covered in a single one of my CS classes I took (such as the intricate details of the TCP/IP stack. Sure...there was an elective course I could have taken to learn those things, but it wasn't part of the core curriculum). There were other concepts that I learned, but have since forgotten about or need to brush up on because it's been a long time (Turing machines, finite automata, languages/grammars/alphabets/stuff i don't really care about). Finally, stuff I actually care about was rarely covered :) (AI, machine learning, etc).

I got lucky with some of the questions: some of them asked for details about caching, pipelining, and RISC instruction sets. Although several courses go over caching, only one course at my university really went in depth about pipelining and RISC instruction sets, and it's not a required course...luckily I took the course :)

Quote:

---

Originally Posted by DavidP

ethic: no, i graduated about a year ago. my actual reason for posting this is because i'm looking at grad schools, and i need to get back in the books and study in order to take the GRE.

---

You should come to the U. We're a better school than BYU. I'm
pretty popular on campus, too. I can hook you up with some
friends.

Quote:

---

Originally Posted by ethic

You should come to the U. We're a better school than BYU. I'm
pretty popular on campus, too. I can hook you up with some
friends.

---

The U is actually in my list of options at the current time...but then again, I have a very large list of schools I am looking at and trying to pair it down a bit. We'll see :)

Ugh!

24 right
8 wrong
38 Unanswered
---------
Raw Score: 22
Scaled Score: 680
---------

There goes my chance at being admitted this year.

Fun test. Learned a lot about what I don't know. And some questions were actually quite interesting. One issue though: How on earth does the loop terminate in question 15?

Quote:

---

Originally Posted by Mario F.

Ugh!

24 right
8 wrong
38 Unanswered
---------
Raw Score: 22
Scaled Score: 680
---------

There goes my chance at being admitted this year.

Fun test. Learned a lot about what I don't know. And some questions were actually quite interesting. One issue though: How on earth does the loop terminate in question 15?

---

I guessed on question 15 :) And I think I got it wrong...

Two things I had never even heard of before taking this exam: datagrams (which apparently after reading on wikipedia are just equivalent to packets) and loop invariants (which I still don't understand even after reading wikipedia).

Well... I just wasted 2 hours of my employer's time on that. I'm surprised by both how much I remember and how much I forgot in the last 10-15 years. I got tired towards the second half of it as well (and it didn't help that my wife was skyping me for the last hour distracting me with questions about our daughters' birthday presents). Still pretty interesting to see a lot of those concepts again.

Yeah. That article is a little to the insane.

Think of a the loop invariant as any condition that must remain true on entry and during all iterations of the loops. I find it easier to think of a loop knowing the invariant than finding the invariant in a given loop.

Say you need to loop through all positive integers less than 100 that are even. One loop invariant is x % 2 = 0.
It's not very clear to me if a loop can have more than one invariant, of if the invariant is the full logical statement. In that last case, it would need to be x < 100 && x % 2 = 0

Quote:

---

How on earth does the loop terminate in question 15?

---

k = floor(k / 2), when repeated, will cause k to become 0, thus terminating the loop. (Unless of course k is negative, but you can rule this out due to the explicitly stated invariant k >= 0)

Question 15 is very simple if you just ignore all the unnecessary information. All you need to know is:

Code:

---

Invariant:  z*x^k = b^n  and  k >= 0
...
while (k != 0)
...

---

Clearly the only way this loop terminates is when k is zero. Substitute 0 for k in the invariant, and you get the answer.

Code:

---

z*x^0 = b^n
z*1 = b^n
z = b^n

---

Also:

Quote:

---

Think of a the loop invariant as any condition that must remain true on entry and during all iterations of the loops.

---

Using the terminology from question 14 of this practice GRE: "A loop invariant for a while statement is an assertion that is true each time the guard is evaluated during the execution of the while statement." The invariant does NOT have to be true at any arbitrary point in the body of the while loop, only before the body begins and after the body ends. I'm pretty sure you already know this, but I felt that statement needed some more exactificationness.

[edit]
Also, a loop can certainly have more than one invariant. In fact, I believe all loops have an infinite number of invariants.
[/edit]

Quote:

---

Originally Posted by arpsmack

k = floor(k / 2), when repeated, will cause k to become 0, thus terminating the loop. (Unless of course k is negative, but you can rule this out due to the explicitly stated invariant k >= 0)

---

Hmm... the thought occurred to me then that K would have to be a floating point variable. That's how I eventually ended up interpreting the pseudo code.

However its a well known fact that no division operation results in zero except when the dividend is 0 itself. So this means the comparison never turns out to be 0 in most languages we know and work with. It's probably safe to assume the pseudo-code means at first iteration where K < 1. But why not just write that and avoid the confusion?

Quote:

---

[edit]
Also, a loop can certainly have more than one invariant. In fact, I believe all loops have an infinite number of invariants.
[/edit]

---

Excellent. That was always a point of some confusion to me. Thanks.

Quote:

---

Originally Posted by arpsmack

Clearly the only way this loop terminates is when k is zero. Substitute 0 for k in the invariant, and you get the answer.

---

Yes. Clearly. :) I always hated it when my professors used that on us :)

Quote:

---

Originally Posted by Mario F.

Hmm... the thought occurred to me then that K would have to be a floating point variable. That's how I eventually ended up interpreting the pseudo code.

However its a well known fact that no division operation results in zero except when the dividend is 0 itself. So this means the comparison never turns out to be 0 in most languages we know and work with. It's probably safe to assume the pseudo-code means at first iteration where K < 1. But why not just write that and avoid the confusion?



Excellent. That was always a point of some confusion to me. Thanks.

---

m/n is 0 whenever m < n (well, it's 0 remainder m, but you didn't ask for the remainder part).

Quote:

---

Originally Posted by DavidP

Yes. Clearly. :) I always hated it when my professors used that on us :)

---

Heh good point. Stupid unnecessary fluff words. And even worse is that in this particular instance I'm pretty sure my subconscious added "clearly" to give off that "OH MY GOD I'M SO SMART LOOK AT THE SIZE OF MY E-PENIS" vibe. I'm such a dick.

My personal pet peeve in that area is "note". Good god do some people overuse it.

Note that a > b in this situation.
Note that integers are numbers.
Note: errors are bad.
Note that I have a lot of notes.
Note that I'm a douche.

FFFFFFFFUUUUUUUUUUU!!!!!!

Quote:

---

Originally Posted by tabstop

m/n is 0 whenever m < n (well, it's 0 remainder m, but you didn't ask for the remainder part).

---

Yeah. Never occurred to me that was an acceptable practice in pseudo-code.

Quote:

---

Originally Posted by DavidP

I should have skipped a lot of the questions I knew I didn't know, but I wanted to give an honest effort to each question...so I answered all of them. To be fair, if I had skipped questions that I knowingly didn't know (does that make sense?), I could have pushed myself up in the rankings a bit.

---

Quote:

---

Originally Posted by tabstop

[If you let me skip, it ends up being +44 -9 =12, which gets me the exact same 820. Hooray guessing!]

---

The test is designed so that random guessing gives you 0 points on average. You should try to answer all questions because you can often rule out at least one option and then guessing will give you a higher score on average.