Machine probability

Out of four machines, Titus knows that exactly two are faulty. He wants to test them one by one in a random order (without replacement) until he can identify the 2 faulty machines. Find the probability that exactly two tests are needed?

1/3 2/3 1/2 1/6

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

9 solutions

Venture Hi
May 4, 2015

1/3 is the right answer. The number of ways to pick the machines is 6 (g=good, b=bad): {(ggbb), (gbbg), (gbgb), (bbgg),(bgbg), (bggb)}

To find the good machines in two tries, you must pick (ggbb) or (bbgg) . So two out of 6 is 1/3.

yes, that's right. :)

Keshav Tiwari - 6 years, 1 month ago

I only count bb chance without gg ._.

Yoga Nugraha - 5 years, 5 months ago

This is the best answer.

Ananya Aaniya - 4 years, 11 months ago

I thought of bbgg and ggbb, picking the first g is 2/4 and the second g 1/3, multiply the two probabilities =2/12 or 1/6, add the same for the bbgg 1/6+1/6=1/3. Then I tapped the wrong one . Lol

David S - 4 years, 5 months ago

but he is choosing at random not in order

D Chauhan - 3 years, 7 months ago

Log in to reply

(g=good, b=bad): {(ggbb), (gbbg), (gbgb), (bbgg),(bgbg), (bggb)}

This is all of possible events that the tests could have turned out. So imagine the first test reveals a good machine, the second test also reveals a good machine. You are done and only needed to do 2 tests.

Now, on the other hand, imagine if the first test show a bad machine and the second test shows a bad machine as well, you know the remaining 2 machines are good. You are also done with only 2 tests.

So out of a possible 6 outcome above, there are 2 cases where you only need to do 2 tests and be done, so 2/6= 1/3.

Omi Slash - 3 years, 6 months ago

1/3 is correct for sequential selection, but he has to test cars randomly, so, the probability of 1st car to be good is 1/4, and the probability of 2nd randomly selected car to be good is 1/3, which results in 1/12 for good car, same for bad car, the correct answer for random selection is 1/12 + 1/12 = 1/6

carslover08 . - 3 years ago

I thought that ggbb isn't a valid sequence cause wouldn't he have to do the test 4 times (+2 times also to know that the first two are good machines).

Raivat Shah - 2 years, 7 months ago
Uros Stojkovic
Dec 22, 2016

He knows that 2 machines are incorrect, and he can find out which two if he tests either two faulty ones or two good ones (if he test two and it turns out they are good, he will automatically know that the other two are faulty). The probability that he will test two faulty machines is 1/2 * 1/3 = 1/6 . The same is the probability of testing two good ones, hence the probability that exactly two tests are needed is 1/6 + 1/6 = 1/3 .

Another intuitive way of thinking about this: Based on the understanding that both tests need to be good or faulty, it doesn't matter what happens in the first test. All that we need is for the second machines test to match the first. If the first test was faulty, we know that of the remaining 3, there is 1 that is left that's also faulty, so there's a 1/3 chance of choosing that for the second test. The same holds true if the first test happened to be good, with a 1/3 chance of picking the right machine.

Sourabh Das - 3 years, 11 months ago

Log in to reply

Great observation! There are numerous ways to solve problem like this one and many of them don't require some special math skills, just logical reasoning.

Uros Stojkovic - 3 years, 11 months ago
Adarsh Kumar
Apr 23, 2015

Let us say that the two faulty machines are, F 1 F_1 and F 2 F_2 and the two working ones are, W 1 W_1 and W 2 W_2 .Now,we see that we can identify the faulty ones in these two ways: 1 ) 1) The two faulty ones are the first two to get checked;and 2 ) 2) The two working ones are the first two to get checked.Thus,for 1 ) 1) we have 4 4 cases in total(two ways of arranging the faulty ones and two ways of arranging the working ones) and similarly for 2 ) 2) we have four cases.Hence total number of favourable cases is equal to 8 8 ,and the total number of ways of arranging the four machines is 4 × 3 × 2 × 1 = 24 4\times3\times2\times1=24 . P ( E ) = 8 24 = 1 3 \Longrightarrow P(E)=\dfrac{8}{24}=\dfrac{1}{3} .

David B
Jun 2, 2018

The result of the first test doesn't matter. If he gets the same result in the second test, he knows which ones work. The probability for this is obviously 1/3 cause after the first test 3 machines are left from which 1 is in the same condition as the machine from the first test.

Joshua Lowrance
Apr 16, 2018

When there are exactly two tests, that means that the first machine tested was faulty, but the second one works. The probability of finding a faulty machine out of the four machines is 1/2. The probability of finding a working machine on the second try out of the remaining three machines is 2/3. (1/2) * (2/3) = 1/3

Lovro Cupic
Jan 18, 2018

First test will identify either valid or invalid machine. Since we can determine 2 valid machines in 2 trials only if we are lucky enough to pick 2 valid or 2 invalid machines in first 2 trials (because there are 4 machines in total and those which are not valid are invalid), it is enough for test to show the same validity of the machines, i.e. the second machine should have the same test result as the first machine. Without loss of generality, we can assume that the first machine is tested valid - which means we're left with 3 machines, 2 invalid and 1 valid, and the question reduces to probability of picking a valid machine between 3 machines, p= 1 3 \frac{1}{3} .

Anek Doteles
May 29, 2018

A probability tree shows that there are only two ways in which he will only need two tries:

  • take two times one of the the faulty ( 2 4 \frac{2}{4} x 1 3 \frac{1}{3} )

  • take two times one of the perfectly fine ( 2 4 \frac{2}{4} x 1 3 \frac{1}{3} )

=> Solution is 2x(2/12) = 1/3

That's because if he would have chosen one faulty and than one fine machine (+the other way around) he has left one faulty and one fine and has therefore to do another try with probability 1/2.

Gustav Elmqvist
Dec 30, 2017

In total there are ( 4 2 ) {4 \choose 2} ways to test two out of four machines. We can identify the faulty machines by either checking the two faulty ones or the two working ones. This gives us 2 favourable cases out of the ( 4 2 ) {4 \choose 2} in total.

This means that the probability of identifying the two faulty machines is 2 ( 4 2 ) \frac {2}{4 \choose 2} = 2 6 \frac {2}{6} = 1 3 \frac {1}{3} .

Rajen Kapur
Apr 22, 2015

As both the faulty machines are found in the first two trials, the answer is (2/4)(1/3) = 1/6.

Moderator note:

This solution is incorrect. See Keshav's comment below.

I do not agree sir . What's wrong in this :There are two ways to identify the faulty machines in exactly two tests. Either we identify the two faulty machines or the two working machines. The probability in the former case is 1 2 1 3 = 1 6 \frac{1} {2} * \frac{1}{3}=\frac{1}{6} and that in the latter case is also 1 2 1 3 = 1 6 \frac{1} {2} * \frac{1}{3}=\frac{1}{6} . Hence the required probability is : 1 6 + 1 6 = 1 3 \frac{1} {6} +\frac{1}{6}=\frac{1}{3}

Keshav Tiwari - 6 years, 1 month ago

Log in to reply

Yes, that makes sense; I hadn't thought of the option of identifying the two faulty machines by way of first identifying the two working machines.

Brian Charlesworth - 6 years, 1 month ago

Log in to reply

Sir then please change the answer . Calvin sir changed it in response to the report by @Jon Haussmann . Thankyou.

Keshav Tiwari - 6 years, 1 month ago

Ah, ok. Tricky problem. I concede the answer is 1/3.

Jon Haussmann - 6 years, 1 month ago

Log in to reply

Thankyou sir. :)

Keshav Tiwari - 6 years, 1 month ago

Thanks. I did not realize that we could identify it the second way. Definitely worth a reshare.

I have updated the answer to 1/3.

Calvin Lin Staff - 6 years, 1 month ago

Log in to reply

Thnkyou sir. :)

Keshav Tiwari - 6 years, 1 month ago

This is an old and a beautiful JEE problem.

Prakhar Gupta - 6 years, 1 month ago

Log in to reply

^Yes. Most of the books have the answer as 1/6 . :)

Keshav Tiwari - 6 years, 1 month ago

I had actually thought of posting this problem yesterday. It is definitely worth it.

Raghav Vaidyanathan - 6 years, 1 month ago

The way this problem is worded, it is unclear what conditions are needed to satisfy the "2 faulty machines" probability.

The tester stops once he identifies 2 faulty machines. However, it does not indicate HE KNOWS there are exactly only 2 faulty machines.

Therefore, in the cases where he identifies two working machines as his first two tests, he will not stop testing because the last two machines might be working as well and the problem's conditions have not been met according to the tester.

I'm not trying to nitpick, and the question can be modified with a simple wording change. :)

Rick Bartus - 5 years, 1 month ago

Log in to reply

Yes that makes sense. Thanks, I have edited the problem.

Keshav Tiwari - 5 years, 1 month ago

Thanks. I have edited the problem for clarity.

Calvin Lin Staff - 5 years, 1 month ago

Well played. I failed on this one :-)

Freddie Kalaitzis - 6 years, 1 month ago

Good sum indeed. I fell for the trap. Your solution is upvoted.

Pranav Rao - 5 years, 4 months ago

There is a better approach.. u can check Venture Hi 's solution above..

Ananya Aaniya - 4 years, 11 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...