# Computer Hardware & Software interview questions

## Interview Questions

An array of 99 elements contains integers from 1 to 100 with one missing element. Find the missing element. 22 Answers100 1. calculate the sum of elements in array say SUM 2. sum of numbers 1 to 100 is(n* (n+1))/2 = 5050 when n==100 3. missing element is (5050-SUM) 100 Show More Responses The parameters of the question do not allow you to determine what element is missing. Either more information should be supplied, or all answers are equally correct. How could an array size of 99 elements contain 1 - 100? Should either be integers 1-99 or 2-100 , in either case there is no missing element. All indices are accounted for. Sum them and then subtract them from 5050. In general, if an array of size n - 1 elements has unique elements from 1 to n, then the missing element can be found by subtracting the sum of the elements in the array from sum(1 ... n) = n * (n + 1) / 2. Alternately, one could use a boolean array of length n with all values set to false and then for each value, set array[val - 1] to true. To find the missing value, scan through the array and find the index which is set to false. Return index + 1. This requires O(n) memory and two passes over an O(n) array (instead of constant memory and one pass), but has the advantage of actually allowing you to verify whether or not the input was well formed. Admittedly, this question is poorly posed; however, the answer they are looking for refers to the syntax/nomenclature of some (not all) programming languages to index arrays starting at “0.” As such the 1-100 stored values would be in entries 0-99 of the array. Read the question. Here are the steps to solve it: 1) find the sum of integers 1 to 100 2) subtract the sum of the 99 members of your set 3) the result is your missing element! Very satisfying! Sort array. While loop with an index variable with condition of next element being 1 greater than previous element. When loop breaks, return the value of the index. Doing the expected sum and subtracting the actual gives the run time of O(2n), however a bucket sort will almost always do it in less time (somewhere between O(n) and O(2n)): 1. create a 101-int (or boolean) array (to have a 100-index) 2. traverse original and for each int, assign value in bucket array to 1 or true. 3.After first traversal, traverse created array starting at one, and when value is false, print it. 100 100 coz in array it initial value starts frm 0 to 100. or else 4 further clarification u can study array chapter in c or c++ 100 Show More Responses The question: "An array of 99 elements contains integers from 1 to 100 with one missing element. Find the missing element." The information states that the integer count is 1 to 100. I take this to be inclusive of all elements in the array so that the missing inters would be subjective to their arrangement or random. In other words, I do not have enough information to say which one. 1 I need more information. 1. Are the integers unique in this array? 2. Do I have enough information to find the sum of the integers in the array (or some aggregation)? If sum is available, then, the answer is 5050-sum{integers}. Bucket Sort works and summation works. I think both are good, practical and clever solutions. I think sorting the array then searching may be unnecessary computation. Another interesting method which may be faster. SIMD computers may do this particularly quickly: Do a bitwise operation on all the elements: Result = Array[0] xor Array[1] xor ... Array[98] xor 1 xor 2 xor ... xor 100 Result = Missing number. Explanation: When you xor 2 identical numbers your result = 0. For example, 5 xor 5 -> 101 xor 101 = 000. (5 in decimal is 101 in binary). Knowing that "xoring" 2 identical numbers results in zero is useful. Now we apply this useful info to the problem. Array is Identical to a list of 1,2,3,...,100 except for one number. In other words 1,2,3,...,100 duplicates all of array's elements and adds one extra element that is missing in Array. Therefore, we now have 2 instances of each element in the Array in addition to one extra element in 1,2,3,...,100. We can see when you xor two duplicate numbers you get zero. Because we have pairs for all numbers in Array and one extra number we are essentially "xoring" the missing number with zero. When we xor the missing number with zero we get the missing number. (For example, 6 xor 0 -> 110 xor 000 = 110) The question states that one (not two or three or n) element ("value") from 1 to 100 is missing. There are 99 elements ("values") in the array. The question implies that the data is well-formed because it states that only element is missing. It doesn't ask you to find the missing value(s), but the one (singular) missing element. With the problem constrained, the solution falls out. Subtracting from 5050 is an elegant solution, but not obvious as to why it works. The array of booleans is more obvious, but doesn't scale well. I agree with one of the answers in this thread...5050-sum(elements) = missing item. Other approach that crossed my mind is something similar to binary search. Check the index of 50th element: if A(50) == 50, the missing element > 50, else if A(50) > 50, missing element <50. Do this iteratively. The number of comparisons would be log 100 = 7. 0 100 Add 1-100 to a hash of 100 elements. Then compare each element with the hash.. Answer in o(n) |

### Software Engineer at Apple was asked...

You have a 100 coins laying flat on a table, each with a head side and a tail side. 10 of them are heads up, 90 are tails up. You can't feel, see or in any other way find out which side is up. Split the coins into two piles such that there are the same number of heads in each pile. 34 AnswersAnswer #1: Place 50 coins into two piles on its edges so that both have the same amount of heads in each pile, neither facing up or down. Answer #2: Trick question, place 50 coins in both piles and in theory they all have heads just not necessarily facing up or down. agree with 2nd ans Split into two piles, one with 90 coins and the other with 10. Flip over every coin in the pile with 10 coins. Show More Responses Just split into two piles, each with 50 coins. The question only asks 50 heads in each one, it doesn't ask for the number of heads up!!! Pick 10 coins from the pile, flip it and put it in the other pile. This will ensure that the number of heads up are equal in both the piles Pick 10 coins from the original 100 and put them in a separate pile. Then flip those 10 coins over. The two piles are now guaranteed to have the same number of heads. For a general solution of N heads and a total of M coins: 1.) Pick any N coins out of the original group and form a second pile. 2.) Flip the new pile of N coins over. Done. Example (N=2, M=6): Original group is HHTTTT (mixed randomly). Pick any two of these and flip them over. There are only three possible scenarios: 1: The two coins you picked are both tails. New groups are {HHTT} {TT} and when you flip the 2nd group you have {HHTT} and {HH}. 2.) The two coins you picked consist of one head and one tail. New groups are {HTTT} and {HT} and when you flip the 2nd group you have {HTTT} and {TH}. 3.) The two coins you picked are both heads. New groups are {TTTT} and {HH} and when you flip the 2nd group you have {TTTT} and {TT}. The question says "'You' can't feel, see or in any other way find out which side is up....' Can a team member? Cooperate with a fellow engineer, or other colleague, who can see the coins to solve the problem? Question has its answer in it... 10 coins are head up..... 90 coins are tail down..... so it means all 90 coins are head up.... Now, all you have to do is to split it into half. 50/50 Let's generalise the question to where there are n heads and any number of tails on the table. Select any n coins. This set will contain m heads, where m is between 0 and n inclusive, and n - m tails. The other n - m heads will be in the remaining coins. We now have two piles: the selection of n coins with n-m tails and the remainder with n-m heads. All we have to do is flip the selection so that the n-m tails become n-m heads, the same number as the heads in the remainder. This is a straightforward extension of the 'pick any 10 coins and flip' answer correctly given above by several people. All of you are over thinking it. Read the last bloody line, "Split the coins into two piles such that there are the same number of heads in each pile" They're not asking for the heads to be up or down, just an equal amount & every coin has a head side so dividing the pile equally achieves that. 100 coins total, 10 of them are heads up, 90 are tails up. Meaning all of them are heads up AND tails down. Split it 50/50 and you are done. It is not as easy as to just split it. And it says heads UP tails UP. Given 10 h, 90 t. Pick some random 10 coins call it P1. Rest is P2. In P1, (10-x) heads, (x) tails In P2, (x) heads, (90-x) tails Flip the coins in P1. In P1, (x) heads and (10-x) tails P1 and P2 have the same number of heads. reading these answers is such a confidence builder. Show More Responses I agree to trev, don't think anyone read the question. we already have 2 piles --> 90 coins with tails up and 10 coins with heads up, just flip over 10 of the coins from 90 coins that have tails up, we will have same number of coins with heads up in each pile. get all coins in your hands, shake them, drop them. for each coin there is a 50% probability to lay heads up, and 50% probability for tails down. now split i half question doesn't need to look faces of which side is up after splitting it in two piles. split all coins in two part of 50 50 and they all have heads ...and thats what questioner asking..! and move them to the 10-coin pile. Take 40coins from 90-coin pile, flip them over and move to the 10-coin pile. It's really depends on whether Apple is hiring Software Engineers who are collaborators, mathematicians or tricksters. It's clear that Apple does hire Engineers who listen to the question accurately. Make two groups at random for 10 and 90 coins. Example:- G1(10) G2(90) case 1:- 6H,4T 4H,86T case2:- 3H,7T 7H,83T Now flip all coins of smaller group G1(10). The result will always have same Heads in each pile. G1(10) G2(90) case 1:- 6T,4H 4H,86T case2:- 3T,7H 7H,83T We just get 5 coins head up put in each piles ==> we get the same number of head up in each pile. They just ask we "Split the coins into two piles such that there are the same number of heads in each pile" . They didn't say that we don't kow what is coin head up and they mixed together. "The question says "'You' can't feel, see or in any other way find out which side is up....' Can a team member? Cooperate with a fellow engineer, or other colleague, who can see the coins to solve the problem?" This is the best answer yet! Completely out of the box answer and yet so simple. Show More Responses Flip every other coin, 90 Tails will get split into 45 Heads and 45 Tails. Similarly 10 Heads will get converted to 5 Head and 5 Tails, so now we have 50 heads (45 + 5) and 50 tails (45 + 5). Then just split them into two equal groups. Find complete answer with description here: http://www.puzzlevilla.com/puzzles/puzzle/172 Answer Make a pile of 10 and flip them over. Then the number of heads is equal in both piles. question says both group should have equal heads, but doesnt specifiy, it should be up, hence, just grouping 50 each would solve the problem This is a screw you question, but yeah if you take out 10 coins you can have anywhere between 0-10 heads for every head you have you have one less head in the other pile and one less tail in your pile of 10 coins. So if you have 100 coins 10 heads and you take lets say 10 coins 0 heads, 10 tails. The 90 coins has 10 heads. 1 heads, 9 tails. The 90 coins has 9 heads (you stole one when selecting 10 coins). 2 heads, 8 tails. The 90 coins has 8 heads (same you stole 2 when selecting 10 coins ect). 3 heads, 7 tails. The 90 coins has 7 heads. 4 heads, 6 tails. the 90 coins has 6 heads. 5 heads, 5 tails, the 90 coins has 5 heads. 6 heads, 4 tails, the 90 coins has 4 heads. 7 heads, 3 tails, the 90 coins has 3 heads. 8 heads, 2 tails, the 90 coins has 2 heads. 9 heads, 1 tails, the 90 coins has 1 heads. 10 heads, 0 tails, the 90 coins has 0 heads. As you can see whenever you take out 10 because your not only stealing from the pile of 90's heads your also offsetting the pile of 10 coins tails by 1 equally you have an equal connection between the tails you have in the pile of 10 coins as you do heads in the pile of 90 coins that your tails in 10 coins pile always equals heads in 90 coin pile. So you just flip over each coin in the pile of 10 coins and your tails becomes heads. So if you selected 1 head and in the 10 coins pile you had 9 heads in the 90 coins pile and 9 tails in the 10 coins pile, you are guaranteed after flipping each over once to have 9 heads in the 10 coins pile as tails becomes heads and 9 heads in the 90 coin pile, and ect, ect. This stands true for any pile that you know the amount of one category and 2 options, If you know you have 25 of one things, despite how many things there are if each thing had only two options like heads or tails, you know selecting 25 of them the same amount you know of one thing that when taking out 25 or the equal number of what you know of one thing is in there that what you unsucessfully try to filter out is the inverse of what you selected successfully to take out. Pick 10 coins, flip them and form a separate pile. The no.of tails in both pile will be equal inspite of your choice being a tails up coins or a heads up coins. Coz when u pick a tails up coin u r reducing the no.of tails up in the first pile and since u flip it its gonna b a heads up coin the second pile, if u r picking up a heads up a coin u turn it into a tails up coin in the second pile so that it can cancel out one tails up coin in the existing first pile. If it means heads up then separate the coins into one pile of 90 one pile of 10 then flip the ten coins it works with all scenarios Of sides you ended up choosing also like to point out that we can't feel them so we probably can't use our hands to flip them but I assume they would allow us to use something as how else would we separate them The answer lies in the exact wording of the question "Split the coins into two piles such that there are the same number of heads in each pile. " It does not specify heads need to be face up, so you would simply split the piles in 50 each and you have the same number of coins with heads in each pile. Take ten coins (consider as one pile, Pile A and other 90 coins as another pile, Pile B). Now you have two piles. Turn all coins as in pile A, you will end up with same number of heads in both piles. Ex: Scenario 1: Consider in Pile A, there are 2 heads and 8 tails. Hence in Pile B there will 8 heads.Now when you turn all the coins in Pile A you will end with 8 heads in Pile A. Hence both Pile A and Pile B have same number of heads. Scenario 2: Consider in Pile A, there are 10 heads. Hence in Pile B there will be 0 heads.Now when you turn all the coins in Pile A, you will end with 0 heads in Pile A. Hence both Pile A and Pile B have same number of heads. Scenario 3: Consider in Pile A, there are 0 heads. Hence in Pile B there will be 10 heads.Now when you turn all the coins in Pile A, you will end with 10 heads in Pile A. Hence both Pile A and Pile B have same number of heads. Show More Responses Take 10 coins.Split into two piles of 5 each.Flip all coins in one pile.Both piles now have equal heads and tails.Take another 10 and go through the same procedure.Follow the same process for the entire original pile.You end up with two sets of 5 piles having equal no. of heads and tails.Combine all 5 piles on each side and it's done. |

How do you convince a CIO of a utility to care about energy efficiency? 5 AnswersThis is a bit of a trick question. CIO really does NOT care about efficiency because it means less revenue. However, some states are mandating efficiency programs and so the CIO wants to do this as cost effectively as possible. Some utilities in some states also face competition for energy consumption, so a customer engagement program that makes consumers feel good about their utility will help with rate approvals and satisfaction ratings, which will boost the bottom line overall. He cares because if he can lower the system peak number, he can lesson the amount of spinning reserves he must have in place, whixh in effect is stranded capital. CIO cares since as a company executive he needs to work with his customers (operating unit executives) to cut costs and increase revenues. By providing a solution that mines data to assist customer to save energry usage - the CIO is prioviding a solution for the utility to flatten its energy demand curve. Show More Responses CIO at a utility does not buy this product, nor would she/he care about this. Not their lane. Simple. You demonstrate a cost savings analysis comparing efficient energy use to inefficient energy use. Use a graph of the inefficient use, compare to an efficient use, then run assign the value. That is just one way. This of course assumes that the inefficient use cost more then the cost to correct the problem. But if you do make the changes you can leverage the changes into a PR piece and get a double benefit (efficient use and a "look at us going green!" piece.) |

### Product Design Engineer at Apple was asked...

What are the different ways you can you tell if this part is steel or aluminium. 4 AnswersSimply by using a magnet, Steel has metallic properties, and the magnet will connect. Aluminium will do nothing. Many stainless steel alloys are not magnetic, so if your magnet is attracted to the material you will definitely know it is steel, but if it doesn't you will not know what the material is. Simple methods would be density (feeling the mass of the object), surface finish (color, texture). If coated that may give you the answer i.ie anodized would indicate aluminium. For more information I would go for EDX( Energy-dispersive X-ray spectroscopy) and possible a cross section to look at the grain structure. Show More Responses By far the easiest way is to test for material properties. -density -hardness -modulus of elasticity I would choose hardness. Strike each item with an equal force, which one deforms more? Thats aluminum. You could probably pull this test off with a hammer. The simplest solutions is always the best. |

In a given sorted array of integers remove all the duplicates. 6 AnswersIterate the array and add each number to a set, if number is already there, it won't be added again, thus removing any duplicates. Complexity is Big-O of N The array is already sorted, no need for a set. example: 2,2,5,7,7,8,9 Just keep tracking the current and previous and the index of the last none repeated element when found a difference copy the element to the last none repeated index + one and update current and previous, no extra space and it will run in O(n) public RemoveDuplicates() { int[] ip = { 1, 2, 2, 4, 5, 5, 8, 9, 10, 11, 11, 12 }; int[] op = new int[ip.Length - 1]; int j = 0, i = 0; ; for (i = 1; i <= ip.Length - 1; i++) { if (ip[i - 1] != ip[i]) { op[j] = ip[i - 1]; j++; } } if (ip[ip.Length - 1] != ip[ip.Length - 2]) op[j] = ip[ip.Length - 1]; int xxx = 0; } Show More Responses def removeDuplicatesSecondApproach(inputArray): prev = 0 noRepeatIndex = 0 counter = 0 for curr in range(1,len(inputArray)): if (inputArray[curr] == inputArray[prev]): counter = counter + 1 prev = curr else: inputArray[noRepeatIndex+1] = inputArray[curr] noRepeatIndex = noRepeatIndex + 1 prev = curr inputArray = inputArray[:-counter] return inputArray if(inpArr[i] == inpArr[i+1]) { int repeats = 1; opArr[opPos] = inpArr[i]; int j = i + 1; while(j+1 <= inpArr.length - 1 && inpArr[i] == inpArr[j+1]) { j++; repeats++; } opArr = Arrays.copyOf(opArr, opArr.length - repeats); i = i + repeats; } else { opArr[opPos] = inpArr[i]; } opPos++; } for(int i =0; i<=opArr.length-1;i++) { System.out.println(opArr[i] + ","); } Apologies for the previous incomplete answer int[] inpArr = {1,2,2,3,4,5,5,5,8,8,8,9,13,14,15,18,20,20}; int[] opArr = new int[inpArr.length]; int opPos = 0; for(int i= 0; i<=inpArr.length - 1; i++) { if(inpArr[i] == inpArr[i+1]) { int repeats = 1; opArr[opPos] = inpArr[i]; int j = i + 1; while(j+1 <= inpArr.length - 1 && inpArr[i] == inpArr[j+1]) { j++; repeats++; } opArr = Arrays.copyOf(opArr, opArr.length - repeats); i = i + repeats; } else { opArr[opPos] = inpArr[i]; } opPos++; } for(int i =0; i<=opArr.length-1;i++) { System.out.println(opArr[i] + ","); } |

how can a particular application be tested apart from testing its functionality 3 AnswersReliability Test, Stability Test, UI Test, Platform Test, Also include, performance, stress & load testing Accessibility, user experience, globalization, localization, integration, compatibility |

How do you influence people who do not report to you? How do you get them to do things even though they don't have to or want to? 2 Answersa) I've never actually had someone refuse to help me when asked, but b) I would try ahead of time to get to know them so that when I need to influence them or need to request assistance, they already have a basic level of comfort with me. I would answer this question by first stating that it is important to get to know how people are motivated on a personal level first. If they do not report to you then you might not have positional authority to ask them for a deliverable. At a company like Microsoft, I'm sure that there are jobs/roles that require people to get help/collaboration from people outside of their positional authority. To do this, you must first know what motivates a person and proactively make personal "deposits" to build some solid connections so that when you need to make a "withdraw" you have invested enough. Deposits can be made by taking an interest in someone, providing them with help when they need it, or maybe buying them lunch/drinks. |

### Network Engineer at Microsoft was asked...

Given a very large existing network with thousands of external connections, how would you add an additional few hundred connections? 3 AnswersThink it through, out loud. Walk through your options- don't just pick one. DS3, metro-E, SONET, etc. someone building networks with thousands of external connections considering anything but 10G, 1G at minimum... doesn't understand COGS.... the name of the game is $0.000xxx/Mbps.. which you are not going to get with anything but ethernet. SNMP/NetFlow to monitor the WAN Bandwidth Utilization + Characterize the Traffic (what type of traffic is using the network, QoS parameters for applications, CoS) a. Characterizing the traffic: 1. identify major traffic sources and stores and documented traffic flow between them. (Voice, Video) 2. categorize the traffic flow for each application as being terminal/host, client/server, peer-to-peer, server/server, or distributed computing. 3. estimate the bandwidth requirements for each application. 4. estimate the bandwidth requirements for routing protocols. 5. characterize network traffic in terms of broadcast/multicast rates, efficiency, frame sizes, windowing and flow control, and error-recovery mechanisms. 6. categorize the QoS requirements of each application and what are the current trends 7. discuss the challenges associated with implementing end-to-end QoS and the need for devices across the network to do their part in implementing QoS strategies. When the team determines that bandwidth augmentation is necessary based on the utilization reports of NetFlow and comparing it with the companies’ policies that the traffic is going over the threshold during the business hours, provides a recommendation to network engineers and managers. Once it is determined that an upgrade is necessary, proceeding with the upgrade becomes a business decision. |

Write code in your favorite programming language that will accept two strings and return true if they are anagrams. 2 AnswersThis was not really that hard to write it, however the interviewer asked me to reduce the complexity. My initial version had n*log(n) complexity and he asked me to reduce it to no more than n complexity. If you have had some upper level Computer Science classes this is not too difficult, however what they are looking for is a way to stump you. If you adjust your code or thinking rapidly to their request they will change it again until they find something that you have trouble with. Do not be discouraged by this, it is the interviewers job to determine how much you know! Found this good link. Time complexity is O(n). http://www.dreamincode.net/code/snippet1481.htm The algorithm can still be improved but gives some basic idea on how to implement. |

What is a JavaScript callback function? 4 Answers5 vote down star 4 I understand passing in a function to another function as a callback and having it execute, but I'm not understanding the best implementation to do that. I'm looking for a very basic example, like this: var myCallBackExample = { myFirstFunction : function( param1, param2, callback ) { // Do something with param1 and param2. if ( arguments.length == 3 ) { // Execute callback function. // What is the "best" way to do this? } }, mySecondFunction : function() { myFirstFunction( false, true, function() { // When this anonymous function is called, execute it. }); } }; In myFirstFunction, if I do return new callback(), then it works and executes the anonymous function, but that doesn't seem like the correct approach to me. I don't think Bloomberg is a very good company. I am an excellent web developer and have gotten multiple offers from other companies with big names, but was rejected by Bloomberg. They are too demanding during the job interview and it becomes a game of how well you can interview as opposed to how talented an employee you are and how much you can contribute to the growth of the company. A callback function is a piece of JavaScript code that executes after the main function that the callback is attached to executes successfully. Show More Responses udaykanth, I would say that a .forEach() would be the most common and most basic use of a callback function. I'm just writing this to help anyone that might have a hard time thinking up a quick example if the come across this question themselves. Example: var numArray = [ 1, 2, 3 ] ; numArray.forEach( function( i ) { console.log( arr[ i - 1 ] ) } ) ; // logs out // 1 // 2 // 3 |