I have a question which I am trying to solve for personal understanding of comparing algorithms as follows. I am given n to be the number of objects and m be the number of keys. I want to find the ratios objects/keys (m/n) that CountingSort is faster in the worst case compared to QuickSort when quicksort chooses the last element as the pivot.
So I have that the worst case running time for CountingSort is O(n+k) and the case when quicksort chooses the last element as the pivot is Theta(n^2). I'm confused of how to approach this question and would welcome some guidance so I could reach a solution.
My idea is that it will be for when the number of keys is double the number of objects? So we have that the runtime for worst case of quicksort is O(n^2) so we want to have counting sort less than this. Therefore if we have rations of m/n < m^2/n this will be the case?
Related
I am prepping for interview leet-code type problems and I came across the k closest problem, but given a sorted array. This problem requires finding the k closest elements by value to an input value from the array. The answer to this problem was fairly straight forward and I did not have any issues determining a linear-time algorithm to solve it.
However, working on this problem got me thinking. Is it possible to solve this problem given an unsorted array in linear time? My first thought was to use a heap and that would give an O(nlogk) time complexity solution, but I am trying to determine if its possible to come up with an O(n) solution? I was thinking about possibly using something like quickselect, but the issue is that this has an expected time of O(n), not a worst case time of O(n).
Is this even possible?
The median-of-medians algorithm makes Quickselect take O(n) time in the worst case.
It is used to select a pivot:
Divide the array into groups of 5 (O(n))
Find the median of each group (O(n))
Use Quickselect to find the median of the n/5 medians (O(n))
The resulting pivot is guaranteed to be greater and less than 30% of the elements, so it guarantees linear time Quickselect.
After selecting the pivot, of course, you have to continue on with the rest of Quickselect, which includes a recursive call like the one we made to select the pivot.
The worst case total time is T(n) = O(n) + T(0.7n) + T(n/5), which is still linear. Compared to the expected time of normal Quickselect, though, it's pretty slow, which is why we don't often use this in practice.
Your heap solution would be very welcome at an interview, I'm sure.
If you really want to get rid of the logk, which in practical applications should seldom be a problem, then yes, using Quickselect would be another option. Something like this:
Partition your array in values smaller and larger than x. <- O(n).
For the lower half, run Quickselect to find the kth largest number, then take the right-side partition which are your k largest numbers. <- O(n)
Repeat step 2 for the higher half, but for the k smallest numbers. <- O(n)
Merge your k smallest and k largest numbers and extract the k closest numbers. <- O(k)
This gives you a total time complexity of O(n), as you said.
However, a few points about your worry about expected time vs worst-case time. I understand that if an interview question explicitly insists on worst-case O(n), then this solution might not be accepted, but otherwise, this can well be considered O(n) in practice.
The key here being that for randomized quickselect and random or well-behaved input, the probability that the time complexity goes beyond O(n) decreases exponentially as the input grows. Meaning that already at largeish inputs, the probability is as small as guessing at a specific atom in the known universe. The assumption on well-behaved input concerns being somewhat random in nature and not adversarial. See this discussion on a similar (not identical) problem.
If an algorithm must make n-1 comparisons to find a certain element, then can we assume that best possible runtime of the algorithm is O(n)?
I know that the lower bound for sorting algorithms is nlogn but since we only return the found one element, I figured it would be possible to do better in terms of run time?
Thanks!
To find a certain element in an unsorted list you need O(n).
But if you sort the array (takes O(n log n) in general) you can find a certain element in O(log n).
So if you want to find often elements in the same list it is most likely worth to sort the list to then be able to find elements much more efficient.
If your array is unsorted and you find some element in it then in worst case Linear search algorithm make n-1 comparisons and time complexity will be O(n).
But if you want to reduce your time complexity then first sort your array and use Binary search algorithm it is take O(logn) in worst case.
So Binary search algorithm is more efficient then linear search.
For unsorted elements, worst case is when you have to go over all the elements, i.e., O(N). If you need many look-ups then you have several pre-processing alternatives that speed up all future accesses.
Option 1: put the elements in a standard hash table. Creating the hash table costs O(N), on average, and later pay O(1) on average for each lookup. This assumes that a reasonable hash-function can be created for this type of elements.
Most languages/libraries implement bucket-based hash-tables, which in pathological cases can put all elements in one bucket, costing O(N) per lookup.
Option 2: there are other hash-table implementations that don't suffer from pathological O(N) cases. The Robin Hood hashing (Wikipedia) (more at Programming.Guide) guarantees O(log N) lookup in the worst case, with average of O(1).
Option 3: another option is to sort elements in O(N log N) once, and then use binary-search to lookup in O(log N). Usually this is slower than Robin Hood hashing (Option 2).
Option 4: If the values are simple integers with limited range, with max-min around N, then it is possible to put the values in an array (list), such that array[value-min] will contain a count of how many times the value appears in the input. It costs O(N) to construct, and O(1) to lookup. Better, the constants for both preprocessing and lookup are significantly lower than in any other method.
Note: I didn't mention the O(N) counting-sort as an alternative to the general case of O(N log N) sorting (option 3), since if max(value)-min(value) is small enough for counting-sort, then option 4 is relevant and is simpler and faster.
If applicable, choose option 4, otherwise if you wish to invest time and code then choose option 2. If 4 isn't applicable, and 2 is not worth the effort in your case, then choose option 2 if you don't mind the pathological worst-case (never choose option 2 when an adversary may want to harm you in a DOS attack).
Your question has nothing to do with sorting, let alone linear search.
If you claim that n-1 comparisons are mandated, then your problem has certainly complexity Ω(n). But with that information alone, you can't guarantee O(n) because it is not said that these n-1 comparisons are sufficient, nor that the algorithm does not perform extra operations, for instance to decide which comparisons to perform. It could turn out that your algorithm is O(n³) with no chance to do better, but we can't tell.
Best case complexity: Ω(n).
Worst case complexity: unknown.
Lets say that I want to find the K max values in an array of n elements , and also return them in a sorted output.
k may be -
k = 30 , k = n/5 ..
I thought about some efficient algorithms but all I could think of was in complexity of O(nlogn). Can I do it in `O(n)? maybe with some modification of quick sort?
Thanks!
The problem could be solved using min-heap-based priority queue in
O(NlogK) + (KlogK) time
If k is constant (k=30 case), then complexity is equal to O(N).
If k = O(N) (k=n/5 case), then complexity is equal to O(NlogN).
Another option for constant k - K-select algorithm based on Quicksort partition with average time O(N) (while worst case O(N^2) might occur)
There is a way of sorting elements in nearly O(n), if you assume that you only want to sort integers. This can be done with Algorithms like Bucket Sort or Radix Sort, that do not rely on the comparison between two elements (which are limited to O(n*log(n))).
However note, that these algorithms also have worst-case runtimes, that might be slower than O(n*log(n)).
More information can be found here.
No comparison based sorting algorithms can achieve a better average case complexity than O(n*lg n)
There are many papers with proofs out there but this site provides a nice visual example.
So unless you are given a sorted array, your best case is going to be an O(n lg n) algorithm.
There are sorts like radix and bucket, but they are not based off of comparison based sorting like your title seems to imply.
This question was in the preparation exam for my midterm in introduction to computer science.
There exists an algorithm which can find the kth element in a list in
O(n) time, and suppose that it is in place. Using this algorithm,
write an in place sorting algorithm that runs in worst case time
O(n*log(n)), and prove that it does. Given that this algorithm exists,
why is mergesort still used?
I assume I must write some alternate form of the quicksort algorithm, which has a worst case of O(n^2), since merge-sort is not an in-place algorithm. What confuses me is the given algorithm to find the kth element in a list. Isn't a simple loop iteration through through the elements of an array already a O(n) algorithm?
How can the provided algorithm make any difference in the running time of the sorting algorithm if it does not change anything in the execution time? I don't see how used with either quicksort, insertion sort or selection sort, it could lower the worst case to O(nlogn). Any input is appreciated!
Check wiki, namely the "Selection by sorting" section:
Similarly, given a median-selection algorithm or general selection algorithm applied to find the median, one can use it as a pivot strategy in Quicksort, obtaining a sorting algorithm. If the selection algorithm is optimal, meaning O(n), then the resulting sorting algorithm is optimal, meaning O(n log n). The median is the best pivot for sorting, as it evenly divides the data, and thus guarantees optimal sorting, assuming the selection algorithm is optimal. A sorting analog to median of medians exists, using the pivot strategy (approximate median) in Quicksort, and similarly yields an optimal Quicksort.
The short answer why mergesort is prefered over quicksort in some cases is that it is stable (while quicksort is not).
Reasons for merge sort. Merge Sort is stable. Merge sort does more moves but fewer compares than quick sort. If the compare overhead is greater than move overhead, then merge sort is faster. One situation where compare overhead may be greater is sorting an array of indices or pointers to objects, like strings.
If sorting a linked list, then merge sort using an array of pointers to the first nodes of working lists is the fastest method I'm aware of. This is how HP / Microsoft std::list::sort() is implemented. In the array of pointers, array[i] is either NULL or points to a list of length pow(2,i) (except the last pointer points to a list of unlimited length).
I found the solution:
if(start>stop) 2 op.
pivot<-partition(A, start, stop) 2 op. + n
quickSort(A, start, pivot-1) 2 op. + T(n/2)
quickSort(A, pibvot+1, stop) 2 op. + T(n/2)
T(n)=8+2T(n/2)+n k=1
=8+2(8+2T(n/4)+n/2)+n
=24+4T(n/4)+2n K=2
...
=(2^K-1)*8+2^k*T(n/2^k)+kn
Recursion finishes when n=2^k <==> k=log2(n)
T(n)=(2^(log2(n))-1)*8+2^(log2(n))*2+log2(n)*n
=n-8+2n+nlog2(n)
=3n+nlog2(n)-8
=n(3+log2(n))-8
is O(nlogn)
Quick sort have worstcase O(n^2), but that only occurs if you have bad luck when choosing the pivot. If you can select the kth element in O(n) that means you can choose a good pivot by doing O(n) extra steps. That yields a woest-case O(nlogn) algorithm. There are a couple of reasons why mergesort is still used. First, this selection algorithm is more or less cumbersome to implement in-place, and also adds several extra operations to the regular quicksort, so it is not that fastest than merge sort, as one might expect.
Nevertheless, MergeSort is not still used because of its worst time complexity, in fact HeapSort achieves the same worst case bounds and is also in place, and didn't replace MergeSort, though it has also other disadvantages against quicksort. The main reason why MergeSort survives is because it is the fastest stable sort algorithm know so far. There are several applications in which is paramount to have an stable sorting algorithm. And that is the strength of MergeSort.
A stable sort is such that the equal items preserve the original relative order. For example, this is very useful when you have two keys, and you want to sort by first key first and then by second key, preserving the first key order.
The problem with HeapSort against quicksort is that it is cache inefficient, since you swap/compare elements too far from each other in the array, while quicksort compares consequent elements, these elements are more likely to be in the cache at the same time.
Suppose we choose a pivot as the first element of an array in case of a quicksort. Now the best/worst case complexity is O(n^2) whereas in average case it is O(nlogn). Is not it weird (best case complexity is greater than worst case complexity)?
The best case complexity is O(nlogn), as the average case. The worst case is O(n^2). Check http://en.wikipedia.org/wiki/Quick_sort.
While other algorithms like Merge Sort and Heap Sort have a better worst case complexity (O(nlogn)), usually Quick Sort is faster - this is why it's the most common used sorting algorithm. An interesting answer about this can be found at Why is quicksort better than mergesort?.
The best-case of quicksort 0(nlogn) is when the chosen pivot splits the subarray in two +- equally sized parts in every iteration.
Worst-case of quicksort is when the chosen pivot is the smallest element in the subarray, so that the array is split in two parts where one part consists of one element (the pivot) and the other part of all the other elements of the subarray.
So choosing the first element as the pivot in an already sorted array, will get you 0(n^2). ;)
Therefore it is important to choose a good pivot. For example by using the median of the first, middle and last element of the subarray, as the pivot.