Related
I was set a homework challenge as part of an application process (I was rejected, by the way; I wouldn't be writing this otherwise) in which I was to implement the following functions:
// Store a collection of integers
class IntegerCollection {
public:
// Insert one entry with value x
void Insert(int x);
// Erase one entry with value x, if one exists
void Erase(int x);
// Erase all entries, x, from <= x < to
void Erase(int from, int to);
// Return the count of all entries, x, from <= x < to
size_t Count(int from, int to) const;
The functions were then put through a bunch of tests, most of which were trivial. The final test was the real challenge as it performed 500,000 single insertions, 500,000 calls to count and 500,000 single deletions.
The member variables of IntegerCollection were not specified and so I had to choose how to store the integers. Naturally, an STL container seemed like a good idea and keeping it sorted seemed an easy way to keep things efficient.
Here is my code for the four functions using a vector:
// Previous bit of code shown goes here
private:
std::vector<int> integerCollection;
};
void IntegerCollection::Insert(int x) {
/* using lower_bound to find the right place for x to be inserted
keeps the vector sorted and makes life much easier */
auto it = std::lower_bound(integerCollection.begin(), integerCollection.end(), x);
integerCollection.insert(it, x);
}
void IntegerCollection::Erase(int x) {
// find the location of the first element containing x and delete if it exists
auto it = std::find(integerCollection.begin(), integerCollection.end(), x);
if (it != integerCollection.end()) {
integerCollection.erase(it);
}
}
void IntegerCollection::Erase(int from, int to) {
if (integerCollection.empty()) return;
// lower_bound points to the first element of integerCollection >= from/to
auto fromBound = std::lower_bound(integerCollection.begin(), integerCollection.end(), from);
auto toBound = std::lower_bound(integerCollection.begin(), integerCollection.end(), to);
/* std::vector::erase deletes entries between the two pointers
fromBound (included) and toBound (not indcluded) */
integerCollection.erase(fromBound, toBound);
}
size_t IntegerCollection::Count(int from, int to) const {
if (integerCollection.empty()) return 0;
int count = 0;
// lower_bound points to the first element of integerCollection >= from/to
auto fromBound = std::lower_bound(integerCollection.begin(), integerCollection.end(), from);
auto toBound = std::lower_bound(integerCollection.begin(), integerCollection.end(), to);
// increment pointer until fromBound == toBound (we don't count elements of value = to)
while (fromBound != toBound) {
++count; ++fromBound;
}
return count;
}
The company got back to me saying that they wouldn't be moving forward because my choice of container meant the runtime complexity was too high. I also tried using list and deque and compared the runtime. As I expected, I found that list was dreadful and that vector took the edge over deque. So as far as I was concerned I had made the best of a bad situation, but apparently not!
I would like to know what the correct container to use in this situation is? deque only makes sense if I can guarantee insertion or deletion to the ends of the container and list hogs memory. Is there something else that I'm completely overlooking?
We cannot know what would make the company happy. If they reject std::vector without concise reasoning I wouldn't want to work for them anyway. Moreover, we dont really know the precise requirements. Were you asked to provide one reasonably well performing implementation? Did they expect you to squeeze out the last percent of the provided benchmark by profiling a bunch of different implementations?
The latter is probably too much for a homework challenge as part of an application process. If it is the first you can either
roll your own. It is unlikely that the interface you were given can be implemented more efficiently than one of the std containers does... unless your requirements are so specific that you can write something that performs well under that specific benchmark.
std::vector for data locality. See eg here for Bjarne himself advocating std::vector rather than linked lists.
std::set for ease of implementation. It seems like you want the container sorted and the interface you have to implement fits that of std::set quite well.
Let's compare only isertion and erasure assuming the container needs to stay sorted:
operation std::set std::vector
insert log(N) N
erase log(N) N
Note that the log(N) for the binary_search to find the position to insert/erase in the vector can be neglected compared to the N.
Now you have to consider that the asymptotic complexity listed above completely neglects the non-linearity of memory access. In reality data can be far away in memory (std::set) leading to many cache misses or it can be local as with std::vector. The log(N) only wins for huge N. To get an idea of the difference 500000/log(500000) is roughly 26410 while 1000/log(1000) is only ~100.
I would expect std::vector to outperform std::set for considerably small container sizes, but at some point the log(N) wins over cache. The exact location of this turning point depends on many factors and can only reliably determined by profiling and measuring.
Nobody knows which container is MOST efficient for multiple insertions / deletions. That is like asking what is the most fuel-efficient design for a car engine possible. People are always innovating on the car engines. They make more efficient ones all the time. However, I would recommend a splay tree. The time required for a insertion or deletion is a splay tree is not constant. Some insertions take a long time and some take only a very a short time. However, the average time per insertion/deletion is always guaranteed to be be O(log n), where n is the number of items being stored in the splay tree. logarithmic time is extremely efficient. It should be good enough for your purposes.
The first thing that comes to mind is to hash the integer value so single look ups can be done in constant time.
The integer value can be hashed to compute an index in to an array of bools or bits, used to tell if the integer value is in the container or not.
Counting and and deleting large ranges could be sped up from there, by using multiple hash tables for specific integer ranges.
If you had 0x10000 hash tables, that each stored ints from 0 to 0xFFFF and were using 32 bit integers you could then mask and shift the upper half of the int value and use that as an index to find the correct hash table to insert / delete values from.
IntHashTable containers[0x10000];
u_int32 hashIndex = (u_int32)value / 0x10000;
u_int32int valueInTable = (u_int32)value - (hashIndex * 0x10000);
containers[hashIndex].insert(valueInTable);
Count for example could be implemented as so, if each hash table kept count of the number of elements it contained:
indexStart = startRange / 0x10000;
indexEnd = endRange / 0x10000;
int countTotal = 0;
for (int i = indexStart; i<=indexEnd; ++i) {
countTotal += containers[i].count();
}
Not sure if using sorting really is a requirement for removing the range. It might be based on position. Anyway, here is a link with some hints which STL container to use.
In which scenario do I use a particular STL container?
Just FYI.
Vector maybe a good choice, but it does a lot of re allocation, as you know. I prefer deque instead, as it doesn't require big chunk of memory to allocate all items. For such requirement as you had, list probably fit better.
Basic solution for this problem might be std::map<int, int>
where key is the integer you are storing and value is the number of occurences.
Problem with this is that you can not quickly remove/count ranges. In other words complexity is linear.
For quick count you would need to implement your own complete binary tree where you can know the number of nodes between 2 nodes(upper and lower bound node) because you know the size of tree, and you know how many left and right turns you took to upper and lower bound nodes. Note that we are talking about complete binary tree, in general binary tree you can not make this calculation fast.
For quick range remove I do not know how to make it faster than linear.
I have a C++11 list of complex elements that are defined by a structure node_info. A node_info element, in particular, contains a field time and is inserted into the list in an ordered fashion according to its time field value. That is, the list contains various node_info elements that are time ordered. I want to remove from this list all the nodes that verify some specific condition specified by coincidence_detect, which I am currently implementing as a predicate for a remove_if operation.
Since my list can be very large (order of 100k -- 10M elements), and for the way I am building my list this coincidence_detect condition is only verified by few (thousands) elements closer to the "lower" end of the list -- that is the one that contains elements whose time value is less than some t_xv, I thought that to improve speed of my code I don't need to run remove_if through the whole list, but just restrict it to all those elements in the list whose time < t_xv.
remove_if() though does not seem however to allow the user to control up to which point I can iterate through the list.
My current code.
The list elements:
struct node_info {
char *type = "x";
int ID = -1;
double time = 0.0;
bool spk = true;
};
The predicate/condition for remove_if:
// Remove all events occurring at t_event
class coincident_events {
double t_event; // Event time
bool spk; // Spike condition
public:
coincident_events(double time,bool spk_) : t_event(time), spk(spk_){}
bool operator()(node_info node_event){
return ((node_event.time==t_event)&&(node_event.spk==spk)&&(strcmp(node_event.type,"x")!=0));
}
};
The actual removing from the list:
void remove_from_list(double t_event, bool spk_){
// Remove all events occurring at t_event
coincident_events coincidence(t_event,spk_);
event_heap.remove_if(coincidence);
}
Pseudo main:
int main(){
// My list
std::list<node_info> event_heap;
...
// Populate list with elements with random time values, yet ordered in ascending order
...
remove_from_list(0.5, true);
return 1;
}
It seems that remove_if may not be ideal in this context. Should I consider instead instantiating an iterator and run an explicit for cycle as suggested for example in this post?
It seems that remove_if may not be ideal in this context. Should I consider instead instantiating an iterator and run an explicit for loop?
Yes and yes. Don't fight to use code that is preventing you from reaching your goals. Keep it simple. Loops are nothing to be ashamed of in C++.
First thing, comparing double exactly is not a good idea as you are subject to floating point errors.
You could always search the point up to where you want to do a search using lower_bound (I assume you list is properly sorted).
The you could use free function algorithm std::remove_if followed by std::erase to remove items between the iterator returned by remove_if and the one returned by lower_bound.
However, doing that you would do multiple passes in the data and you would move nodes so it would affect performance.
See also: https://en.cppreference.com/w/cpp/algorithm/remove
So in the end, it is probably preferable to do you own loop on the whole container and for each each check if it need to be removed. If not, then check if you should break out of the loop.
for (auto it = event_heap.begin(); it != event_heap.end(); )
{
if (coincidence(*it))
{
auto itErase = it;
++it;
event_heap.erase(itErase)
}
else if (it->time < t_xv)
{
++it;
}
else
{
break;
}
}
As you can see, code can easily become quite long for something that should be simple. Thus, if you need to do that kind of algorithm often, consider writing you own generic algorithm.
Also, in practice you might not need to do a complete search for the end using the first solution if you process you data in increasing time order.
Finally, you might consider using an std::set instead. It could lead to simpler and more optimized code.
Thanks. I used your comments and came up with this solution, which seemingly increases speed by a factor of 5-to-10.
void remove_from_list(double t_event,bool spk_){
coincident_events coincidence(t_event,spk_);
for(auto it=event_heap.begin();it!=event_heap.end();){
if(t_event>=it->time){
if(coincidence(*it)) {
it = event_heap.erase(it);
}
else
++it;
}
else
break;
}
}
The idea to make erase return it (as already ++it) was suggested by this other post. Note that in this implementation I am actually erasing all list elements up to t_event value (meaning, I pass whatever I want for t_xv).
Does anyone know if it's possible to turn this from O(m * n) to O(m + n)?
vector<int> theFirst;
vector<int> theSecond;
vector<int> theMatch;
theFirst.push_back( -2147483648 );
theFirst.push_back(2);
theFirst.push_back(44);
theFirst.push_back(1);
theFirst.push_back(22);
theFirst.push_back(1);
theSecond.push_back(1);
theSecond.push_back( -2147483648 );
theSecond.push_back(3);
theSecond.push_back(44);
theSecond.push_back(32);
theSecond.push_back(1);
for( int i = 0; i < theFirst.size(); i++ )
{
for( int x = 0; x < theSecond.size(); x++ )
{
if( theFirst[i] == theSecond[x] )
{
theMatch.push_back( theFirst[i] );
}
}
}
Put the contents of the first vector into a hash set, such as std::unordered_set. That is O(m). Scan the second vector, checking if the values are in the unordered_set and keeping a tally of those that are. That is n lookups of a hash structure, so O(n). So, O(m+n). If you have l elements in the overlap, you may count O(l) for adding them to the third vector. std::unordered_set is in the C++0x draft and available in the latest gcc versions, and there is also an implementation in boost.
Edited to use unordered_set
Using C++2011 syntax:
unordered_set<int> firstMap(theFirst.begin(), theFirst.end());
for (const int& i : theSecond) {
if (firstMap.find(i)!=firstMap.end()) {
cout << "Duplicate: " << i << endl;
theMatch.push_back(i);
}
}
Now, the question still remains, what do you want to do with duplicates in the originals? Explicitly, how many times should 1 be in theMatch, 1, 2 or 4 times?
This outputs:
Duplicate: 1
Duplicate: -2147483648
Duplicate: 44
Duplicate: 1
Using this: http://www.cplusplus.com/reference/algorithm/set_intersection/
You should be able to achieve O(mlogm + nlogn) I believe. (set_intersection requires that the input ranges be already sorted).
This might perform a bit differently than your solution for duplicate elements, however.
Please correct me if I am wrong,
you are suggesting following solution for the intersection problem:
sort two vectors, and keep iteration in both sorted vector in such a way that we reach to a common element,
so overall complexity will be
(n*log(n) + m*log(m)) + (n + m)
Assuming k*log(k) as complexity of sorting
Am I right?
Ofcourse the complexity will depend on the complexity of sorting.
I would sort the longer array O(n*log (n)), search for elements from the shorter array O(m*log (n)). Total is then O(n*log(n) + m*log (n) )
Assuming you want to produce theMatch from two data sets, and you don't care about the data sets themselves, put one in an unordered_map (available currently from Boost and listed in the final committee draft for C++11), mapping the key to an integer that increases whenever added to, and therefore keeps track of the number of times the key occurs. Then, when you get a hit on the other data set, you push_back the hit the number of times it occurred in the first time.
You can get to O(n log n + m log m) by sorting the vectors first, or O(n log n + m) by creating a std::map of one of them.
Caveat: these are not order-preserving operations, and theMatch will come out in different orders with different techniques. It looks to me like the order is likely considered arbitrary. If the order given in the code above is necessary, I don't think there's a better algorithm.
Edit:
Take data set A and data set B, of type Type. Create an unordered_map<Type, int>.
Go through data set A, and check each member to see if it's in the map. If not, add the element with the int 1 to the map. If it is, increment the int. Each of these operations is O(1) on the average, so this step is O(len A).
Go through data set B, and check each member to see if it's in the map. If not, go on to the next. If so, push_back the member onto the destination queue. The int is the number of times that value is in data set A, so do the push_back the number of times the member's in A to duplicate the behavior given. Each of these operations is on the average O(1), so this step is O(len B).
This is average behavior. If you always hit the worst case, you're back with O(m*n). I don't think there's a way to guarantee O(m + n).
If the order of the elements in the resulting array/set doesn't matter then the answer is yes.
For the arbitrary types of elements with some order defined the best algorithm is O( max(m,n)*log(min(m,n)) ). For the numbers of limited size the best algorithm is O(m+n).
Construct the set of elements of smaller array - for arbitrary elements just sorting is OK and for the numbers of limited size it must be something similar to intermediate table in numeric sort.
Iterate through larger array and check if the element is within a set constructed earlier - for the arbitrary element binary search is OK (which is O(log(min(n,m))) and for numbers the single check is O(1).
I have a data structure like this:
struct X {
float value;
int id;
};
a vector of those (size N (think 100000), sorted by value (stays constant during the execution of the program):
std::vector<X> values;
Now, I want to write a function
void subvector(std::vector<X> const& values,
std::vector<int> const& ids,
std::vector<X>& out /*,
helper data here */);
that fills the out parameter with a sorted subset of values, given by the passed ids (size M < N (about 0.8 times N)), fast (memory is not an issue, and this will be done repeatedly, so building lookuptables (the helper data from the function parameters) or something else that is done only once is entirely ok).
My solution so far:
Build lookuptable lut containing id -> offset in values (preparation, so constant runtime)
create std::vector<X> tmp, size N, filled with invalid ids (linear in N)
for each id, copy values[lut[id]] to tmp[lut[id]] (linear in M)
loop over tmp, copying items to out (linear in N)
this is linear in N (as it's bigger than M), but the temporary variable and repeated copying bugs me. Is there a way to do it quicker than this? Note that M will be close to N, so things that are O(M log N) are unfavourable.
Edit: http://ideone.com/xR8Vp is a sample implementation of mentioned algorithm, to make the desired output clear and prove that it's doable in linear time - the question is about the possibility of avoiding the temporary variable or speeding it up in some other way, something that is not linear is not faster :).
An alternative approach you could try is to use a hash table instead of a vector to look up ids in:
void subvector(std::vector<X> const& values,
std::unordered_set<int> const& ids,
std::vector<X>& out) {
out.clear();
out.reserve(ids.size());
for(std::vector<X>::const_iterator i = values.begin(); i != values.end(); ++i) {
if(ids.find(i->id) != ids.end()) {
out.push_back(*i);
}
}
}
This runs in linear time since unordered_set::find is constant expected time (assuming that we have no problems hashing ints). However I suspect it might not be as fast in practice as the approach you described initially using vectors.
Since your vector is sorted, and you want a subset of it sorted the same way, I assume we can just slice out the chunk you want without rearranging it.
Why not just use find_if() twice. Once to find the start of the range you want and once to find the end of the range. This will give you the start and end iterators of the sub vector. Construct a new vector using those iterators. One of the vector constructor overloads takes two iterators.
That or the partition algorithm should work.
If I understood your problem correctly, you actually try to create a linear time sorting algorithm (subject to the input size of numbers M).
That is NOT possible.
Your current approach is to have a sorted list of possible values.
This takes linear time to the number of possible values N (theoretically, given that the map search takes O(1) time).
The best you could do, is to sort the values (you found from the map) with a quick sorting method (O(MlogM) f.e. quicksort, mergesort etc) for small values of M and maybe do that linear search for bigger values of M.
For example, if N is 100000 and M is 100 it is much faster to just use a sorting algorithm.
I hope you can understand what I say. If you still have questions I will try to answer them :)
edit: (comment)
I will further explain what I mean.
Say you know that your numbers will range from 1 to 100.
You have them sorted somewhere (actually they are "naturally" sorted) and you want to get a subset of them in sorted form.
If it would be possible to do it faster than O(N) or O(MlogM), sorting algorithms would just use this method to sort.
F.e. by having the set of numbers {5,10,3,8,9,1,7}, knowing that they are a subset of the sorted set of numbers {1,2,3,4,5,6,7,8,9,10} you still can't sort them faster than O(N) (N = 10) or O(MlogM) (M = 7).
I'm intersecting some sets of numbers, and doing this by storing a count of each time I see a number in a map.
I'm finding the performance be very slow.
Details:
- One of the sets has 150,000 numbers in it
- The intersection of that set and another set takes about 300ms the first time, and about 5000ms the second time
- I haven't done any profiling yet, but every time I break the debugger while doing the intersection its in malloc.c!
So, how can I improve this performance? Switch to a different data structure? Some how improve the memory allocation performance of map?
Update:
Is there any way to ask std::map or
boost::unordered_map to pre-allocate
some space?
Or, are there any tips for using these efficiently?
Update2:
See Fast C++ container like the C# HashSet<T> and Dictionary<K,V>?
Update3:
I benchmarked set_intersection and got horrible results:
(set_intersection) Found 313 values in the intersection, in 11345ms
(set_intersection) Found 309 values in the intersection, in 12332ms
Code:
int runIntersectionTestAlgo()
{
set<int> set1;
set<int> set2;
set<int> intersection;
// Create 100,000 values for set1
for ( int i = 0; i < 100000; i++ )
{
int value = 1000000000 + i;
set1.insert(value);
}
// Create 1,000 values for set2
for ( int i = 0; i < 1000; i++ )
{
int random = rand() % 200000 + 1;
random *= 10;
int value = 1000000000 + random;
set2.insert(value);
}
set_intersection(set1.begin(),set1.end(), set2.begin(), set2.end(), inserter(intersection, intersection.end()));
return intersection.size();
}
You should definitely be using preallocated vectors which are way faster. The problem with doing set intersection with stl sets is that each time you move to the next element you're chasing a dynamically allocated pointer, which could easily not be in your CPU caches. With a vector the next element will often be in your cache because it's physically close to the previous element.
The trick with vectors, is that if you don't preallocate the memory for a task like this, it'll perform EVEN WORSE because it'll go on reallocating memory as it resizes itself during your initialization step.
Try something like this instaed - it'll be WAY faster.
int runIntersectionTestAlgo() {
vector<char> vector1; vector1.reserve(100000);
vector<char> vector2; vector2.reserve(1000);
// Create 100,000 values for set1
for ( int i = 0; i < 100000; i++ ) {
int value = 1000000000 + i;
set1.push_back(value);
}
sort(vector1.begin(), vector1.end());
// Create 1,000 values for set2
for ( int i = 0; i < 1000; i++ ) {
int random = rand() % 200000 + 1;
random *= 10;
int value = 1000000000 + random;
set2.push_back(value);
}
sort(vector2.begin(), vector2.end());
// Reserve at most 1,000 spots for the intersection
vector<char> intersection; intersection.reserve(min(vector1.size(),vector2.size()));
set_intersection(vector1.begin(), vector1.end(),vector2.begin(), vector2.end(),back_inserter(intersection));
return intersection.size();
}
Without knowing any more about your problem, "check with a good profiler" is the best general advise I can give. Beyond that...
If memory allocation is your problem, switch to some sort of pooled allocator that reduces calls to malloc. Boost has a number of custom allocators that should be compatible with std::allocator<T>. In fact, you may even try this before profiling, if you've already noticed debug-break samples always ending up in malloc.
If your number-space is known to be dense, you can switch to using a vector- or bitset-based implementation, using your numbers as indexes in the vector.
If your number-space is mostly sparse but has some natural clustering (this is a big if), you may switch to a map-of-vectors. Use higher-order bits for map indexing, and lower-order bits for vector indexing. This is functionally very similar to simply using a pooled allocator, but it is likely to give you better caching behavior. This makes sense, since you are providing more information to the machine (clustering is explicit and cache-friendly, rather than a random distribution you'd expect from pool allocation).
I would second the suggestion to sort them. There are already STL set algorithms that operate on sorted ranges (like set_intersection, set_union, etc):
set_intersection
I don't understand why you have to use a map to do intersection. Like people have said, you could put the sets in std::set's, and then use std::set_intersection().
Or you can put them into hash_set's. But then you would have to implement intersection manually: technically you only need to put one of the sets into a hash_set, and then loop through the other one, and test if each element is contained in the hash_set.
Intersection with maps are slow, try a hash_map. (however, this is not provided in all STL implementation.
Alternatively, sort both map and do it in a merge-sort-like way.
What is your intersection algorithm? Maybe there are some improvements to be made?
Here is an alternate method
I do not know it to be faster or slower, but it could be something to try. Before doing so, I also recommend using a profiler to ensure you really are working on the hotspot. Change the sets of numbers you are intersecting to use std::set<int> instead. Then iterate through the smallest one looking at each value you find. For each value in the smallest set, use the find method to see if the number is present in each of the other sets (for performance, search from smallest to largest).
This is optimised in the case that the number is not found in all of the sets, so if the intersection is relatively small, it may be fast.
Then, store the intersection in std::vector<int> instead - insertion using push_back is also very fast.
Here is another alternate method
Change the sets of numbers to std::vector<int> and use std::sort to sort from smallest to largest. Then use std::binary_search to find the values, using roughly the same method as above. This may be faster than searching a std::set since the array is more tightly packed in memory. Actually, never mind that, you can then just iterate through the values in lock-step, looking at the ones with the same value. Increment only the iterators which are less than the minimum value you saw at the previous step (if the values were different).
Might be your algorithm. As I understand it, you are spinning over each set (which I'm hoping is a standard set), and throwing them into yet another map. This is doing a lot of work you don't need to do, since the keys of a standard set are in sorted order already. Instead, take a "merge-sort" like approach. Spin over each iter, dereferencing to find the min. Count the number that have that min, and increment those. If the count was N, add it to the intersection. Repeat until the first map hits it's end (If you compare the sizes before starting, you won't have to check every map's end each time).
Responding to update: There do exist faculties to speed up memory allocation by pre-reserving space, like boost::pool_alloc. Something like:
std::map<int, int, std::less<int>, boost::pool_allocator< std::pair<int const, int> > > m;
But honestly, malloc is pretty good at what it does; I'd profile before doing anything too extreme.
Look at your algorithms, then choose the proper data type. If you're going to have set-like behaviour, and want to do intersections and the like, std::set is the container to use.
Since it's elements are stored in a sorted way, insertion may cost you O(log N), but intersection with another (sorted!) std::set can be done in linear time.
I figured something out: if I attach the debugger to either RELEASE or DEBUG builds (e.g. hit F5 in the IDE), then I get horrible times.