Idea is very easy, we have two hash tables. One is responsible to map a value to index while another one is to map index to value. I swap the element deleted to the end of idx before delete it. In this way, the index is always 0 ~ n -1 and we can easily make a random get to them.

```
class RandomizedSet {
public:
/** Initialize your data structure here. */
unordered_map<int, int> valToIdxs;
unordered_map<int, int> idxToVals;
int n = 0; // current size
RandomizedSet() {
valToIdxs.clear();
idxToVals.clear();
n = 0;
}
/** Inserts a value to the set. Returns true if the set did not already contain the specified element. */
bool insert(int val) {
if (valToIdxs.count(val) == 0) {
valToIdxs[val] = n;
idxToVals[n++] = val;
return true;
}
return false;
}
/** Removes a value from the set. Returns true if the set contained the specified element. */
bool remove(int val) {
if (valToIdxs.count(val) > 0) {
int idx = valToIdxs[val];
int last = n - 1;
valToIdxs[idxToVals[last]] = idx;
idxToVals[idx] = idxToVals[last]; // change idx of last element to idx
valToIdxs.erase(valToIdxs.find(val));
idxToVals.erase(idxToVals.find(last));
n --;
return true;
}
return false;
}
/** Get a random element from the set. */
int getRandom() {
int idx = rand() % n;
return idxToVals[idx];
}
};
```