| 16 | } |
| 17 | |
| 18 | static double CalcSampleQuantileBinarySearch( |
| 19 | TConstArrayRef<float> sampleRef, |
| 20 | TConstArrayRef<float> weightsRef, |
| 21 | const double alpha |
| 22 | ) { |
| 23 | constexpr int BINARY_SEARCH_ITERATIONS = 100; |
| 24 | |
| 25 | const double totalWeight = Accumulate(weightsRef, 0.0); |
| 26 | const double needWeight = totalWeight * alpha; |
| 27 | |
| 28 | const auto minMaxSamples = MinMaxElement(sampleRef.begin(), sampleRef.end()); |
| 29 | double lQ = *minMaxSamples.first - DBL_EPSILON; |
| 30 | double rQ = *minMaxSamples.second; |
| 31 | |
| 32 | const size_t sampleSize = sampleRef.size(); |
| 33 | TVector<TValueWithWeight> elements; |
| 34 | elements.yresize(sampleSize); |
| 35 | for (auto i : xrange(sampleSize)) { |
| 36 | elements[i] = {sampleRef[i], weightsRef[i]}; |
| 37 | } |
| 38 | /* |
| 39 | * We will support the following invariant: |
| 40 | * total weight of elements with sample values <= lQ is strictly less than needWeight |
| 41 | * total weight of elements with sample values <= rQ is greater or equal than needWeight |
| 42 | * elements with indices < l are less or equal than q |
| 43 | * elements with indices > r are greater than q |
| 44 | */ |
| 45 | int l = 0, r = sampleSize; |
| 46 | double collectedLeftWeight = 0; |
| 47 | for (auto it : xrange(BINARY_SEARCH_ITERATIONS)) { |
| 48 | Y_UNUSED(it); |
| 49 | const double q = (lQ + rQ) / 2; |
| 50 | auto partitionIt = std::partition( |
| 51 | elements.begin() + l, |
| 52 | elements.begin() + r, |
| 53 | [q](const TValueWithWeight& element) { |
| 54 | return element.Value <= q; |
| 55 | } |
| 56 | ); |
| 57 | const double partitionLeftWeight = Accumulate( |
| 58 | elements.begin() + l, |
| 59 | partitionIt, |
| 60 | 0.0, |
| 61 | [](double sum, const TValueWithWeight& element) { |
| 62 | return sum + element.Weight; |
| 63 | } |
| 64 | ); |
| 65 | const int partitionPoint = partitionIt - elements.begin(); |
| 66 | |
| 67 | if (collectedLeftWeight + partitionLeftWeight < needWeight - DBL_EPSILON) { |
| 68 | l = partitionPoint; |
| 69 | lQ = q; |
| 70 | collectedLeftWeight += partitionLeftWeight; |
| 71 | } else { |
| 72 | r = partitionPoint; |
| 73 | rQ = q; |
| 74 | } |
| 75 | } |
no test coverage detected