Merge pull request #225 from smehringer/replace_function_bit_size

[MISC] Use seqan::hibf::build::bin_size_in_bits
seqan · Oct 26, 2023 · ec33f79 · ec33f79
2 parents 7767b99 + 9dbf172
commit ec33f79
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 33 deletions.
diff --git a/include/chopper/layout/hibf_statistics.hpp b/include/chopper/layout/hibf_statistics.hpp
@@ -133,18 +133,6 @@ class hibf_statistics
     //!\brief The gathered summary of statistics for each level of this HIBF.
     std::map<size_t, level_summary> summaries;
 
-    /*!\brief Computes the bin size in bits.
-    *
-    * -NUM_ELEM*HASHES
-    * ----------------------  = SIZE
-    * LN(1-FPR^(1/HASHES))
-    *
-    * -NUM_ELEMS*HASHES
-    * -----------------------
-    * LN(1 - e^(LN(FPR) / HASHES) )
-    */
-    size_t compute_bin_size(size_t const number_of_kmers_to_be_stored) const;
-
     /*!\brief Compute the Bloom Filter size from `number_of_kmers_to_be_stored` and
      *        return it as a formatted string with the appropriate unit.
      * \param[in] number_of_kmers_to_be_stored

diff --git a/src/layout/hibf_statistics.cpp b/src/layout/hibf_statistics.cpp
@@ -32,6 +32,7 @@
 #include <chopper/layout/hibf_statistics.hpp>
 #include <chopper/layout/ibf_query_cost.hpp>
 
+#include <hibf/build/bin_size_in_bits.hpp>
 #include <hibf/contrib/robin_hood.hpp>
 #include <hibf/layout/compute_fpr_correction.hpp>
 #include <hibf/layout/layout.hpp>
@@ -258,7 +259,12 @@ size_t hibf_statistics::total_hibf_size_in_byte()
         total_size += std::reduce(summary.ibf_mem_size.begin(), summary.ibf_mem_size.end());
     }
 
-    return compute_bin_size(total_size) / 8;
+    size_t const size_in_bits =
+        seqan::hibf::build::bin_size_in_bits({.fpr = config.hibf_config.maximum_false_positive_rate,
+                                              .hash_count = config.hibf_config.number_of_hash_functions,
+                                              .elements = total_size});
+
+    return size_in_bits / 8;
 }
 
 //!\brief Round bytes to the appropriate unit and convert to string with unit.
@@ -340,18 +346,13 @@ size_t hibf_statistics::total_hibf_size_in_byte()
     return result;
 }
 
-size_t hibf_statistics::compute_bin_size(size_t const number_of_kmers_to_be_stored) const
-{
-    return std::ceil(-static_cast<double>(number_of_kmers_to_be_stored * config.hibf_config.number_of_hash_functions)
-                     / std::log(1
-                                - std::exp(std::log(config.hibf_config.maximum_false_positive_rate)
-                                           / config.hibf_config.number_of_hash_functions)));
-}
-
 std::string hibf_statistics::to_formatted_BF_size(size_t const number_of_kmers_to_be_stored) const
 {
-    size_t const size_in_bytes = compute_bin_size(number_of_kmers_to_be_stored) / 8;
-    return byte_size_to_formatted_str(size_in_bytes);
+    size_t const size_in_bits =
+        seqan::hibf::build::bin_size_in_bits({.fpr = config.hibf_config.maximum_false_positive_rate,
+                                              .hash_count = config.hibf_config.number_of_hash_functions,
+                                              .elements = number_of_kmers_to_be_stored});
+    return byte_size_to_formatted_str(size_in_bits / 8);
 }
 
 void hibf_statistics::collect_bins()

diff --git a/test/benchmark/benchmark_data/query_cost/bin_influence_benchmark.cpp b/test/benchmark/benchmark_data/query_cost/bin_influence_benchmark.cpp
@@ -22,6 +22,8 @@
 
 #include <raptor/adjust_seed.hpp>
 
+#include <hibf/build/bin_size_in_bits.hpp>
+
 #define USE_UNIT_TEST_PARAMETERS 0
 
 static constexpr size_t operator""_MiB(unsigned long long int number)
@@ -73,14 +75,6 @@ static std::vector<std::vector<seqan3::dna4>> const reads{
 
 using ibf_t = seqan3::interleaved_bloom_filter<seqan3::data_layout::uncompressed>;
 
-static constexpr size_t compute_bin_size(size_t const max_bin_size, double const fpr)
-{
-    double const numerator{-static_cast<double>(max_bin_size * hash_num)};
-    double const denominator{std::log(1 - std::exp(std::log(fpr) / hash_num))};
-    double const result{std::ceil(numerator / denominator)};
-    return result;
-}
-
 static std::vector<size_t> cardinality(size_t const bin_count, auto && hash_adaptor)
 {
     std::vector<size_t> cardinalities(bin_count);
@@ -105,7 +99,7 @@ static std::vector<size_t> cardinality(size_t const bin_count, auto && hash_adap
 
 static ibf_t construct_ibf(size_t const bin_count, auto && hash_adaptor, double const fpr)
 {
-    size_t const bin_size{compute_bin_size(std::ranges::max(cardinality(bin_count, hash_adaptor)), fpr)};
+    size_t const bin_size{seqan::hibf::build::bin_size_in_bits({.fpr = fpr, .hash_count = hash_num, .elements = cardinality(bin_count, hash_adaptor)});
 
     if (bin_size * bin_count > max_ibf_size)
         throw std::runtime_error{"Resulting IBF would be too big. " + std::to_string(bin_size * bin_count)};
@@ -197,4 +191,3 @@ BENCHMARK_CAPTURE(bulk_count, "0.3125", 0.3125)->RangeMultiplier(2)->Range(64, 6
 #endif
 
 BENCHMARK_MAIN();
-// clang-format on