[WIP] Dynamic

include/hibf/config.hpp

@@ -40,6 +40,7 @@ namespace seqan::hibf
  * | General | seqan::hibf::config::threads                  | 1       | [RECOMMENDED_TO_ADAPT] |
  * | Layout  | seqan::hibf::config::sketch_bits              | 12      |                        |
  * | Layout  | seqan::hibf::config::tmax                     | 0       | 0 indicates unset      |
+ * | Layout  | seqan::hibf::config::empty_bin_fraction       | 0.0     | Dynamic Layout         |
  * | Layout  | seqan::hibf::config::max_rearrangement_ratio  | 0.5     |                        |
  * | Layout  | seqan::hibf::config::alpha                    | 1.2     |                        |
  * | Layout  | seqan::hibf::config::disable_estimate_union   | false   |                        |
@@ -230,6 +231,9 @@ struct config
      */
     size_t tmax{};
 
+    //!\brief The percentage of empty bins in the layout.
+    double empty_bin_fraction{};
+
     /*!\brief A scaling factor to influence the amount of merged bins produced by the layout algorithm.
      *
      * The layout algorithm optimizes the space consumption of the resulting HIBF, but currently has no means of
@@ -302,6 +306,7 @@ struct config
      *   * seqan::hibf::config::threads must be greater than `0`.
      *   * seqan::hibf::config::sketch_bits must be in `[5,32]`.
      *   * seqan::hibf::config::tmax must be at most `18446744073709551552`.
+     *   * seqan::hibf::config::empty_bin_fraction must be in `[0.0,1.0)`.
      *   * seqan::hibf::config::alpha must be positive.
      *   * seqan::hibf::config::max_rearrangement_ratio must be in `[0.0,1.0]`.
      *
@@ -324,21 +329,26 @@ struct config
                threads == other.threads &&
                sketch_bits == other.sketch_bits &&
                tmax == other.tmax &&
+               empty_bin_fraction == other.empty_bin_fraction &&
                alpha == other.alpha &&
                max_rearrangement_ratio == other.max_rearrangement_ratio &&
                disable_estimate_union == other.disable_estimate_union &&
                disable_rearrangement == other.disable_rearrangement;
         // clang-format on
     }
 
+    bool validated{false};
 private:
     friend class cereal::access;
 
+    static constexpr uint32_t version{2};
+
+
     template <typename archive_t>
     void serialize(archive_t & archive)
     {
-        uint32_t version{1};
-        archive(CEREAL_NVP(version));
+        uint32_t parsed_version{version};
+        archive(cereal::make_nvp("version", parsed_version));
 
         archive(CEREAL_NVP(number_of_user_bins));
         archive(CEREAL_NVP(number_of_hash_functions));
@@ -348,10 +358,16 @@ struct config
 
         archive(CEREAL_NVP(sketch_bits));
         archive(CEREAL_NVP(tmax));
+
+        if (parsed_version > 1u)
+            archive(CEREAL_NVP(empty_bin_fraction));
+
         archive(CEREAL_NVP(alpha));
         archive(CEREAL_NVP(max_rearrangement_ratio));
         archive(CEREAL_NVP(disable_estimate_union));
         archive(CEREAL_NVP(disable_rearrangement));
+        if (parsed_version > 1u)
+            archive(CEREAL_NVP(validated));
     }
 };
 

include/hibf/hierarchical_interleaved_bloom_filter.hpp

@@ -33,6 +33,8 @@ namespace bin_kind
 
 //!\brief The value that indicates a merged bin.
 static constexpr uint64_t merged{std::numeric_limits<uint64_t>::max()};
+//!\brief The value that indicates a deleted bin.
+static constexpr uint64_t deleted{std::numeric_limits<uint64_t>::max() - 1u};
 
 } // namespace bin_kind
 
@@ -212,6 +214,21 @@ class hierarchical_interleaved_bloom_filter
      */
     std::vector<std::vector<uint64_t>> next_ibf_id;
 
+    struct previous_ibf_id_pair
+    {
+        size_t ibf_idx{};
+        size_t bin_idx{};
+
+        template <seqan::hibf::cereal_archive archive_t>
+        void CEREAL_SERIALIZE_FUNCTION_NAME(archive_t & archive)
+        {
+            archive(ibf_idx);
+            archive(bin_idx);
+        }
+    };
+
+    std::vector<previous_ibf_id_pair> prev_ibf_id;
+
     /*!\brief Stores for each bin in each IBF of the HIBF the user bin ID.
     * \details
     * Assume we look up a bin `b` in IBF `i`, i.e. `ibf_bin_to_user_bin_id[i][b]`.
@@ -251,6 +268,7 @@ class hierarchical_interleaved_bloom_filter
         archive(ibf_vector);
         archive(next_ibf_id);
         archive(ibf_bin_to_user_bin_id);
+        archive(prev_ibf_id);
     }
 
     /*!\name Timer

include/hibf/interleaved_bloom_filter.hpp

@@ -190,7 +190,7 @@ class interleaved_bloom_filter : private seqan::hibf::bit_vector
 
     //!\brief Helper function to reduce code-duplication between emplace and emplace_exists.
     template <bool check_exists>
-    inline auto emplace_impl(size_t const value, bin_index const bin) noexcept;
+    inline void emplace_impl(size_t const value, bin_index const bin) noexcept;
 
 public:
     class membership_agent_type; // documented upon definition below
@@ -249,7 +249,7 @@ class interleaved_bloom_filter : private seqan::hibf::bit_vector
      * \returns `true` if the value already existed, `false` otherwise.
      * \sa seqan::hibf::interleaved_bloom_filter::emplace
     */
-    [[nodiscard]] bool emplace_exists(size_t const value, bin_index const bin) noexcept;
+    void emplace_exists(size_t const value, bin_index const bin) noexcept;
 
     /*!\brief Clears a specific bin.
      * \param[in] bin The bin index to clear.
@@ -281,14 +281,16 @@ class interleaved_bloom_filter : private seqan::hibf::bit_vector
                       "The reference type of the range to clear must be seqan::hibf::bin_index.");
 #ifndef NDEBUG
         for (auto && bin : bin_range)
-            assert(bin.value < bins);
+            assert(bin.value < technical_bins);
 #endif // NDEBUG
 
         for (size_t offset = 0, i = 0; i < bin_size_; offset += technical_bins, ++i)
             for (auto && bin : bin_range)
                 (*this)[bin.value + offset] = 0;
     }
 
+    bool set_bin_count(bin_count const new_bin_count);
+
     /*!\brief Increases the number of bins stored in the Interleaved Bloom Filter.
      * \param[in] new_bins_ The new number of bins.
      * \throws std::invalid_argument If passed number of bins is smaller than current number of bins.
@@ -311,7 +313,7 @@ class interleaved_bloom_filter : private seqan::hibf::bit_vector
      *
      * \include test/snippet/ibf/interleaved_bloom_filter_increase_bin_number_to.cpp
      */
-    void increase_bin_number_to(bin_count const new_bins_);
+    void increase_bin_number_to(bin_count const new_bin_count);
     //!\}
 
     /*!\name Lookup
@@ -367,6 +369,11 @@ class interleaved_bloom_filter : private seqan::hibf::bit_vector
         return bins;
     }
 
+    void overwrite_bin_count(size_t const new_count) noexcept
+    {
+        bins = new_count;
+    }
+
     /*!\brief Returns the size of a single bin that the Interleaved Bloom Filter manages.
      * \returns The size in bits of a single bin.
      */
@@ -403,6 +410,9 @@ class interleaved_bloom_filter : private seqan::hibf::bit_vector
     using base_t::data;
     //!\}
 
+    std::vector<size_t> occupancy{};
+    bit_vector occupied_bins{};
+
     /*!\cond DEV
      * \brief Serialisation support function.
      * \tparam archive_t Type of `archive`; must satisfy seqan::hibf::cereal_archive.
@@ -420,6 +430,8 @@ class interleaved_bloom_filter : private seqan::hibf::bit_vector
         archive(bin_words);
         archive(hash_funs);
         archive(cereal::base_class<base_t>(this));
+        archive(occupancy);
+        archive(occupied_bins);
     }
     //!\endcond
 };

include/hibf/layout/hierarchical_binning.hpp

@@ -68,7 +68,8 @@ class hierarchical_binning
         config{config_},
         data{std::addressof(data_)},
         num_user_bins{data->positions.size()},
-        num_technical_bins{data->previous.empty() ? config.tmax : needed_technical_bins(num_user_bins)}
+        num_technical_bins{data->previous.empty() ? config.tmax
+                                                  : std::min<size_t>(needed_technical_bins(num_user_bins), config.tmax)}
     {
         assert(data != nullptr);
     }

include/hibf/misc/empty_bins_by_fraction.hpp

@@ -0,0 +1,26 @@
+// SPDX-FileCopyrightText: 2006-2024, Knut Reinert & Freie Universität Berlin
+// SPDX-FileCopyrightText: 2016-2024, Knut Reinert & MPI für molekulare Genetik
+// SPDX-License-Identifier: BSD-3-Clause
+
+#pragma once
+
+#include <algorithm>
+#include <cstddef>
+
+#include <hibf/platform.hpp>
+
+namespace seqan::hibf
+{
+
+/*!\brief Returns the number of empty bins that should be created by a given fraction of the total number of bins.
+ * \param[in] tmax     The total number of bins.
+ * \param[in] fraction The fraction of the total number of bins that should be empty.
+ * \ingroup hibf
+ * \sa https://godbolt.org/z/cMjbM39vj
+ */
+[[nodiscard]] constexpr size_t empty_bins_by_fraction(size_t const tmax, double const fraction) noexcept
+{
+    return std::clamp<size_t>(tmax * fraction, 1, tmax - 2) - (fraction == 0.0);
+}
+
+} // namespace seqan::hibf

src/build/insert_into_ibf.cpp

@@ -41,7 +41,7 @@ void insert_into_ibf(robin_hood::unordered_flat_set<uint64_t> const & kmers,
         seqan::hibf::bin_index const bin_idx{bin_index + chunk_number};
         ++chunk_number;
         for (size_t const value : chunk)
-            ibf.emplace(value, bin_idx);
+            ibf.emplace_exists(value, bin_idx);
     }
     local_fill_ibf_timer.stop();
     fill_ibf_timer += local_fill_ibf_timer;
@@ -63,7 +63,7 @@ void insert_into_ibf(build_data const & data,
     serial_timer local_fill_ibf_timer{};
     local_fill_ibf_timer.start();
     for (auto && value : values)
-        ibf.emplace(value, bin_index);
+        ibf.emplace_exists(value, bin_index);
     local_fill_ibf_timer.stop();
     data.fill_ibf_timer += local_fill_ibf_timer;
 }

src/config.cpp

@@ -14,9 +14,10 @@
 #include <cereal/archives/json.hpp> // for JSONInputArchive, JSONOutputArchive
 #include <cereal/cereal.hpp>        // for make_nvp, InputArchive, OutputArchive
 
-#include <hibf/config.hpp>                   // for config
-#include <hibf/layout/prefixes.hpp>          // for meta_header, meta_hibf_config_end, meta_hibf_config_start
-#include <hibf/misc/next_multiple_of_64.hpp> // for next_multiple_of_64
+#include <hibf/config.hpp>                      // for config
+#include <hibf/layout/prefixes.hpp>             // for meta_header, meta_hibf_config_end, meta_hibf_config_start
+#include <hibf/misc/empty_bins_by_fraction.hpp> // for empty_bins_by_fraction
+#include <hibf/misc/next_multiple_of_64.hpp>    // for next_multiple_of_64
 
 namespace seqan::hibf
 {
@@ -63,6 +64,9 @@ void config::write_to(std::ostream & stream) const
 
 void config::validate_and_set_defaults()
 {
+    if (validated)
+        return;
+
     if (!input_fn)
         throw std::invalid_argument{"[HIBF CONFIG ERROR] You did not set the required config::input_fn."};
 
@@ -111,6 +115,11 @@ void config::validate_and_set_defaults()
                   << "anyway, so we increased your number of technical bins to " << tmax << ".\n";
     }
 
+    if (empty_bin_fraction < 0.0 || empty_bin_fraction >= 1.0)
+        throw std::invalid_argument{"[HIBF CONFIG ERROR] config::empty_bin_fraction must be in [0.0,1.0)."};
+
+    tmax -= empty_bins_by_fraction(tmax, empty_bin_fraction);
+
     if (alpha < 0.0)
         throw std::invalid_argument{"[HIBF CONFIG ERROR] config::alpha must be positive."};
 
@@ -119,6 +128,8 @@ void config::validate_and_set_defaults()
 
     if (disable_estimate_union || max_rearrangement_ratio == 0.0)
         disable_rearrangement = true;
+
+    validated = true;
 }
 
 } // namespace seqan::hibf

src/hierarchical_interleaved_bloom_filter.cpp

@@ -42,7 +42,8 @@ size_t hierarchical_build(hierarchical_interleaved_bloom_filter & hibf,
                           robin_hood::unordered_flat_set<uint64_t> & parent_kmers,
                           layout::graph::node const & current_node,
                           build::build_data & data,
-                          bool is_root)
+                          bool is_root,
+                          size_t const parent_ibf_idx = 0u)
 {
     size_t const ibf_pos{data.request_ibf_idx()};
 
@@ -68,7 +69,8 @@ size_t hierarchical_build(hierarchical_interleaved_bloom_filter & hibf,
                                    kmers,
                                    current_node.children[current_node.favourite_child_idx.value()],
                                    data,
-                                   false);
+                                   false,
+                                   ibf_pos);
             return 1;
         }
         else // max bin is not a merged bin
@@ -124,8 +126,9 @@ size_t hierarchical_build(hierarchical_interleaved_bloom_filter & hibf,
             auto & child = children[index];
 
             robin_hood::unordered_flat_set<uint64_t> kmers{};
-            size_t const ibf_pos = hierarchical_build(hibf, kmers, child, data, false);
+            size_t const ibf_pos = hierarchical_build(hibf, kmers, child, data, false, ibf_pos);
             auto parent_bin_index = child.parent_bin_index;
+            hibf.prev_ibf_id[ibf_pos] = {.ibf_idx = parent_ibf_idx, .bin_idx = parent_bin_index};
             {
                 size_t const mutex_id{parent_bin_index / 64};
                 std::lock_guard<std::mutex> guard{local_ibf_mutex[mutex_id]};
@@ -184,6 +187,7 @@ void build_index(hierarchical_interleaved_bloom_filter & hibf,
 
     hibf.ibf_vector.resize(number_of_ibfs);
     hibf.ibf_bin_to_user_bin_id.resize(number_of_ibfs);
+    hibf.prev_ibf_id.resize(number_of_ibfs);
     hibf.next_ibf_id.resize(number_of_ibfs);
 
     build::build_data data{.config = config, .ibf_graph = {hibf_layout}};