gtopt::FlatHelper class
#include <gtopt/flat_helper.hpp>

Converts multi-dimensional optimization data into flat vectors for LP formulation.

This helper class transforms scenario/stage/block indexed data into:

  • Flat vectors for LP matrix construction
  • Valid/invalid markers for sparse problems
  • Scaled values using provided factors

Key features:

  • Handles active element filtering (only processes active scenarios/stages/blocks)
  • Applies scaling factors during flattening
  • Supports multiple index holder types (GSTB, STB, ST, T)
  • Provides constexpr accessors for active elements
  • Thread-safe operations (all methods are const)

Derived classes

class SystemContext

Constructors, destructors, conversion operators

FlatHelper(const SimulationLP& simulation) explicit

Public functions

auto active_block_count() const -> constexpr size_t noexcept
auto active_blocks() const -> constexpr const auto& noexcept
auto active_scenario_count() const -> constexpr size_t noexcept
auto active_scenarios() const -> constexpr const auto& noexcept
auto active_stage_blocks() const -> constexpr const auto& noexcept
auto active_stage_count() const -> constexpr size_t noexcept
auto active_stages() const -> constexpr const auto& noexcept
template<typename Projection Projection, typename Value Value = Index, typename Factor Factor = block_factor_matrix_t>
auto flat(const GSTBIndexHolder<Value>& hstb, Projection proj, const Factor& factor = Factor()) const -> constexpr auto
Flattens GSTB-indexed data into vectors with optional scaling.
template<typename Projection Projection, typename Value Value = Index, typename Factor Factor = block_factor_matrix_t>
auto flat(const STBIndexHolder<Value>& hstb, Projection proj, const Factor& factor, const STIndexHolder<double>& st_scale) const -> constexpr auto
template<typename Projection Projection, typename Value Value = Index, typename Factor Factor = block_factor_matrix_t>
auto flat(const STBIndexHolder<Value>& hstb, Projection proj, const Factor& factor = Factor()) const -> constexpr auto
template<typename Projection Projection, typename Value Value = Index, typename Factor Factor = scenario_stage_factor_matrix_t>
auto flat(const STIndexHolder<Value>& hst, Projection proj, const Factor& factor = Factor()) const -> constexpr auto
template<typename Projection Projection = std::identity, typename Value Value = Index, typename Factor Factor = stage_factor_matrix_t>
auto flat(const TIndexHolder<Value>& ht, Projection proj = {}, const Factor& factor = {}) const -> constexpr auto
auto is_first_scenario(const ScenarioUid& scenario_uid) const -> constexpr bool noexcept
auto is_first_stage(const StageLP& stage) const -> constexpr bool noexcept
auto is_first_stage(const StageUid& stage_uid) const -> constexpr bool noexcept
auto is_last_stage(const StageUid& stage_uid) const -> constexpr bool noexcept
auto prev_stage(const StageLP& stage) const -> constexpr auto noexcept
auto simulation() const -> constexpr const SimulationLP& noexcept
auto st_uids() const -> STUids
auto stb_uids() const -> STBUids
auto t_uids() const -> TUids

Function documentation

template<typename Projection Projection, typename Value Value = Index, typename Factor Factor = block_factor_matrix_t>
constexpr auto gtopt::FlatHelper::flat(const GSTBIndexHolder<Value>& hstb, Projection proj, const Factor& factor = Factor()) const

Flattens GSTB-indexed data into vectors with optional scaling.

Parameters
hstb The GSTB-indexed data holder
proj Projection function to apply to each value
factor Optional scaling factors (applied after projection)
Returns Pair of (values, valid) vectors

Processes a 3D (Scenario/Stage/Block) indexed container into:

  • Values vector with projected/scaled values
  • Valid vector marking which indices had data

template<typename Projection Projection, typename Value Value = Index, typename Factor Factor = block_factor_matrix_t>
constexpr auto gtopt::FlatHelper::flat(const STBIndexHolder<Value>& hstb, Projection proj, const Factor& factor, const STIndexHolder<double>& st_scale) const

flat() overload that applies an additional per-(scenario,stage) scale factor on top of the block-level factor. Used by StorageLP::add_to_output to back-scale volume-balance duals when the daily-cycle option was active for a stage (scale = 24/stage_dur).

The st_scale lookup is performed once per (scenario,stage) pair, outside the block loop, so there is no per-block overhead. When a (suid,tuid) key is absent from st_scale the scale defaults to 1.0 (i.e. no scaling is applied for that stage), making this a safe drop-in for the plain flat() call when st_scale is empty.