#include <minkowski_momentum.hpp>

Public Attributes
double	learning_rate = 0.01
	Step size per iteration.

int	max_iterations = 1000
	Maximum number of ascent steps.

double	tolerance = 1e-8
	Convergence criterion on m²

double	min_weight = 0.0
	Lower bound on each weight.

double	max_weight = 1.0
	Upper bound on each weight.

Detailed Description

Finds portfolio weights that maximise the Minkowski invariant mass (diversification measure) subject to the constraint that weights sum to 1.

The optimisation is performed by a simple gradient-ascent procedure: at each step the weight vector is nudged in the direction of increasing m², then re-normalised to sum to 1 and clamped to [min_weight, max_weight]. Configuration for the gradient-ascent optimiser. Defined at namespace scope (and aliased as MomentumPortfolioOptimizer::Config) because GCC rejects a = {} default argument of a nested class with default member initialisers.

Definition at line 161 of file minkowski_momentum.hpp.

Member Data Documentation

◆ learning_rate

double srfm::minkowski_momentum::MomentumOptimizerConfig::learning_rate = 0.01

Step size per iteration.

Definition at line 162 of file minkowski_momentum.hpp.

◆ max_iterations

int srfm::minkowski_momentum::MomentumOptimizerConfig::max_iterations = 1000

Maximum number of ascent steps.

Definition at line 163 of file minkowski_momentum.hpp.

◆ max_weight

double srfm::minkowski_momentum::MomentumOptimizerConfig::max_weight = 1.0

Upper bound on each weight.

Definition at line 166 of file minkowski_momentum.hpp.

◆ min_weight

double srfm::minkowski_momentum::MomentumOptimizerConfig::min_weight = 0.0

Lower bound on each weight.

Definition at line 165 of file minkowski_momentum.hpp.

◆ tolerance

double srfm::minkowski_momentum::MomentumOptimizerConfig::tolerance = 1e-8

Convergence criterion on m²

Definition at line 164 of file minkowski_momentum.hpp.

The documentation for this struct was generated from the following file:

include/srfm/minkowski_momentum.hpp