Tuning parameters for the relativistic portfolio optimizer. More...

Public Attributes
int	max_iterations = 1000
	Maximum number of projected gradient descent iterations.

double	step_size = 1e-3
	Step size for gradient descent (learning rate).

double	convergence_tol = 1e-8
	Convergence tolerance: stop when \|\|w_{k+1} − w_k\|\|₂ < tol.

double	regularisation = 1e-6

double	c_market = constants::SPEED_OF_INFORMATION
	Speed-of-information parameter for Lorentz factor computation.

Detailed Description

Tuning parameters for the relativistic portfolio optimizer.

Definition at line 55 of file relativistic_optimizer.hpp.

Member Data Documentation

double srfm::portfolio::OptimizerConfig::c_market = constants::SPEED_OF_INFORMATION

Speed-of-information parameter for Lorentz factor computation.

Definition at line 70 of file relativistic_optimizer.hpp.

double srfm::portfolio::OptimizerConfig::convergence_tol = 1e-8

Convergence tolerance: stop when ||w_{k+1} − w_k||₂ < tol.

Definition at line 63 of file relativistic_optimizer.hpp.

int srfm::portfolio::OptimizerConfig::max_iterations = 1000

Maximum number of projected gradient descent iterations.

Definition at line 57 of file relativistic_optimizer.hpp.

double srfm::portfolio::OptimizerConfig::regularisation = 1e-6

Tikhonov regularisation strength added to the geodesic metric. Prevents degenerate geodesic distances when assets are nearly collinear.

Definition at line 67 of file relativistic_optimizer.hpp.

double srfm::portfolio::OptimizerConfig::step_size = 1e-3

Step size for gradient descent (learning rate).

Definition at line 60 of file relativistic_optimizer.hpp.

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