Helper Functions

get_eigenvalues(ham::EffectiveHamiltonian, prof, local_inds, comm, conf=get_empty_config();
                Nbatch=get_nbatch(conf), rank=0, nranks=1, verbosity=get_verbosity(conf))

Computes eigenvalues and eigenvectors of the Hamiltonian for a set of structures, distributing the computation across MPI ranks.

Arguments

• ham::EffectiveHamiltonian: The effective Hamiltonian object.
• prof: A profiling object that stores timing information for each step.
• local_inds: Indices of the local structures assigned to the current MPI rank.
• comm: The MPI communicator used for parallel execution.
• conf: Configuration object (default: get_empty_config()) containing parameters for diagonalization.
• Nbatch: The batch size for processing structures (default: get_nbatch(conf)).
• rank: The rank of the MPI process (default: 0).
• nranks: Total number of MPI ranks (default: 1).
• verbosity: Level of verbosity for printed output (default: get_verbosity(conf)).

hyper_optimize(param_values, params, comm, conf; rank=0, nranks=1, verbosity=get_verbosity(conf)) -> Float64

Evaluate a given set of hyperparameters by updating a configuration and running an optimization calculation.

Arguments

• param_values::Vector{Float64}: Numerical values for each parameter to be optimized.
• params::Vector{String}: List of parameter keys. Keys can be flat (e.g., "alpha") or hierarchical (e.g., "Ga_alpha").
• comm: MPI communicator.
• conf: Configuration object.
• rank::Int: MPI rank (default = 0).
• nranks::Int: Total number of MPI processes (default = 1).
• verbosity::Int: Verbosity level (default = pulled from configuration).

Returns

• Float64: The minimum training loss obtained from the optimization calculation.