Specs

fetch_solution((... -> float()), [Meta.Problem.Variable.t()]) ::
  Meta.Problem.Solution.t()

Applies the objective function with variables and yields a Solution struct with the function result and the variables used, useful for logging or history registry.

# Examples

iex> fetch_solution(&Meta.Functions.simple_example/1, [%Meta.Problem.Variable{name: :x, value: 3, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 100}}])
%Meta.Problem.Solution{value: 9, variables: [%Meta.Problem.Variable{name: :x, value: 3, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 100}}]}

iex> fetch_solution(&Meta.Functions.example_one/2, [%Meta.Problem.Variable{name: :x, value: 10, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 100}}, %Meta.Problem.Variable{name: :y, value: 4, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 100}}])
%Meta.Problem.Solution{value: 32.99060735569487, variables: [%Meta.Problem.Variable{name: :x, value: 10, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 100}}, %Meta.Problem.Variable{name: :y, value: 4, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 100}}]}

Specs

is_better?(
  Meta.Problem.Solution.t(),
  Meta.Problem.Solution.t(),
  :minimization | :maximization
) :: boolean()

Compares two solutions and returns if the first is better than the second, based on the problem's type (:minimization or :maximization).

# Examples

iex> is_better?(%Meta.Problem.Solution{value: 2, variables: []}, %Meta.Problem.Solution{value: 1, variables: []}, :minimization)
false

iex> is_better?(%Meta.Problem.Solution{value: 2, variables: []}, %Meta.Problem.Solution{value: 1, variables: []}, :maximization)
true

Specs

maybe_generate_random_variable_value(Meta.Problem.Variable.t()) ::
  Meta.Problem.Variable.t()

Seeds a variable value field if its nil, generating a random variable value respecting its constraint definitions. If the variable already has a value, nothing is changed, hence the maybe name.

# Examples

iex> :rand.seed(:exsplus, {101, 102, 103})
iex> maybe_generate_random_variable_value(%Meta.Problem.Variable{name: :x, value: nil, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}})
%Meta.Problem.Variable{name: :x, value: 1.489738081165117, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}

iex> maybe_generate_random_variable_value(%Meta.Problem.Variable{name: :x, value: 1, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}})
%Meta.Problem.Variable{name: :x, value: 1, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}

Tweak a variable value, to a maximum of a matching variable name in noise_sizes. The constraints of the variable are respected, so the maximum delta change may be lower than the noise_size found.

The variable is only tweaked if a probability check passes (a random float between 0 and 1 is lower than the probability argument).

It raises an Error if the probability is not between 0 and 1 # Examples

iex> :rand.seed(:exsplus, {101, 102, 103})
iex> maybe_tweak_variable(%Meta.Problem.Variable{name: :x, value: 1, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}, 1, [x: 0.25])
%Meta.Problem.Variable{name: :x, value: 0.978631679333307, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}

iex> maybe_tweak_variable(%Meta.Problem.Variable{name: :x, value: 1, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}, 0, [x: 0.25])
%Meta.Problem.Variable{name: :x, value: 1, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}

iex> :rand.seed(:exsplus, {101, 102, 103})
iex> maybe_tweak_variable(%Meta.Problem.Variable{name: :x, value: 1, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}, 1, [x: 5000])
%Meta.Problem.Variable{name: :x, value: 0.9145267173332277, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}

Specs

serialize(Meta.Problem.Solution.t()) :: map()

Flattens a solution map to a one-dimensional for serialiation purpose.

The key value of solution is passed on along each variable name as key and its corresponding values as value.

Examples

iex> serialize(%Meta.Problem.Solution{value: 5, variables: [%Meta.Problem.Variable{name: :x, value: 0.25}, %Meta.Problem.Variable{name: :y, value: 0.5}]})
%{value: 5, x: 0.25, y: 0.5}

Specs

tweak_variable(Meta.Problem.Variable.t(), float()) :: Meta.Problem.Variable.t()

Tweak a variable value, to a maximum of a noise_size. The constraints of the variable are respected, so the maximum delta change may be lower than the noise_size found.

# Examples

iex> :rand.seed(:exsplus, {101, 102, 103})
iex> tweak_variable(%Meta.Problem.Variable{name: :x, value: 1, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}, 0.25)
%Meta.Problem.Variable{name: :x, value: 1.1224345202912793, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}

iex> :rand.seed(:exsplus, {101, 102, 103})
iex> tweak_variable(%Meta.Problem.Variable{name: :x, value: 1, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}, 5000)
%Meta.Problem.Variable{name: :x, value: 1.489738081165117, constraint: %Meta.Problem.Constraint{lower_boundary: 0, higher_boundary: 2}}