Struct SimplifiedSmoothedConvexOptimization

pub struct SimplifiedSmoothedConvexOptimization<'a, T, C, D> {
    pub d: i32,
    pub t_end: i32,
    pub bounds: Vec<T>,
    pub switching_cost: Vec<f64>,
    pub hitting_cost: CostFn<'a, Config<T>, C, D>,
}

Simplified Smoothed Convex Optimization (SSCO).

• decision space is lower bounded by $\mathbf{0}$
• movement costs are a dimension-dependently scaled Manhattan distance

Fields

d: i32

Number of dimensions.

t_end: i32

Finite, positive time horizon.

bounds: Vec<T>

Vector of upper bounds of each dimension.

switching_cost: Vec<f64>

Vector of positive real constants resembling the switching cost of each dimension.

hitting_cost: CostFn<'a, Config<T>, C, D>

Non-negative convex cost functions.

Implementations

impl<'a, T, C, D> SimplifiedSmoothedConvexOptimization<'a, T, C, D>

where T: Value<'a>, C: ModelOutputSuccess + 'a, D: ModelOutputFailure + 'a,

pub fn into_sco(self) -> SmoothedConvexOptimization<'a, T, C, D>

Convert to an instance of Smoothed Convex Optimization. This assumes that time slots are added after this conversion.

Trait Implementations

impl<'a, T, C, D> BaseProblem for SimplifiedSmoothedConvexOptimization<'a, T, C, D>

where T: Value<'a>, C: Clone, D: Clone,

fn d(&self) -> i32

Number of dimensions.

fn t_end(&self) -> i32

Finite, positive time horizon.

fn set_t_end(&mut self, t_end: i32)

Updates the time horizon.

fn inc_t_end(&mut self)

Increases the time horizon by one time step.

impl<'a, T: Clone, C: Clone, D: Clone> Clone for SimplifiedSmoothedConvexOptimization<'a, T, C, D>

fn clone(&self) -> SimplifiedSmoothedConvexOptimization<'a, T, C, D>

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<'a, T, C, D> Debug for SimplifiedSmoothedConvexOptimization<'a, T, C, D>

where T: Debug,

fn fmt(&self, __f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'a, C, D> DefaultGivenOnlineProblem<f64, SimplifiedSmoothedConvexOptimization<'a, f64, C, D>, C, D> for Memory

where C: ModelOutputSuccess, D: ModelOutputFailure,

fn default( o: &Online<FractionalSimplifiedSmoothedConvexOptimization<'_, C, D>>, ) -> Self

impl<'a, T> Model<T, SimplifiedSmoothedConvexOptimization<'a, T, DataCenterModelOutputSuccess, DataCenterModelOutputFailure>, DataCenterOfflineInput, DataCenterOnlineInput, DataCenterModelOutputSuccess, DataCenterModelOutputFailure> for DataCenterModel

where T: Value<'a>,

fn to( &self, _: DataCenterOfflineInput, ) -> SimplifiedSmoothedConvexOptimization<'a, T, DataCenterModelOutputSuccess, DataCenterModelOutputFailure>

Generates an offline problem instance given some $input$ (with certainty).

fn update( &self, o: &mut Online<SimplifiedSmoothedConvexOptimization<'a, T, DataCenterModelOutputSuccess, DataCenterModelOutputFailure>>, _: DataCenterOnlineInput, )

Performs an online update of the given problem instance $o$ with some $input$ (which may be uncertain).

impl<'a, T, C, D> Problem<T, C, D> for SimplifiedSmoothedConvexOptimization<'a, T, C, D>

where T: Value<'a>, C: ModelOutputSuccess, D: ModelOutputFailure,

fn hit_cost(&self, t: i32, x: Config<T>) -> Cost<C, D>

Hitting cost.

fn movement(&self, prev_x: Config<T>, x: Config<T>, inverted: bool) -> N64

Movement cost.

fn objective_function<'a>(&self, xs: &Schedule<T>) -> Result<Cost<C, D>>

where T: Value<'a>,

Objective function. Calculates the cost of a schedule.

fn inverted_objective_function<'a>( &self, xs: &Schedule<T>, ) -> Result<Cost<C, D>>

where T: Value<'a>,

Inverted Objective function. Calculates the cost of a schedule. Pays the switching cost for powering down rather than powering up.

fn alpha_unfair_objective_function<'a>( &self, xs: &Schedule<T>, alpha: f64, ) -> Result<Cost<C, D>>

where T: Value<'a>,

$\alpha$-unfair Objective function. Calculates the cost of a schedule.

fn objective_function_with_default<'a>( &self, xs: &Schedule<T>, default: &Config<T>, ) -> Result<Cost<C, D>>

where T: Value<'a>,

Objective function starting from an initial configuration other than $\mathbf{0}$.

fn _objective_function_with_default<'a>( &self, xs: &Schedule<T>, default: &Config<T>, alpha: f64, inverted: bool, t_end: i32, ) -> Result<Cost<C, D>>

where T: Value<'a>,

fn total_movement<'a>(&self, xs: &Schedule<T>, inverted: bool) -> Result<N64>

where T: Value<'a>,

Movement in the decision space.

fn _default_config<'a>(&self) -> Config<T>

where T: Value<'a>,

impl<'a, T, C, D> Resettable<'a> for SimplifiedSmoothedConvexOptimization<'a, T, C, D>

where T: Value<'a>, C: ModelOutputSuccess, D: ModelOutputFailure,

fn reset( &'a self, t_start: i32, ) -> SimplifiedSmoothedConvexOptimization<'a, T, C, D>

Shift object to some new initial time $t_start$. Here, $t_start$ is the time before first time slot (e.g., $t_start = 0$ is a no-op).

impl<'a, T, C, D> VerifiableProblem for SimplifiedSmoothedConvexOptimization<'a, T, C, D>

where T: Value<'a>,

fn verify(&self) -> Result<()>