soco 1.0.0

Algorithms for Smoothed Online Convex Optimization
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157

//! Switching cost model.

use crate::model::data_center::model::ServerType;
use noisy_float::prelude::*;
use pyo3::prelude::*;
use std::collections::HashMap;

/// Switching cost model. Parameters are provided separately for each server type.
#[derive(Clone, FromPyObject)]
pub struct SwitchingCostModel(HashMap<String, SwitchingCost>);

/// Switching cost.
#[pyclass]
#[derive(Clone)]
pub struct SwitchingCost {
    /// Average cost per unit of energy.
    #[pyo3(get, set)]
    pub energy_cost: f64,
    /// Power consumed when idling.
    #[pyo3(get, set)]
    pub phi_min: f64,
    /// Power consumed at full load.
    #[pyo3(get, set)]
    pub phi_max: f64,
    /// Additional energy consumed by toggling a server on and off.
    #[pyo3(get, set)]
    pub epsilon: f64,
    /// Required time in time slots for migrating connections or data.
    #[pyo3(get, set)]
    pub delta: f64,
    /// Wear-and-tear costs of toggling a server.
    #[pyo3(get, set)]
    pub tau: f64,
    /// Perceived risk associated with toggling a server.
    #[pyo3(get, set)]
    pub rho: f64,
}
impl SwitchingCost {
    /// Computes switching cost for a server of some type.
    pub fn switching_cost(&self) -> N64 {
        n64(
            self.energy_cost * (self.epsilon + self.delta * self.phi_max)
                + self.tau
                + self.rho,
        )
    }

    /// Computes normalized switching cost for a server of some type given the time slot length $\delta$.
    /// Approximately, measures the minimum duration a server must be asleep to outweigh the switching cost.
    /// Referred to as $\xi$ in the paper.
    pub fn normalized_switching_cost(&self, delta: f64) -> N64 {
        self.switching_cost() / (self.energy_cost * delta * self.phi_min)
    }

    /// Builds switching cost such that the normalized switching cost matches $normalized_switching_cost$.
    /// Here, $\delta$ is the time slot length.
    pub fn from_normalized(
        delta: f64,
        normalized_switching_cost: f64,
        energy_cost: f64,
        phi_min: f64,
    ) -> Self {
        Self {
            energy_cost,
            phi_min,
            phi_max: 0.,
            epsilon: 0.,
            delta: 0.,
            tau: 0.,
            rho: normalized_switching_cost * energy_cost * delta * phi_min,
        }
    }
}
#[pymethods]
impl SwitchingCost {
    #[new]
    fn constructor(
        energy_cost: f64,
        phi_min: f64,
        phi_max: f64,
        epsilon: f64,
        delta: f64,
        tau: f64,
        rho: f64,
    ) -> Self {
        SwitchingCost {
            energy_cost,
            phi_min,
            phi_max,
            epsilon,
            delta,
            tau,
            rho,
        }
    }

    #[pyo3(name = "switching_cost")]
    fn switching_cost_py(&self) -> PyResult<f64> {
        Ok(self.switching_cost().raw())
    }

    #[pyo3(name = "normalized_switching_cost")]
    fn normalized_switching_cost_py(&self, delta: f64) -> PyResult<f64> {
        Ok(self.normalized_switching_cost(delta).raw())
    }

    #[staticmethod]
    #[pyo3(name = "from_normalized")]
    fn from_normalized_py(
        delta: f64,
        normalized_switching_cost: f64,
        energy_cost: f64,
        phi_min: f64,
    ) -> PyResult<Self> {
        Ok(Self::from_normalized(
            delta,
            normalized_switching_cost,
            energy_cost,
            phi_min,
        ))
    }
}

impl SwitchingCostModel {
    pub fn new(switching_costs: HashMap<String, SwitchingCost>) -> Self {
        SwitchingCostModel(switching_costs)
    }

    /// Builds vector of switching costs for all server types.
    pub fn switching_costs(&self, server_types: &Vec<ServerType>) -> Vec<f64> {
        server_types
            .iter()
            .map(|server_type| self.model(server_type).switching_cost().raw())
            .collect()
    }

    /// Builds vector of normalized switching costs for all server types.
    pub fn normalized_switching_costs(
        &self,
        delta: f64,
        server_types: &Vec<ServerType>,
    ) -> Vec<f64> {
        server_types
            .iter()
            .map(|server_type| {
                self.model(server_type)
                    .normalized_switching_cost(delta)
                    .raw()
            })
            .collect()
    }

    /// Returns model of some server type.
    fn model(&self, server_type: &ServerType) -> &SwitchingCost {
        &self.0[&server_type.key]
    }
}