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
158
159
160
161
162
163
use super::*;
use std::ops::Deref;

/// A multi objective cost.
pub struct MultiObjectiveCost {
    primary_costs: ObjectiveCosts,
    secondary_costs: ObjectiveCosts,
    value_func: ObjectiveCostValueFn,
}

/// Encapsulates objective which has multiple objectives.
pub struct MultiObjective {
    /// List of primary objectives. Solution can be considered as improvement
    /// only if none of costs, returned by these objectives, is worse.
    primary_objectives: Vec<Box<dyn Objective + Send + Sync>>,
    /// List of secondary objectives. This list is evaluated only if primary objectives
    /// costs are considered as equal.
    secondary_objectives: Vec<Box<dyn Objective + Send + Sync>>,
    /// A function which extract actual cost from multiple objective costs.
    value_func: ObjectiveCostValueFn,
}

impl ObjectiveCost for MultiObjectiveCost {
    fn value(&self) -> f64 {
        self.value_func.deref()(&self.primary_costs, &self.secondary_costs)
    }

    fn cmp_relaxed(&self, other: &ObjectiveCostType) -> (Ordering, Ordering) {
        let (primary_costs, secondary_costs) = self.get_costs(other);

        let (result, _) = match Self::analyze(&self.primary_costs, primary_costs, 0) {
            (Equal, relaxed_count) => Self::analyze(&self.secondary_costs, secondary_costs, relaxed_count),
            result_pair => result_pair,
        };

        (result, result)
    }

    fn clone_box(&self) -> ObjectiveCostType {
        Box::new(Self {
            primary_costs: self.primary_costs.iter().map(|c| c.clone_box()).collect(),
            secondary_costs: self.secondary_costs.iter().map(|c| c.clone_box()).collect(),
            value_func: self.value_func.clone(),
        })
    }

    fn as_any(&self) -> &dyn Any {
        self
    }
}

impl MultiObjectiveCost {
    /// Creates a new instance of `MultiObjectiveCost`.
    pub fn new(
        primary_costs: ObjectiveCosts,
        secondary_costs: ObjectiveCosts,
        value_func: ObjectiveCostValueFn,
    ) -> Self {
        Self { primary_costs, secondary_costs, value_func }
    }

    fn get_costs<'a>(&self, other: &'a ObjectiveCostType) -> (&'a Vec<ObjectiveCostType>, &'a Vec<ObjectiveCostType>) {
        let other = other.as_any().downcast_ref::<MultiObjectiveCost>().expect("Expecting MultiObjectiveCost");

        let primary_costs = &other.primary_costs;
        assert_eq!(self.primary_costs.len(), primary_costs.len());

        let secondary_costs = &other.secondary_costs;
        assert_eq!(self.secondary_costs.len(), secondary_costs.len());

        (primary_costs, secondary_costs)
    }

    fn analyze(left: &[ObjectiveCostType], right: &[ObjectiveCostType], relaxed_count: usize) -> (Ordering, usize) {
        // NOTE Allow not more than one objective to be relaxed at the same time
        const MAX_RELAXED_COUNT: usize = 1;

        let results = left.iter().zip(right.iter()).map(|(left, right)| left.cmp_relaxed(right)).collect::<Vec<_>>();

        let relaxed_count = results.iter().filter(|(a, r)| *a == Greater && *r == Equal).count() + relaxed_count;
        let result_actual = Self::analyze_results(results.iter().map(|(a, _)| *a));
        let result_relaxed = Self::analyze_results(results.iter().map(|(_, r)| *r));

        let result = match (result_actual, result_relaxed) {
            (Less, _) => Less,
            (_, relaxed) if relaxed_count <= MAX_RELAXED_COUNT => relaxed,
            _ => Greater,
        };

        (result, relaxed_count)
    }

    fn analyze_results(results: impl Iterator<Item = Ordering>) -> Ordering {
        results.fold(Equal, |acc, result| match (acc, result) {
            (Equal, new) => new,
            (Less, Greater) => Greater,
            (Less, _) => Less,
            (Greater, _) => Greater,
        })
    }
}

impl MultiObjective {
    /// Creates a new instance of `MultiObjective`.
    pub fn new(
        primary_objectives: Vec<Box<dyn Objective + Send + Sync>>,
        secondary_objectives: Vec<Box<dyn Objective + Send + Sync>>,
        value_func: ObjectiveCostValueFn,
    ) -> Self {
        assert!(!primary_objectives.is_empty() || !secondary_objectives.is_empty());
        Self { primary_objectives, secondary_objectives, value_func }
    }
}

impl Default for MultiObjective {
    fn default() -> Self {
        Self {
            primary_objectives: vec![Box::new(TotalRoutes::default()), Box::new(TotalUnassignedJobs::default())],
            secondary_objectives: vec![Box::new(TotalTransportCost::default())],
            value_func: Arc::new(|_, secondary| secondary.first().unwrap().value()),
        }
    }
}

impl Objective for MultiObjective {
    fn estimate_cost(
        &self,
        refinement_ctx: &mut RefinementContext,
        insertion_ctx: &InsertionContext,
    ) -> ObjectiveCostType {
        let primary_costs =
            self.primary_objectives.iter().map(|o| o.estimate_cost(refinement_ctx, insertion_ctx)).collect::<Vec<_>>();
        let secondary_costs = self
            .secondary_objectives
            .iter()
            .map(|o| o.estimate_cost(refinement_ctx, insertion_ctx))
            .collect::<Vec<_>>();

        Box::new(MultiObjectiveCost::new(primary_costs, secondary_costs, self.value_func.clone()))
    }

    fn is_goal_satisfied(
        &self,
        refinement_ctx: &mut RefinementContext,
        insertion_ctx: &InsertionContext,
    ) -> Option<bool> {
        let mut get_satisfaction = |objectives: &Vec<Box<dyn Objective + Send + Sync>>| {
            objectives.iter().filter_map(|o| o.is_goal_satisfied(refinement_ctx, insertion_ctx)).collect::<Vec<_>>()
        };

        let mut results = get_satisfaction(&self.primary_objectives);

        if results.is_empty() {
            results.extend(get_satisfaction(&self.secondary_objectives).into_iter())
        }

        if results.is_empty() {
            None
        } else {
            Some(results.iter().all(|&goal_satisfied| goal_satisfied))
        }
    }
}