Refactoring restarting

include/common.hpp

@@ -35,16 +35,9 @@ using FunctionType = std::function<Float(const Vector &)>;
 
 namespace constants
 {
-	extern Float max_dsigma;
-	extern Float min_dsigma;
-	extern Float tol_condition_cov;
-	extern Float stagnation_quantile;
-	extern Float sigma_threshold;
 	extern size_t cache_max_doubles;
 	extern size_t cache_min_samples;
 	extern bool cache_samples;
-	extern Float lb_sigma;
-	extern Float ub_sigma; 
 	extern bool clip_sigma; 
 }
 

include/restart_criteria.hpp

@@ -17,6 +17,8 @@ namespace restart
 
         Criterion(const std::string& name): met(false), name(name) {}
 
+        virtual ~Criterion() = default;
+
         void reset(const parameters::Parameters &p); 
 
         virtual void update(const parameters::Parameters &p) = 0;
@@ -69,9 +71,17 @@ namespace restart
         void on_reset(const parameters::Parameters &p) override;
     };
 
-    struct SigmaOutOfBounds: Criterion 
+    struct MaxSigma: Criterion 
+    {
+        static inline Float tolerance = 1e4;
+        MaxSigma(): Criterion("MaxSigma"){}
+        void update(const parameters::Parameters &p) override;
+    };
+
+    struct MinSigma: Criterion 
     {
-        SigmaOutOfBounds(): Criterion("SigmaOutOfBounds"){}
+        static inline Float tolerance = 1e-20;
+        MinSigma(): Criterion("MinSigma"){}
         void update(const parameters::Parameters &p) override;
     };
 
@@ -94,6 +104,7 @@ namespace restart
 
     struct TolX: Criterion 
     {
+        static inline Float tolerance = 10e-12;
         Vector tolx_vector;
         TolX(): Criterion("TolX"){}
         void update(const parameters::Parameters &p) override;
@@ -103,37 +114,44 @@ namespace restart
 
     struct MaxDSigma: Criterion 
     {
+        static inline Float tolerance = std::pow(10., 20.);
         MaxDSigma(): Criterion("MaxDSigma"){}
         void update(const parameters::Parameters &p) override;
     };
 
     struct MinDSigma: Criterion 
     {
+        static inline Float tolerance = 1e-8;
         MinDSigma(): Criterion("MinDSigma"){}
         void update(const parameters::Parameters &p) override;
     };
 
 
     struct ConditionC: Criterion 
     {
+        static inline Float tolerance = std::pow(10., 14.);
         ConditionC(): Criterion("ConditionC"){}
         void update(const parameters::Parameters &p) override;
     };
 
     struct NoEffectAxis: Criterion 
     {
+        static inline Float tolerance = 0.;
         NoEffectAxis(): Criterion("NoEffectAxis"){}
         void update(const parameters::Parameters &p) override;
     };
 
     struct NoEffectCoord: Criterion 
     {
+        static inline Float tolerance = 0.;
         NoEffectCoord(): Criterion("NoEffectCoord"){}
         void update(const parameters::Parameters &p) override;
     };
 
     struct Stagnation: Criterion 
     {
+        static inline Float tolerance = 0.3;
+
         size_t n_stagnation;
         std::vector<Float> median_fitnesses;
 		std::vector<Float> best_fitnesses;

modcma/c_maes/cmaescpp/__init__.pyi

@@ -1,4 +1,4 @@
-from typing import Any, Callable, List, Optional, Union, overload, ClassVar
+from typing import Any, Callable, List, overload, ClassVar
 
 import numpy
 from . import (
@@ -21,30 +21,10 @@ class constants:
     cache_max_doubles: ClassVar[int] = ...
     cache_min_samples: ClassVar[int] = ...
     cache_samples: ClassVar[bool] = ...
-    sigma_threshold: ClassVar[float] = ...
-    stagnation_quantile: ClassVar[float] = ...
-    tol_condition_cov: ClassVar[float] = ...
-    tol_min_sigma: ClassVar[float] = ...
-    tolup_sigma: ClassVar[float] = ...
+    clip_sigma: ClassVar[bool] = ...
     def __init__(self, *args, **kwargs) -> None: ...
 
-class ModularCMAES:
-    @overload
-    def __init__(self, parameters: Parameters) -> None: ...
-    @overload
-    def __init__(self, dimension: int) -> None: ...
-    @overload
-    def __init__(self, settings: parameters.Settings) -> None: ...
-    def adapt(self, objective: Callable[[numpy.ndarray], float]) -> None: ...
-    def break_conditions(self) -> bool: ...
-    def mutate(self, arg0: Callable[[numpy.ndarray], float]) -> None: ...
-    def recombine(self) -> None: ...
-    def run(self, objective: Callable[[numpy.ndarray], float]) -> None: ...
-    def select(self) -> None: ...
-    def step(self, objective: Callable[[numpy.ndarray], float]) -> bool: ...
-    def __call__(self, objective: Callable[[numpy.ndarray], float]) -> None: ...
-    @property
-    def p(self) -> Parameters: ...
+
 
 class Population:
     X: numpy.ndarray
@@ -108,17 +88,21 @@ class Parameters:
     def start(
         self, objective: Callable[[numpy.ndarray[numpy.float64[m, 1]]], float]
     ) -> None: ...
-
-class constants:
-    cache_max_doubles: ClassVar[int] = ...
-    cache_min_samples: ClassVar[int] = ...
-    cache_samples: ClassVar[bool] = ...
-    clip_sigma: ClassVar[bool] = ...
-    lb_sigma: ClassVar[float] = ...
-    max_dsigma: ClassVar[float] = ...
-    min_dsigma: ClassVar[float] = ...
-    sigma_threshold: ClassVar[float] = ...
-    stagnation_quantile: ClassVar[float] = ...
-    tol_condition_cov: ClassVar[float] = ...
-    ub_sigma: ClassVar[float] = ...
-    def __init__(self, *args, **kwargs) -> None: ...
+
+class ModularCMAES:
+    @overload
+    def __init__(self, parameters: Parameters) -> None: ...
+    @overload
+    def __init__(self, dimension: int) -> None: ...
+    @overload
+    def __init__(self, settings: parameters.Settings) -> None: ...
+    def adapt(self) -> None: ...
+    def break_conditions(self) -> bool: ...
+    def mutate(self, arg0: Callable[[numpy.ndarray], float]) -> None: ...
+    def recombine(self) -> None: ...
+    def run(self, objective: Callable[[numpy.ndarray], float]) -> None: ...
+    def select(self) -> None: ...
+    def step(self, objective: Callable[[numpy.ndarray], float]) -> bool: ...
+    def __call__(self, objective: Callable[[numpy.ndarray], float]) -> None: ...
+    @property
+    def p(self) -> Parameters: ...

modcma/c_maes/cmaescpp/restart.pyi

@@ -1,15 +1,35 @@
+from typing import ClassVar
+
 import numpy
 
 class Strategy:
     def __init__(self, *args, **kwargs) -> None: ...
     def update(self, parameters) -> float: ...
 
+class BIPOP(Strategy):
+    budget: int
+    budget_large: int
+    budget_small: int
+    lambda_init: int
+    lambda_large: int
+    lambda_small: int
+    mu_factor: float
+    def __init__(self, *args, **kwargs) -> None: ...
+    def large(self) -> bool: ...
+    @property
+    def used_budget(self) -> int: ...
+
+class IPOP(Strategy):
+    ipop_factor: float
+    def __init__(self, *args, **kwargs) -> None: ...
+
 class Criterion:
     last_restart: int
     met: bool
     name: str
     def __init__(self, *args, **kwargs) -> None: ...
     def reset(self, parameters) -> None: ...
+    def on_reset(self, parameters) -> None: ...
     def update(self, parameters) -> None: ...
 
 class Criteria:
@@ -20,20 +40,8 @@ class Criteria:
     @property
     def any(self) -> bool: ...
 
-class BIPOP(Strategy):
-    budget: int
-    budget_large: int
-    budget_small: int
-    lambda_init: int
-    lambda_large: int
-    lambda_small: int
-    mu_factor: float
-    def __init__(self, *args, **kwargs) -> None: ...
-    def large(self) -> bool: ...
-    @property
-    def used_budget(self) -> int: ...
-
 class ConditionC(Criterion):
+    tolerance: ClassVar[float] = ...
     def __init__(self) -> None: ...
 
 class ExceededMaxIter(Criterion):
@@ -42,42 +50,49 @@ class ExceededMaxIter(Criterion):
 
 class FlatFitness(Criterion):
     flat_fitness_index: int
-    flat_fitnesses: numpy.ndarray[numpy.int32[m, 1]]
+    flat_fitnesses: numpy.ndarray[numpy.int32]
     max_flat_fitness: int
     def __init__(self) -> None: ...
 
-class IPOP(Strategy):
-    ipop_factor: float
-    def __init__(self, *args, **kwargs) -> None: ...
-
 class MaxDSigma(Criterion):
+    tolerance: ClassVar[float] = ...
     def __init__(self) -> None: ...
 
 class MinDSigma(Criterion):
+    tolerance: ClassVar[float] = ...
     def __init__(self) -> None: ...
 
 class NoEffectAxis(Criterion):
+    tolerance: ClassVar[float] = ...
     def __init__(self) -> None: ...
 
 class NoEffectCoord(Criterion):
+    tolerance: ClassVar[float] = ...
     def __init__(self) -> None: ...
 
 class NoImprovement(Criterion):
     best_fitnesses: list[float]
     n_bin: int
     def __init__(self) -> None: ...
 
-class SigmaOutOfBounds(Criterion):
+class MinSigma(Criterion):
+    tolerance: ClassVar[float] = ...
+    def __init__(self) -> None: ...
+
+class MaxSigma(Criterion):
+    tolerance: ClassVar[float] = ...
     def __init__(self) -> None: ...
 
 class Stagnation(Criterion):
     best_fitnesses: list[float]
     median_fitnesses: list[float]
     n_stagnation: int
+    tolerance: ClassVar[float] = ...
     def __init__(self) -> None: ...
 
 class TolX(Criterion):
-    tolx_vector: numpy.ndarray[numpy.float64[m, 1]]
+    tolx_vector: numpy.ndarray[numpy.float64]
+    tolerance: ClassVar[float] = ...
     def __init__(self) -> None: ...
 
 class UnableToAdapt(Criterion):

src/common.cpp

@@ -11,16 +11,9 @@ std::ostream& operator<<(std::ostream& os, const std::vector<T>& x)
 
 namespace constants
 {
-	Float max_dsigma = std::pow(10., 20.);
-	Float min_dsigma = 1e-8;
-	Float tol_condition_cov = pow(10., 14.);
-	Float stagnation_quantile = 0.3;
-	Float sigma_threshold = 1e-4;
 	size_t cache_max_doubles = 2'000'000;
 	size_t cache_min_samples = 128;
 	bool cache_samples = false;
-	Float lb_sigma = 1e-20;
-	Float ub_sigma = 1e4;
 	bool clip_sigma = false; 
 }