Coverage for pySDC/core/problem.py: 97%

37 statements  

« prev     ^ index     » next       coverage.py v7.6.9, created at 2024-12-20 14:51 +0000

1#!/usr/bin/env python3 

2# -*- coding: utf-8 -*- 

3""" 

4Description 

5----------- 

6 

7Module containing the base Problem class for pySDC 

8""" 

9 

10import logging 

11 

12from pySDC.core.common import RegisterParams 

13 

14 

15class WorkCounter(object): 

16 """ 

17 Utility class for counting iterations. 

18 

19 Contains one attribute `niter` initialized to zero during 

20 instantiation, which can be incremented by calling object as 

21 a function, e.g 

22 

23 >>> count = WorkCounter() # => niter = 0 

24 >>> count() # => niter = 1 

25 >>> count() # => niter = 2 

26 """ 

27 

28 def __init__(self): 

29 self.niter = 0 

30 

31 def __call__(self, *args, **kwargs): 

32 # *args and **kwargs are necessary for gmres 

33 self.niter += 1 

34 

35 def decrement(self): 

36 self.niter -= 1 

37 

38 def __str__(self): 

39 return f'{self.niter}' 

40 

41 

42class Problem(RegisterParams): 

43 """ 

44 Prototype class for problems, just defines the attributes essential to get started. 

45 

46 Parameters 

47 ---------- 

48 init : list of args 

49 Argument(s) used to initialize data types. 

50 dtype_u : type 

51 Variable data type. Should generate a data variable using dtype_u(init). 

52 dtype_f : type 

53 RHS data type. Should generate a data variable using dtype_f(init). 

54 

55 Attributes 

56 ---------- 

57 logger: logging.Logger 

58 custom logger for problem-related logging. 

59 """ 

60 

61 logger = logging.getLogger('problem') 

62 dtype_u = None 

63 dtype_f = None 

64 

65 def __init__(self, init): 

66 self.work_counters = {} # Dictionary to store WorkCounter objects 

67 self.init = init # Initialization parameter to instantiate data types 

68 

69 @property 

70 def u_init(self): 

71 """Generate a data variable for u""" 

72 return self.dtype_u(self.init) 

73 

74 @property 

75 def f_init(self): 

76 """Generate a data variable for RHS""" 

77 return self.dtype_f(self.init) 

78 

79 @classmethod 

80 def get_default_sweeper_class(cls): 

81 raise NotImplementedError(f'No default sweeper class implemented for {cls} problem!') 

82 

83 def eval_f(self, u, t): 

84 """ 

85 Abstract interface to RHS computation of the ODE 

86 

87 Parameters 

88 ---------- 

89 u : dtype_u 

90 Current values. 

91 t : float 

92 Current time. 

93 

94 Returns 

95 ------- 

96 f : dtype_f 

97 The RHS values. 

98 """ 

99 raise NotImplementedError('ERROR: problem has to implement eval_f(self, u, t)') 

100 

101 def apply_mass_matrix(self, u): # pragma: no cover 

102 """Default mass matrix : identity""" 

103 return u 

104 

105 def generate_scipy_reference_solution(self, eval_rhs, t, u_init=None, t_init=None, **kwargs): 

106 """ 

107 Compute a reference solution using `scipy.solve_ivp` with very small tolerances. 

108 Keep in mind that scipy needs the solution to be a one dimensional array. If you are solving something higher 

109 dimensional, you need to make sure the function `eval_rhs` takes a flattened one-dimensional version as an input 

110 and output, but reshapes to whatever the problem needs for evaluation. 

111 

112 The keyword arguments will be passed to `scipy.solve_ivp`. You should consider passing `method='BDF'` for stiff 

113 problems and to accelerate that you can pass a function that evaluates the Jacobian with arguments `jac(t, u)` 

114 as `jac=jac`. 

115 

116 Args: 

117 eval_rhs (function): Function evaluate the full right hand side. Must have signature `eval_rhs(float: t, numpy.1darray: u)` 

118 t (float): current time 

119 u_init (pySDC.implementations.problem_classes.Lorenz.dtype_u): initial conditions for getting the exact solution 

120 t_init (float): the starting time 

121 

122 Returns: 

123 numpy.ndarray: Reference solution 

124 """ 

125 import numpy as np 

126 from scipy.integrate import solve_ivp 

127 

128 kwargs = { 

129 'atol': 100 * np.finfo(float).eps, 

130 'rtol': 100 * np.finfo(float).eps, 

131 **kwargs, 

132 } 

133 u_init = self.u_exact(t=0) if u_init is None else u_init * 1.0 

134 t_init = 0 if t_init is None else t_init 

135 

136 u_shape = u_init.shape 

137 return solve_ivp(eval_rhs, (t_init, t), u_init.flatten(), **kwargs).y[:, -1].reshape(u_shape) 

138 

139 def get_fig(self): 

140 """ 

141 Get a figure suitable to plot the solution of this problem 

142 

143 Returns 

144 ------- 

145 self.fig : matplotlib.pyplot.figure.Figure 

146 """ 

147 raise NotImplementedError 

148 

149 def plot(self, u, t=None, fig=None): 

150 r""" 

151 Plot the solution. Please supply a figure with the same structure as returned by ``self.get_fig``. 

152 

153 Parameters 

154 ---------- 

155 u : dtype_u 

156 Solution to be plotted 

157 t : float 

158 Time to display at the top of the figure 

159 fig : matplotlib.pyplot.figure.Figure 

160 Figure with the correct structure 

161 

162 Returns 

163 ------- 

164 None 

165 """ 

166 raise NotImplementedError