Coverage for C: \ Users \ peaco \ OneDrive \ Documents \ GitHub \ mt_metadata \ mt_metadata \ features \ weights \ threshold_weight_kernel.py: 100%

1# ===================================================== 

2# Imports 

3# ===================================================== 

4from typing import Annotated 

5 

6from pydantic import Field 

7 

8from mt_metadata.features.weights.taper_monotonic_weight_kernel import ( 

9 TaperMonotonicWeightKernel, 

10) 

11 

12 

13# ===================================================== 

14 

15 

16class ThresholdWeightKernel(TaperMonotonicWeightKernel): 

17 """ 

18 ThresholdWeightKernel 

19 

20 A special case of MonotonicWeightKernel where the transition region is a single value, 

21 resulting in a hard threshold (step function). This kernel outputs 0 or 1 depending on 

22 whether the input is above or below the threshold, according to the threshold_type. 

23 

24 Parameters 

25 ---------- 

26 threshold : float 

27 The threshold value. 

28 threshold_type : str, optional 

29 "low cut" (default) or "high cut". Determines which side is downweighted. 

30 **kwargs : 

31 Additional keyword arguments passed to MonotonicWeightKernel. 

32 """ 

33 

34 threshold_type: Annotated[ 

35 str, 

36 Field( 

37 default="low cut", 

38 description="Which side of a threshold should be downweighted.", 

39 alias=None, 

40 json_schema_extra={ 

41 "units": None, 

42 "required": True, 

43 "examples": ["low cut"], 

44 }, 

45 ), 

46 ] 

47 

48 # TODO: Uncomment if needed for testing or future use 

49 def _normalize(self, values): 

50 """ 

51 Normalize input values to the [0, 1] interval, supporting infinite bounds for activation and taper kernels. 

52 Handles all combinations of finite and infinite transition bounds: 

53 - Both bounds finite: linear normalization. 

54 - Lower bound -inf, upper bound finite: exponential normalization. 

55 - Lower bound finite, upper bound +inf: exponential normalization. 

56 - Both bounds infinite: all values map to 0.5. 

57 """ 

58 

59 raise NotImplementedError( 

60 "Normalization not implemented for ThresholdWeightKernel." 

61 ) 

62 # lb = float(self.transition_lower_bound) 

63 # ub = float(self.transition_upper_bound) 

64 # values = np.asarray(values) 

65 # # Both bounds finite 

66 # if np.isfinite(lb) and np.isfinite(ub): 

67 # return np.clip((values - lb) / (ub - lb), 0, 1) 

68 # # Lower bound -inf, upper bound finite 

69 # elif not np.isfinite(lb) and np.isfinite(ub): 

70 # scale = np.std(values) if np.std(values) > 0 else 1.0 

71 # x = 1 - np.exp(-(ub - values) / scale) 

72 # return np.clip(x, 0, 1) 

73 # # Lower bound finite, upper bound +inf 

74 # elif np.isfinite(lb) and not np.isfinite(ub): 

75 # scale = np.std(values) if np.std(values) > 0 else 1.0 

76 # x = 1 - np.exp(-(values - lb) / scale) 

77 # return np.clip(x, 0, 1) 

78 # # Both bounds infinite 

79 # else: 

80 # return np.full_like(values, 0.5)