Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.8/site-packages/scipy/optimize/_nnls.py: 20%

20 statements  

« prev     ^ index     » next       coverage.py v7.3.2, created at 2023-12-12 06:31 +0000

1from . import __nnls 

2from numpy import asarray_chkfinite, zeros, double 

3 

4__all__ = ['nnls'] 

5 

6 

7def nnls(A, b, maxiter=None): 

8 """ 

9 Solve ``argmin_x || Ax - b ||_2`` for ``x>=0``. This is a wrapper 

10 for a FORTRAN non-negative least squares solver. 

11 

12 Parameters 

13 ---------- 

14 A : ndarray 

15 Matrix ``A`` as shown above. 

16 b : ndarray 

17 Right-hand side vector. 

18 maxiter: int, optional 

19 Maximum number of iterations, optional. 

20 Default is ``3 * A.shape[1]``. 

21 

22 Returns 

23 ------- 

24 x : ndarray 

25 Solution vector. 

26 rnorm : float 

27 The residual, ``|| Ax-b ||_2``. 

28 

29 See Also 

30 -------- 

31 lsq_linear : Linear least squares with bounds on the variables 

32 

33 Notes 

34 ----- 

35 The FORTRAN code was published in the book below. The algorithm 

36 is an active set method. It solves the KKT (Karush-Kuhn-Tucker) 

37 conditions for the non-negative least squares problem. 

38 

39 References 

40 ---------- 

41 Lawson C., Hanson R.J., (1987) Solving Least Squares Problems, SIAM 

42 

43 Examples 

44 -------- 

45 >>> import numpy as np 

46 >>> from scipy.optimize import nnls 

47 ... 

48 >>> A = np.array([[1, 0], [1, 0], [0, 1]]) 

49 >>> b = np.array([2, 1, 1]) 

50 >>> nnls(A, b) 

51 (array([1.5, 1. ]), 0.7071067811865475) 

52 

53 >>> b = np.array([-1, -1, -1]) 

54 >>> nnls(A, b) 

55 (array([0., 0.]), 1.7320508075688772) 

56 

57 """ 

58 

59 A, b = map(asarray_chkfinite, (A, b)) 

60 

61 if len(A.shape) != 2: 

62 raise ValueError("Expected a two-dimensional array (matrix)" + 

63 ", but the shape of A is %s" % (A.shape, )) 

64 if len(b.shape) != 1: 

65 raise ValueError("Expected a one-dimensional array (vector)" + 

66 ", but the shape of b is %s" % (b.shape, )) 

67 

68 m, n = A.shape 

69 

70 if m != b.shape[0]: 

71 raise ValueError( 

72 "Incompatible dimensions. The first dimension of " + 

73 "A is %s, while the shape of b is %s" % (m, (b.shape[0], ))) 

74 

75 maxiter = -1 if maxiter is None else int(maxiter) 

76 

77 w = zeros((n,), dtype=double) 

78 zz = zeros((m,), dtype=double) 

79 index = zeros((n,), dtype=int) 

80 

81 x, rnorm, mode = __nnls.nnls(A, m, n, b, w, zz, index, maxiter) 

82 if mode != 1: 

83 raise RuntimeError("too many iterations") 

84 

85 return x, rnorm