Merge pull request #10 from nschloe/coverage

fix for krylov

npx/_krylov.py

@@ -1,5 +1,5 @@
 from collections import namedtuple
-from typing import Optional
+from typing import Callable, Optional
 
 import numpy as np
 import scipy
@@ -15,12 +15,15 @@ def cg(
     tol: float = 1e-05,
     maxiter: Optional[int] = None,
     M=None,
-    callback=None,
+    callback: Optional[Callable] = None,
     atol: Optional[float] = 0.0,
     exact_solution=None,
 ):
     resnorms = []
 
+    if x0 is None:
+        x0 = np.zeros(A.shape[1])
+
     if exact_solution is None:
         errnorms = None
     else:
@@ -60,12 +63,15 @@ def gmres(
     restart: Optional[int] = None,
     maxiter: Optional[int] = None,
     M=None,
-    callback=None,
+    callback: Optional[Callable] = None,
     atol: Optional[float] = 0.0,
     exact_solution=None,
 ):
     resnorms = []
 
+    if x0 is None:
+        x0 = np.zeros(A.shape[1])
+
     if exact_solution is None:
         errnorms = None
     else:
@@ -114,11 +120,14 @@ def minres(
     tol: float = 1e-05,
     maxiter: Optional[int] = None,
     M=None,
-    callback=None,
+    callback: Optional[Callable] = None,
     exact_solution=None,
 ):
     resnorms = []
 
+    if x0 is None:
+        x0 = np.zeros(A.shape[1])
+
     if exact_solution is None:
         errnorms = None
     else:

npx/_minimize.py

@@ -1,8 +1,10 @@
+from typing import Callable
+
 import numpy as np
 import scipy.optimize
 
 
-def minimize(fun, x0, *args, **kwargs):
+def minimize(fun: Callable, x0, *args, **kwargs):
     x0 = np.asarray(x0)
     x0_shape = x0.shape
 

npx/_nonlinear.py

@@ -1,3 +1,5 @@
+from typing import Callable
+
 import numpy as np
 
 
@@ -10,7 +12,7 @@ def _dot_last(a, b):
     return out
 
 
-def bisect(f, a, b, tol, max_num_steps=np.infty):
+def bisect(f: Callable, a, b, tol: float, max_num_steps=np.infty):
     a = np.asarray(a)
     b = np.asarray(b)
 
@@ -51,7 +53,7 @@ def bisect(f, a, b, tol, max_num_steps=np.infty):
     return a, b
 
 
-def regula_falsi(f, a, b, tol, max_num_steps=np.infty):
+def regula_falsi(f: Callable, a, b, tol: float, max_num_steps=np.infty):
     a = np.asarray(a)
     b = np.asarray(b)
     fa = np.asarray(f(a))

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = npx
-version = 0.0.8
+version = 0.0.9
 author = Nico Schlömer
 author_email = nico.schloemer@gmail.com
 description = Some useful extensions for NumPy

test/test_krylov.py

@@ -4,15 +4,17 @@
 import npx
 
 
-def _run(fun, resnorms1, resnorms2, tol=1.0e-13):
+def _run(method, resnorms1, resnorms2, tol=1.0e-13):
     n = 10
     data = -np.ones((3, n))
     data[1] = 2.0
     A = scipy.sparse.spdiags(data, [-1, 0, 1], n, n)
     A = A.tocsr()
     b = np.ones(n)
 
-    sol, info = fun(A, b)
+    exact_solution = scipy.sparse.linalg.spsolve(A, b)
+
+    sol, info = method(A, b, exact_solution=exact_solution)
     assert sol is not None
     assert info.success
     resnorms1 = np.asarray(resnorms1)
@@ -23,7 +25,7 @@ def _run(fun, resnorms1, resnorms2, tol=1.0e-13):
 
     # with "preconditioning"
     M = scipy.sparse.linalg.LinearOperator((n, n), matvec=lambda x: 0.5 * x)
-    sol, info = fun(A, b, M=M)
+    sol, info = method(A, b, M=M)
 
     assert sol is not None
     assert info.success