CUQIpy 1.3.0.post0.dev70__tar.gz → 1.3.0.post0.dev104__tar.gz

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/CUQIpy.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: CUQIpy
-Version: 1.3.0.post0.dev70
+Version: 1.3.0.post0.dev104
 Summary: Computational Uncertainty Quantification for Inverse problems in Python
 Maintainer-email: "Nicolai A. B. Riis" <nabr@dtu.dk>, "Jakob S. Jørgensen" <jakj@dtu.dk>, "Amal M. Alghamdi" <amaal@dtu.dk>, Chao Zhang <chaz@dtu.dk>
 License:                                  Apache License
@@ -202,7 +202,8 @@ Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: matplotlib
 Requires-Dist: numpy>=1.17.0
-Requires-Dist: scipy<1.13
+Requires-Dist: scipy==1.12.0; python_version <= "3.9"
+Requires-Dist: scipy; python_version > "3.9"
 Requires-Dist: arviz
 Requires-Dist: tqdm
 Dynamic: license-file

cuqipy-1.3.0.post0.dev104/CUQIpy.egg-info/requires.txt

@@ -0,0 +1,10 @@
+matplotlib
+numpy>=1.17.0
+arviz
+tqdm
+
+[:python_version <= "3.9"]
+scipy==1.12.0
+
+[:python_version > "3.9"]
+scipy

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: CUQIpy
-Version: 1.3.0.post0.dev70
+Version: 1.3.0.post0.dev104
 Summary: Computational Uncertainty Quantification for Inverse problems in Python
 Maintainer-email: "Nicolai A. B. Riis" <nabr@dtu.dk>, "Jakob S. Jørgensen" <jakj@dtu.dk>, "Amal M. Alghamdi" <amaal@dtu.dk>, Chao Zhang <chaz@dtu.dk>
 License:                                  Apache License
@@ -202,7 +202,8 @@ Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: matplotlib
 Requires-Dist: numpy>=1.17.0
-Requires-Dist: scipy<1.13
+Requires-Dist: scipy==1.12.0; python_version <= "3.9"
+Requires-Dist: scipy; python_version > "3.9"
 Requires-Dist: arviz
 Requires-Dist: tqdm
 Dynamic: license-file

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/cuqi/_version.py

@@ -8,11 +8,11 @@ import json
 
 version_json = '''
 {
- "date": "2025-04-10T10:13:59+0200",
+ "date": "2025-04-28T12:59:25+0200",
  "dirty": false,
  "error": null,
- "full-revisionid": "c18048b15c6ec1a66e8e4bc097e0180fc3d4e47e",
- "version": "1.3.0.post0.dev70"
+ "full-revisionid": "c62aed4105837bc321e7e325f1b1a2286c659b2f",
+ "version": "1.3.0.post0.dev104"
 }
 '''  # END VERSION_JSON
 

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/cuqi/distribution/_truncated_normal.py

@@ -107,7 +107,7 @@ class TruncatedNormal(Distribution):
         """
         # check if x falls in the range between np.array a and b
         if np.any(x < self.low) or np.any(x > self.high):
-            return np.NaN*np.ones_like(x)
+            return np.nan*np.ones_like(x)
         else:
             return self._normal.gradient(x, *args, **kwargs)
 

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/cuqi/distribution/_uniform.py

@@ -47,7 +47,7 @@ class Uniform(Distribution):
         Computes the gradient of logpdf at the given values of x.
         """
         if np.any(x < self.low) or np.any(x > self.high):
-            return np.NaN*np.ones_like(x)
+            return np.nan*np.ones_like(x)
         else:
             return np.zeros_like(x)
 

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/cuqi/experimental/mcmc/_rto.py

@@ -3,7 +3,7 @@ from scipy.linalg.interpolative import estimate_spectral_norm
 from scipy.sparse.linalg import LinearOperator as scipyLinearOperator
 import numpy as np
 import cuqi
-from cuqi.solver import CGLS, FISTA, ADMM, ScipyLinearLSQ
+from cuqi.solver import CGLS, FISTA, ADMM, ScipyLinearLSQ, ScipyMinimizer
 from cuqi.experimental.mcmc import Sampler
 
 
@@ -167,6 +167,7 @@ class RegularizedLinearRTO(LinearRTO):
     ADMM:  [2] Boyd et al. "Distributed optimization and statistical learning via the alternating direction method of multipliers."Foundations and Trends® in Machine learning, 2011.
            Used when prior.proximal is a list of penalty terms.
     ScipyLinearLSQ: Wrapper for Scipy's lsq_linear for the Trust Region Reflective algorithm. Optionally used when the constraint is either "nonnegativity" or "box".
+    ScipyMinimizer: Wrapper for Scipy's minimize. Optionally used when the constraint is either "nonnegativity" or "box".
 
     Parameters
     ------------
@@ -177,7 +178,7 @@ class RegularizedLinearRTO(LinearRTO):
         Initial point for the sampler. *Optional*.
 
     maxit : int
-        Maximum number of iterations of the FISTA/ADMM/ScipyLinearLSQ solver. *Optional*.
+        Maximum number of iterations of the FISTA/ADMM/ScipyLinearLSQ/ScipyMinimizer solver. *Optional*.
 
     inner_max_it : int
         Maximum number of iterations of the CGLS solver used within the ADMM solver. *Optional*.
@@ -191,7 +192,7 @@ class RegularizedLinearRTO(LinearRTO):
         See [2] or `cuqi.solver.ADMM`
 
     abstol : float
-        Absolute tolerance of the FISTA/ScipyLinearLSQ solver. *Optional*.
+        Absolute tolerance of the FISTA/ScipyLinearLSQ/ScipyMinimizer solver. *Optional*.
     
     inner_abstol : float
         Tolerance parameter for ScipyLinearLSQ's inner solve of the unbounded least-squares problem. *Optional*.
@@ -200,7 +201,7 @@ class RegularizedLinearRTO(LinearRTO):
         If True, FISTA is used as solver, otherwise ISTA is used. *Optional*.
     
     solver : string
-        If set to "ScipyLinearLSQ", solver is set to cuqi.solver.ScipyLinearLSQ, otherwise FISTA/ISTA or ADMM is used. Note "ScipyLinearLSQ" can only be used with `RegularizedGaussian` of `box` or `nonnegativity` constraint. *Optional*.
+        Options are "FISTA" (default for a single constraint or regularization), "ADMM" (default and the only option for multiple constraints or regularizations), "ScipyLinearLSQ" and "ScipyMinimizer". Note "ScipyLinearLSQ" and "ScipyMinimizer" can only be used with `RegularizedGaussian` of a single `box` or `nonnegativity` constraint. *Optional*.
 
     callback : callable, optional
         A function that will be called after each sampling step. It can be useful for monitoring the sampler during sampling.
@@ -234,11 +235,11 @@ class RegularizedLinearRTO(LinearRTO):
 
     @solver.setter
     def solver(self, value):
-        if value == "ScipyLinearLSQ":
+        if value == "ScipyLinearLSQ" or value == "ScipyMinimizer":
             if (self.target.prior.preset["constraint"] == "nonnegativity" or self.target.prior.preset["constraint"] == "box"):
                 self._solver = value
             else:
-                raise ValueError("ScipyLinearLSQ only supports RegularizedGaussian with box or nonnegativity constraint.")
+                raise ValueError("ScipyLinearLSQ and ScipyMinimizer only support RegularizedGaussian with box or nonnegativity constraint.")
         else:
             self._solver = value
 
@@ -281,15 +282,22 @@ class RegularizedLinearRTO(LinearRTO):
             sim = ADMM(self.M, y, self.proximal,
                         self.current_point, self.penalty_parameter, maxit = self.maxit, inner_max_it = self.inner_max_it, adaptive = self.adaptive)
         elif self.solver == "ScipyLinearLSQ":
-                A_op = sp.sparse.linalg.LinearOperator((sum([llh.dim for llh in self.likelihoods])+self.target.prior.dim, self.target.prior.dim),
-                                        matvec=lambda x: self.M(x, 1),
-                                        rmatvec=lambda x: self.M(x, 2)
-                                        )
-                sim = ScipyLinearLSQ(A_op, y, self.target.prior._box_bounds, 
-                                     max_iter = self.maxit,
-                                     lsmr_maxiter = self.inner_max_it, 
-                                     tol = self.abstol,
-                                     lsmr_tol = self.inner_abstol)
+            A_op = sp.sparse.linalg.LinearOperator((sum([llh.distribution.dim for llh in self.likelihoods])+self.target.prior.dim, self.target.prior.dim),
+                                    matvec=lambda x: self.M(x, 1),
+                                    rmatvec=lambda x: self.M(x, 2)
+                                    )
+            sim = ScipyLinearLSQ(A_op, y, self.target.prior._box_bounds, 
+                                    max_iter = self.maxit,
+                                    lsmr_maxiter = self.inner_max_it, 
+                                    tol = self.abstol,
+                                    lsmr_tol = self.inner_abstol)
+        elif self.solver == "ScipyMinimizer":
+            # Adapt bounds format, as scipy.minimize requires a bounds format 
+            # different than that in scipy.lsq_linear.
+            bounds = [(self.target.prior._box_bounds[0][i], self.target.prior._box_bounds[1][i]) for i in range(self.target.prior.dim)]
+            # Note that the objective function is defined as 0.5*||Mx-y||^2, 
+            # and the corresponding gradient (gradfunc) is given by M^T(Mx-y).
+            sim = ScipyMinimizer(lambda x: 0.5*np.sum((self.M(x, 1)-y)**2), self.current_point, gradfunc=lambda x: self.M(self.M(x, 1) - y, 2), bounds=bounds, tol=self.abstol, options={"maxiter": self.maxit})
         else:
             raise ValueError("Choice of solver not supported.")
 

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/cuqi/implicitprior/_regularizedGaussian.py

@@ -189,12 +189,18 @@ class RegularizedGaussian(Distribution):
             elif c_lower == "increasing":
                 if not isinstance(self.geometry, Continuous1D):
                     raise ValueError("Geometry not supported for " + c_lower)
-                self._constraint_prox = lambda z, _: spoptimize.isotonic_regression(z, increasing=True).x
+                if hasattr(spoptimize, 'isotonic_regression'):
+                    self._constraint_prox = lambda z, _: spoptimize.isotonic_regression(z, increasing=True).x
+                else:
+                    raise AttributeError(f"The function 'isotonic_regression' does not exist in scipy.optimize. Installed scipy version: {spoptimize.__version__}. You need to install a scipy >= 1.12.0")
                 self._preset["constraint"] = "increasing"
             elif c_lower == "decreasing":
                 if not isinstance(self.geometry, Continuous1D):
                     raise ValueError("Geometry not supported for " + c_lower)
-                self._constraint_prox = lambda z, _: spoptimize.isotonic_regression(z, increasing=False).x
+                if hasattr(spoptimize, 'isotonic_regression'):
+                    self._constraint_prox = lambda z, _: spoptimize.isotonic_regression(z, increasing=False).x
+                else:
+                    raise AttributeError(f"The function 'isotonic_regression' does not exist in scipy.optimize. Installed scipy version: {spoptimize.__version__}. You need to install a scipy >= 1.12.0")
                 self._preset["constraint"] = "decreasing"
             elif c_lower == "convex":
                 if not isinstance(self.geometry, Continuous1D):

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/cuqi/solver/_solver.py

@@ -196,8 +196,11 @@ class ScipyLSQ(object):
         'trf', Trust Region Reflective algorithm: for large sparse problems with bounds.
         'dogbox', dogleg algorithm with rectangular trust regions, for small problems with bounds.
         'lm', Levenberg-Marquardt algorithm as implemented in MINPACK. Doesn't handle bounds and sparse Jacobians.
+    tol : The numerical tolerance for convergence checks.
+    maxit : The maximum number of iterations.
+    kwargs : Additional keyword arguments passed to scipy's least_squares. Empty by default. See documentation for scipy.optimize.least_squares
     """
-    def __init__(self, func, x0, jacfun='2-point', method='trf', loss='linear', tol=1e-6, maxit=1e4):
+    def __init__(self, func, x0, jacfun='2-point', method='trf', loss='linear', tol=1e-6, maxit=1e4, **kwargs):
         self.func = func
         self.x0 = x0
         self.jacfun = jacfun
@@ -205,6 +208,7 @@ class ScipyLSQ(object):
         self.loss = loss
         self.tol = tol
         self.maxit = int(maxit)
+        self.kwargs = kwargs
     
     def solve(self):
         """Runs optimization algorithm and returns solution and info.
@@ -215,7 +219,7 @@ class ScipyLSQ(object):
             Solution found (array_like) and optimization information (dictionary).
         """
         solution = least_squares(self.func, self.x0, jac=self.jacfun, \
-                                method=self.method, loss=self.loss, xtol=self.tol, max_nfev=self.maxit)
+                                method=self.method, loss=self.loss, xtol=self.tol, max_nfev=self.maxit, **self.kwargs)
         info = {"success": solution['success'],
                 "message": solution['message'],
                 "func": solution['fun'],

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/cuqi/utilities/_utilities.py

@@ -188,7 +188,7 @@ def approx_derivative(func, wrt, direction=None, epsilon=np.sqrt(np.finfo(float)
     # We compute the Jacobian matrix of func using forward differences.
     # If the function is scalar-valued, we compute the gradient instead.
     # If the direction is provided, we compute the direction-Jacobian product.
-    wrt = np.asfarray(wrt)
+    wrt = np.asarray(wrt)
     f0 = func(wrt)
     Matr = np.zeros([infer_len(wrt), infer_len(f0)])
     dx = np.zeros(len(wrt))

cuqipy-1.3.0.post0.dev104/requirements.txt

@@ -0,0 +1,6 @@
+matplotlib
+numpy>=1.17.0
+scipy==1.12.0; python_version <= '3.9'
+scipy; python_version > '3.9'
+arviz
+tqdm

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/tests/test_bayesian_inversion.py

@@ -2,7 +2,9 @@ from typing import Dict
 
 import pytest
 import numpy as np
+import scipy
 import sys
+from packaging import version
 
 from cuqi.testproblem import Deconvolution1D
 from cuqi.distribution import Gaussian, GMRF, CMRF, LMRF, Gamma
@@ -25,6 +27,14 @@ def test_TP_BayesianProblem_sample(copy_reference, TP_type, phantom, prior, Ns,
     # SKIP NUTS test if not windows (for now)
     if isinstance(prior, CMRF) and not sys.platform.startswith('win'):
         pytest.skip("NUTS(CMRF) regression test is not implemented for this platform")
+    elif (
+        isinstance(prior, RegularizedGaussian) 
+        or isinstance(prior, RegularizedGMRF)
+    ) and version.parse(scipy.__version__) >= version.parse('1.13.0'):
+        pytest.skip(
+            "Saved reference data is not consistent with scipy>=1.13.0 "
+            "(due to different default behaviour of scipy.optimize)"
+        )
 
     np.random.seed(19937)
 

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/tests/test_implicit_priors.py

@@ -1,6 +1,8 @@
 import cuqi
 import numpy as np
+import scipy
 import pytest
+from packaging import version
 
 def test_RegularizedGaussian_default_init():
     """ Test that the implicit regularized Gaussian requires at least 1 regularization argument """
@@ -186,6 +188,7 @@ def test_RegularizedGaussian_double_preset():
     assert len(x.proximal[0]) == 2
     assert len(x.proximal[1]) == 2
 
+@pytest.mark.skipif(version.parse(scipy.__version__) < version.parse('1.12.0'), reason="isotonic_regression in not available in scipy < 1.12.0")
 def test_regression_increasing():
     """ Regression test for the increasing constraints in RegularizedGaussian"""
 

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/tests/test_model.py

@@ -19,7 +19,7 @@ def test_LinearModel_getMatrix(seed):
     mat1 = model1.get_matrix() #Normal matrix
     mat2 = model2.get_matrix() #Sparse matrix (generated from functions)
 
-    assert np.allclose(mat1,mat2.A)
+    assert np.allclose(mat1,mat2.toarray())
 
 def test_initialize_model_dim():
     model1 = cuqi.model.Model(lambda x:x, range_geometry=4, domain_geometry=4)

{cuqipy-1.3.0.post0.dev70 → cuqipy-1.3.0.post0.dev104}/tests/test_solver.py

@@ -100,19 +100,27 @@ def test_ScipyLinearLSQ_with_LinearOperator():
     sol, _ = ScipyLinearLSQ(A_op, b).solve()
     assert np.allclose(sol, ref_sol, rtol=1e-10)
 
-def test_ScipyLinearLSQ_against_FISTA():
+def test_ScipyLinearLSQ_against_ScipyMinimizer_and_against_FISTA():
     A = np.array([[73,71,52],[87,74,46],[72,2,7],[80,89,71]])
     b = np.array([49,67,68,20])
+
+    # solve with ScipyMinimizer
+    def fun(x):
+        return 0.5*np.linalg.norm(A@x-b)**2
+    def jac(x):
+        return A.T@(A@x-b)
+    sol_min, _ = ScipyMinimizer(fun, np.zeros(3), gradfunc=jac, tol=1e-10, bounds=[(0,np.inf),(0,np.inf),(0,np.inf)]).solve()
+
     # solve with ScipyLinearLSQ
-    lb = np.zeros(3)
-    ub = lb + np.inf
-    sol_lsq, _ = ScipyLinearLSQ(A, b, (lb,ub)).solve()
+    sol_lsq, _ = ScipyLinearLSQ(A, b, ([0,0,0],[np.inf,np.inf,np.inf]), tol=1e-10).solve()
+
     # solve with FISTA
     rng = np.random.default_rng(seed = 1219)
     x0 = rng.standard_normal(3)
     sol_fista, _ = FISTA(A, b, lambda x, _: ProjectNonnegative(x), x0, stepsize=1e-7, maxit=100000, abstol=1e-16, adaptive=True).solve()
 
     assert np.allclose(sol_lsq, sol_fista, rtol=1e-8)
+    assert np.allclose(sol_min, sol_lsq, rtol=1e-8)
 
 def test_LM():
     # compare to MATLAB's original code solution

cuqipy-1.3.0.post0.dev70/CUQIpy.egg-info/requires.txt

@@ -1,5 +0,0 @@
-matplotlib
-numpy>=1.17.0
-scipy<1.13
-arviz
-tqdm

cuqipy-1.3.0.post0.dev70/requirements.txt

@@ -1,5 +0,0 @@
-matplotlib
-numpy>=1.17.0
-scipy<1.13
-arviz
-tqdm