New argument: max_inner_iter (#865)

CHANGELOG.rst

@@ -12,6 +12,7 @@ Changelog
 
 **New features:**
 
+- New argument ``max_inner_iter`` for classes :class:`~glum.GeneralizedLinearRegressor` and :class:`~glum.GeneralizedLinearRegressorCV` to control the maximum number of iterations of the inner solver in the IRLS-CD algorithm.
 - New fitted attributes ``col_means_`` and ``col_stds_`` for classes :class:`~glum.GeneralizedLinearRegressor` and :class:`~glum.GeneralizedLinearRegressorCV`.
 - :class:`~glum.GeneralizedLinearRegressor` now prints more informative logs when fitting with ``alpha_search=True`` and ``verbose=True``.
 

src/glum/_glm.py

@@ -730,6 +730,7 @@ def __init__(
         link: Union[str, Link] = "auto",
         solver: str = "auto",
         max_iter=100,
+        max_inner_iter=100000,
         gradient_tol: Optional[float] = None,
         step_size_tol: Optional[float] = None,
         hessian_approx: float = 0.0,
@@ -767,6 +768,7 @@ def __init__(
         self.link = link
         self.solver = solver
         self.max_iter = max_iter
+        self.max_inner_iter = max_inner_iter
         self.gradient_tol = gradient_tol
         self.step_size_tol = step_size_tol
         self.hessian_approx = hessian_approx
@@ -1102,6 +1104,7 @@ def _solve(
                 family=self._family_instance,
                 link=self._link_instance,
                 max_iter=max_iter,
+                max_inner_iter=self.max_inner_iter,
                 gradient_tol=self._gradient_tol,
                 step_size_tol=self.step_size_tol,
                 fixed_inner_tol=fixed_inner_tol,
@@ -2695,6 +2698,10 @@ class GeneralizedLinearRegressor(GeneralizedLinearRegressorBase):
     max_iter : int, optional (default=100)
         The maximal number of iterations for solver algorithms.
 
+    max_inner_iter: int, optional (default=100000)
+        The maximal number of iterations for the inner solver in the IRLS-CD
+        algorithm. This parameter is only used when ``solver='irls-cd'``.
+
     gradient_tol : float, optional (default=None)
         Stopping criterion. If ``None``, solver-specific defaults will be used.
         The default value for most solvers is ``1e-4``, except for
@@ -2942,6 +2949,7 @@ def __init__(
         link: Union[str, Link] = "auto",
         solver: str = "auto",
         max_iter=100,
+        max_inner_iter=100000,
         gradient_tol: Optional[float] = None,
         step_size_tol: Optional[float] = None,
         hessian_approx: float = 0.0,
@@ -2983,6 +2991,7 @@ def __init__(
             link=link,
             solver=solver,
             max_iter=max_iter,
+            max_inner_iter=max_inner_iter,
             gradient_tol=gradient_tol,
             step_size_tol=step_size_tol,
             hessian_approx=hessian_approx,

src/glum/_glm_cv.py

@@ -106,6 +106,10 @@ class GeneralizedLinearRegressorCV(GeneralizedLinearRegressorBase):
     max_iter : int, optional (default=100)
         The maximal number of iterations for solver algorithms.
 
+    max_inner_iter: int, optional (default=100000)
+        The maximal number of iterations for the inner solver in the IRLS-CD
+        algorithm. This parameter is only used when ``solver='irls-cd'``.
+
     gradient_tol : float, optional (default=None)
         Stopping criterion. If ``None``, solver-specific defaults will be used.
         The default value for most solvers is ``1e-4``, except for
@@ -338,6 +342,7 @@ def __init__(
         link: Union[str, Link] = "auto",
         solver: str = "auto",
         max_iter=100,
+        max_inner_iter=100000,
         gradient_tol: Optional[float] = None,
         step_size_tol: Optional[float] = None,
         hessian_approx: float = 0.0,
@@ -381,6 +386,7 @@ def __init__(
             link=link,
             solver=solver,
             max_iter=max_iter,
+            max_inner_iter=max_inner_iter,
             gradient_tol=gradient_tol,
             step_size_tol=step_size_tol,
             hessian_approx=hessian_approx,