From 07045060ab0236cdebe6bffab4f18ff7964c2c7b Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 10:21:15 +0200
Subject: [PATCH 01/13] init commit

---
 skglm/solvers/bfgs.py | 12 ++++++++++++
 1 file changed, 12 insertions(+)
 create mode 100644 skglm/solvers/bfgs.py

diff --git a/skglm/solvers/bfgs.py b/skglm/solvers/bfgs.py
new file mode 100644
index 000000000..64dcd9678
--- /dev/null
+++ b/skglm/solvers/bfgs.py
@@ -0,0 +1,12 @@
+from skglm.solvers import BaseSolver
+
+
+class BFGS(BaseSolver):
+    """A wrapper for scipy BFGS solver."""
+
+    def __init__(self, max_iter=50, tol=1e-4):
+        self.max_iter = max_iter
+        self.tol = tol
+
+    def solve(X, y, datafit, penalty, w_init=None, Xw_init=None):
+        return

From 7e364212fe059a9dc67acbec239e5167d002c2f3 Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 10:57:28 +0200
Subject: [PATCH 02/13] bfgs wrapper implem

---
 skglm/solvers/bfgs.py | 47 ++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 44 insertions(+), 3 deletions(-)

diff --git a/skglm/solvers/bfgs.py b/skglm/solvers/bfgs.py
index 64dcd9678..4ccf62358 100644
--- a/skglm/solvers/bfgs.py
+++ b/skglm/solvers/bfgs.py
@@ -1,12 +1,53 @@
+import numpy as np
+import scipy.optimize
+from numpy.linalg import norm
+
 from skglm.solvers import BaseSolver
+from skglm.datafits import BaseDatafit
 
 
 class BFGS(BaseSolver):
     """A wrapper for scipy BFGS solver."""
 
-    def __init__(self, max_iter=50, tol=1e-4):
+    def __init__(self, max_iter=50, tol=1e-4, verbose=False):
         self.max_iter = max_iter
         self.tol = tol
+        self.verbose = verbose
+
+    def solve(self, X, y, datafit: BaseDatafit, penalty, w_init=None, Xw_init=None):
+
+        def objective_function(w):
+            Xw = X @ w
+            datafit_value = datafit.value(y, w, Xw)
+            penalty_value = penalty.value(w)
+
+            return datafit_value + penalty_value
+
+        def jacobian_function(w):
+            Xw = X @ w
+            datafit_grad = datafit.gradient(y, w, Xw)
+            penalty_grad = penalty.gradient(w)
+
+            return datafit_grad + penalty_grad
+
+        n_features = X.shape[1]
+        w = np.zeros(n_features) if w_init is not None else w_init
+        p_objs_out = []
+
+        result = scipy.optimize.minimize(
+            fun=objective_function,
+            jac=jacobian_function,
+            x0=w,
+            method="BFGS",
+            options=dict(
+                gtol=self.tol,
+                maxiter=self.max_iter,
+                disp=self.verbose
+            ),
+            callback=lambda w_k: p_objs_out.append(objective_function(w_k))
+        )
+
+        w = result.x
+        stop_crit = norm(result.jac)
 
-    def solve(X, y, datafit, penalty, w_init=None, Xw_init=None):
-        return
+        return w, np.asarray(p_objs_out), stop_crit

From 0fdbcde76637202c807204ef78b45617dc6aedbb Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 11:14:52 +0200
Subject: [PATCH 03/13] implement L2 penalty

---
 skglm/penalties/separable.py | 31 +++++++++++++++++++++++++++++++
 1 file changed, 31 insertions(+)

diff --git a/skglm/penalties/separable.py b/skglm/penalties/separable.py
index 2c1429a87..a2e6468cb 100644
--- a/skglm/penalties/separable.py
+++ b/skglm/penalties/separable.py
@@ -557,3 +557,34 @@ def is_penalized(self, n_features):
     def generalized_support(self, w):
         """Return a mask with non-zero coefficients."""
         return w != 0
+
+
+class L2(BasePenalty):
+    r""":math:`ell_2` penalty.
+
+    The penalty reads::
+
+    .. math::
+
+        \alpha xx \lVert w \rVert_2
+    """
+
+    def __init__(self, alpha):
+        self.alpha = alpha
+
+    def get_spec(self):
+        spec = (
+            ('alpha', float64),
+        )
+        return spec
+
+    def params_to_dict(self):
+        return dict(alpha=self.alpha)
+
+    def value(self, w):
+        """Compute the value of the L2 penalty."""
+        return self.alpha * (w ** 2).sum() / 2
+
+    def gradient(self, w):
+        """Compute the gradient of the L2 penalty."""
+        return self.alpha * w

From c75b21f1ceb69e3ac9a7c2f59374655bf8e1ff8c Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 11:19:13 +0200
Subject: [PATCH 04/13] gradient of logistic

---
 skglm/datafits/single_task.py | 3 +++
 skglm/solvers/bfgs.py         | 4 ++--
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/skglm/datafits/single_task.py b/skglm/datafits/single_task.py
index f21484fd6..457f63501 100644
--- a/skglm/datafits/single_task.py
+++ b/skglm/datafits/single_task.py
@@ -182,6 +182,9 @@ def value(self, y, w, Xw):
     def gradient_scalar(self, X, y, w, Xw, j):
         return (- X[:, j] @ (y * sigmoid(- y * Xw))) / len(y)
 
+    def gradient(self, X, y, Xw):
+        return X.T @ self.raw_grad(y, Xw)
+
     def full_grad_sparse(
             self, X_data, X_indptr, X_indices, y, Xw):
         n_features = X_indptr.shape[0] - 1
diff --git a/skglm/solvers/bfgs.py b/skglm/solvers/bfgs.py
index 4ccf62358..ec2702113 100644
--- a/skglm/solvers/bfgs.py
+++ b/skglm/solvers/bfgs.py
@@ -25,7 +25,7 @@ def objective_function(w):
 
         def jacobian_function(w):
             Xw = X @ w
-            datafit_grad = datafit.gradient(y, w, Xw)
+            datafit_grad = datafit.gradient(X, y, Xw)
             penalty_grad = penalty.gradient(w)
 
             return datafit_grad + penalty_grad
@@ -40,8 +40,8 @@ def jacobian_function(w):
             x0=w,
             method="BFGS",
             options=dict(
-                gtol=self.tol,
                 maxiter=self.max_iter,
+                gtol=self.tol,
                 disp=self.verbose
             ),
             callback=lambda w_k: p_objs_out.append(objective_function(w_k))

From d6358e12c3bf6b244d977a79a6dc48a091b14698 Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 11:50:45 +0200
Subject: [PATCH 05/13] unittest against scikit learn

---
 skglm/penalties/__init__.py     |  4 ++--
 skglm/solvers/__init__.py       |  3 ++-
 skglm/solvers/bfgs.py           |  8 +++++--
 skglm/tests/test_bfgs_solver.py | 41 +++++++++++++++++++++++++++++++++
 4 files changed, 51 insertions(+), 5 deletions(-)
 create mode 100644 skglm/tests/test_bfgs_solver.py

diff --git a/skglm/penalties/__init__.py b/skglm/penalties/__init__.py
index 3fb0771ec..a08b1b859 100644
--- a/skglm/penalties/__init__.py
+++ b/skglm/penalties/__init__.py
@@ -1,6 +1,6 @@
 from .base import BasePenalty
 from .separable import (
-    L1_plus_L2, L0_5, L1, L2_3, MCPenalty, SCAD, WeightedL1, IndicatorBox,
+    L1_plus_L2, L0_5, L1, L2, L2_3, MCPenalty, SCAD, WeightedL1, IndicatorBox,
     PositiveConstraint
 )
 from .block_separable import (
@@ -12,6 +12,6 @@
 
 __all__ = [
     BasePenalty,
-    L1_plus_L2, L0_5, L1, L2_3, MCPenalty, SCAD, WeightedL1, IndicatorBox,
+    L1_plus_L2, L0_5, L1, L2, L2_3, MCPenalty, SCAD, WeightedL1, IndicatorBox,
     PositiveConstraint, L2_05, L2_1, BlockMCPenalty, BlockSCAD, WeightedGroupL2, SLOPE
 ]
diff --git a/skglm/solvers/__init__.py b/skglm/solvers/__init__.py
index 457367018..0a1d6ba0c 100644
--- a/skglm/solvers/__init__.py
+++ b/skglm/solvers/__init__.py
@@ -6,7 +6,8 @@
 from .multitask_bcd import MultiTaskBCD
 from .prox_newton import ProxNewton
 from .group_prox_newton import GroupProxNewton
+from .bfgs import BFGS
 
 
 __all__ = [AndersonCD, BaseSolver, FISTA, GramCD, GroupBCD, MultiTaskBCD, ProxNewton,
-           GroupProxNewton]
+           GroupProxNewton, BFGS]
diff --git a/skglm/solvers/bfgs.py b/skglm/solvers/bfgs.py
index ec2702113..b1ed50d72 100644
--- a/skglm/solvers/bfgs.py
+++ b/skglm/solvers/bfgs.py
@@ -30,8 +30,12 @@ def jacobian_function(w):
 
             return datafit_grad + penalty_grad
 
+        def save_p_obj(w_k):
+            p_obj = objective_function(w_k)
+            p_objs_out.append(p_obj)
+
         n_features = X.shape[1]
-        w = np.zeros(n_features) if w_init is not None else w_init
+        w = np.zeros(n_features) if w_init is None else w_init
         p_objs_out = []
 
         result = scipy.optimize.minimize(
@@ -44,7 +48,7 @@ def jacobian_function(w):
                 gtol=self.tol,
                 disp=self.verbose
             ),
-            callback=lambda w_k: p_objs_out.append(objective_function(w_k))
+            callback=save_p_obj,
         )
 
         w = result.x
diff --git a/skglm/tests/test_bfgs_solver.py b/skglm/tests/test_bfgs_solver.py
new file mode 100644
index 000000000..57429cef6
--- /dev/null
+++ b/skglm/tests/test_bfgs_solver.py
@@ -0,0 +1,41 @@
+import numpy as np
+
+from skglm.solvers import BFGS
+from skglm.penalties import L2
+from skglm.datafits import Logistic
+
+from sklearn.linear_model import LogisticRegression
+
+from skglm.utils.data import make_correlated_data
+from skglm.utils.jit_compilation import compiled_clone
+
+
+def test_bfgs_L2_logreg():
+    reg = 1.
+    n_samples, n_features = 50, 10
+
+    X, y, _ = make_correlated_data(
+        n_samples, n_features, random_state=0)
+    y = np.sign(y)
+
+    datafit = compiled_clone(Logistic())
+    penalty = compiled_clone(L2(reg))
+    solver = BFGS()
+
+    w, *_ = solver.solve(X, y, datafit, penalty)
+
+    # fit scikit learn
+    estimator = LogisticRegression(
+        penalty='l2',
+        C=1 / (n_samples * reg),
+        fit_intercept=False
+    )
+    estimator.fit(X, y)
+
+    np.testing.assert_allclose(
+        w, estimator.coef_.flatten(), atol=1e-4
+    )
+
+
+if __name__ == "__main__":
+    test_bfgs_L2_logreg()

From ebbf0bc6a7c636425ed7f7dd666bc5c890d2f4b9 Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 13:18:33 +0200
Subject: [PATCH 06/13] fix verbosity and p_pbj computation

---
 skglm/solvers/bfgs.py           | 18 ++++++++++++++----
 skglm/tests/test_bfgs_solver.py |  3 ++-
 2 files changed, 16 insertions(+), 5 deletions(-)

diff --git a/skglm/solvers/bfgs.py b/skglm/solvers/bfgs.py
index b1ed50d72..47e57b440 100644
--- a/skglm/solvers/bfgs.py
+++ b/skglm/solvers/bfgs.py
@@ -30,10 +30,21 @@ def jacobian_function(w):
 
             return datafit_grad + penalty_grad
 
-        def save_p_obj(w_k):
+        def callback_post_iter(w_k):
+            # save p_obj
             p_obj = objective_function(w_k)
             p_objs_out.append(p_obj)
 
+            if self.verbose:
+                grad = jacobian_function(w_k)
+                stop_crit = norm(grad)
+
+                it = len(p_objs_out)
+                print(
+                    f"Iteration {it}: {p_obj:.10f}, "
+                    f"stopping crit: {stop_crit:.2e}"
+                )
+
         n_features = X.shape[1]
         w = np.zeros(n_features) if w_init is None else w_init
         p_objs_out = []
@@ -45,10 +56,9 @@ def save_p_obj(w_k):
             method="BFGS",
             options=dict(
                 maxiter=self.max_iter,
-                gtol=self.tol,
-                disp=self.verbose
+                gtol=self.tol
             ),
-            callback=save_p_obj,
+            callback=callback_post_iter,
         )
 
         w = result.x
diff --git a/skglm/tests/test_bfgs_solver.py b/skglm/tests/test_bfgs_solver.py
index 57429cef6..027682f3c 100644
--- a/skglm/tests/test_bfgs_solver.py
+++ b/skglm/tests/test_bfgs_solver.py
@@ -20,7 +20,7 @@ def test_bfgs_L2_logreg():
 
     datafit = compiled_clone(Logistic())
     penalty = compiled_clone(L2(reg))
-    solver = BFGS()
+    solver = BFGS(verbose=1)
 
     w, *_ = solver.solve(X, y, datafit, penalty)
 
@@ -39,3 +39,4 @@ def test_bfgs_L2_logreg():
 
 if __name__ == "__main__":
     test_bfgs_L2_logreg()
+    pass

From 92d6eda805a69f3f2eacda53d6cdcf3db0502c95 Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 13:37:01 +0200
Subject: [PATCH 07/13] BFGS --> L-BFGS

---
 skglm/solvers/__init__.py                                 | 4 ++--
 skglm/solvers/{bfgs.py => lbfgs.py}                       | 6 +++---
 skglm/tests/{test_bfgs_solver.py => test_lbfgs_solver.py} | 8 ++++----
 3 files changed, 9 insertions(+), 9 deletions(-)
 rename skglm/solvers/{bfgs.py => lbfgs.py} (94%)
 rename skglm/tests/{test_bfgs_solver.py => test_lbfgs_solver.py} (87%)

diff --git a/skglm/solvers/__init__.py b/skglm/solvers/__init__.py
index 0a1d6ba0c..9089128d5 100644
--- a/skglm/solvers/__init__.py
+++ b/skglm/solvers/__init__.py
@@ -6,8 +6,8 @@
 from .multitask_bcd import MultiTaskBCD
 from .prox_newton import ProxNewton
 from .group_prox_newton import GroupProxNewton
-from .bfgs import BFGS
+from .lbfgs import LBFGS
 
 
 __all__ = [AndersonCD, BaseSolver, FISTA, GramCD, GroupBCD, MultiTaskBCD, ProxNewton,
-           GroupProxNewton, BFGS]
+           GroupProxNewton, LBFGS]
diff --git a/skglm/solvers/bfgs.py b/skglm/solvers/lbfgs.py
similarity index 94%
rename from skglm/solvers/bfgs.py
rename to skglm/solvers/lbfgs.py
index 47e57b440..25677e205 100644
--- a/skglm/solvers/bfgs.py
+++ b/skglm/solvers/lbfgs.py
@@ -6,8 +6,8 @@
 from skglm.datafits import BaseDatafit
 
 
-class BFGS(BaseSolver):
-    """A wrapper for scipy BFGS solver."""
+class LBFGS(BaseSolver):
+    """A wrapper for scipy L-BFGS solver."""
 
     def __init__(self, max_iter=50, tol=1e-4, verbose=False):
         self.max_iter = max_iter
@@ -53,7 +53,7 @@ def callback_post_iter(w_k):
             fun=objective_function,
             jac=jacobian_function,
             x0=w,
-            method="BFGS",
+            method="L-BFGS-B",
             options=dict(
                 maxiter=self.max_iter,
                 gtol=self.tol
diff --git a/skglm/tests/test_bfgs_solver.py b/skglm/tests/test_lbfgs_solver.py
similarity index 87%
rename from skglm/tests/test_bfgs_solver.py
rename to skglm/tests/test_lbfgs_solver.py
index 027682f3c..2eaf26ad8 100644
--- a/skglm/tests/test_bfgs_solver.py
+++ b/skglm/tests/test_lbfgs_solver.py
@@ -1,6 +1,6 @@
 import numpy as np
 
-from skglm.solvers import BFGS
+from skglm.solvers import LBFGS
 from skglm.penalties import L2
 from skglm.datafits import Logistic
 
@@ -10,7 +10,7 @@
 from skglm.utils.jit_compilation import compiled_clone
 
 
-def test_bfgs_L2_logreg():
+def test_lbfgs_L2_logreg():
     reg = 1.
     n_samples, n_features = 50, 10
 
@@ -20,7 +20,7 @@ def test_bfgs_L2_logreg():
 
     datafit = compiled_clone(Logistic())
     penalty = compiled_clone(L2(reg))
-    solver = BFGS(verbose=1)
+    solver = LBFGS(verbose=1)
 
     w, *_ = solver.solve(X, y, datafit, penalty)
 
@@ -38,5 +38,5 @@ def test_bfgs_L2_logreg():
 
 
 if __name__ == "__main__":
-    test_bfgs_L2_logreg()
+    test_lbfgs_L2_logreg()
     pass

From c130ca32d7899054847d78062e9c64b5adf628d2 Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 13:41:12 +0200
Subject: [PATCH 08/13] add convergence warning

---
 skglm/solvers/lbfgs.py           | 11 +++++++++++
 skglm/tests/test_lbfgs_solver.py |  2 +-
 2 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/skglm/solvers/lbfgs.py b/skglm/solvers/lbfgs.py
index 25677e205..b4e2fd263 100644
--- a/skglm/solvers/lbfgs.py
+++ b/skglm/solvers/lbfgs.py
@@ -1,3 +1,6 @@
+import warnings
+from sklearn.exceptions import ConvergenceWarning
+
 import numpy as np
 import scipy.optimize
 from numpy.linalg import norm
@@ -61,6 +64,14 @@ def callback_post_iter(w_k):
             callback=callback_post_iter,
         )
 
+        if not result.success:
+            warnings.warn(
+                f"`LBFGS` did not converge for tol={self.tol:.3e} "
+                f"and max_iter={self.max_iter}.\n"
+                "Consider increasing `max_iter` and/or `tol`.",
+                category=ConvergenceWarning
+            )
+
         w = result.x
         stop_crit = norm(result.jac)
 
diff --git a/skglm/tests/test_lbfgs_solver.py b/skglm/tests/test_lbfgs_solver.py
index 2eaf26ad8..a4984c9f8 100644
--- a/skglm/tests/test_lbfgs_solver.py
+++ b/skglm/tests/test_lbfgs_solver.py
@@ -20,7 +20,7 @@ def test_lbfgs_L2_logreg():
 
     datafit = compiled_clone(Logistic())
     penalty = compiled_clone(L2(reg))
-    solver = LBFGS(verbose=1)
+    solver = LBFGS()
 
     w, *_ = solver.solve(X, y, datafit, penalty)
 

From 7161dccb1e0cf126b2c41a8a9a3018b9f9068c71 Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 13:47:35 +0200
Subject: [PATCH 09/13] add to docs

---
 doc/api.rst            |  1 +
 skglm/solvers/lbfgs.py | 17 ++++++++++++++++-
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/doc/api.rst b/doc/api.rst
index ef738558d..04ff040b0 100644
--- a/doc/api.rst
+++ b/doc/api.rst
@@ -78,6 +78,7 @@ Solvers
    GramCD
    GroupBCD
    GroupProxNewton
+   LBFGS
    MultiTaskBCD
    ProxNewton
 
diff --git a/skglm/solvers/lbfgs.py b/skglm/solvers/lbfgs.py
index b4e2fd263..a1b8a1bf8 100644
--- a/skglm/solvers/lbfgs.py
+++ b/skglm/solvers/lbfgs.py
@@ -10,7 +10,22 @@
 
 
 class LBFGS(BaseSolver):
-    """A wrapper for scipy L-BFGS solver."""
+    """A wrapper for scipy L-BFGS solver.
+
+    Refer to `scipy L-BFGS-B <https://docs.scipy.org/doc/scipy/reference/optimize.
+    minimize-lbfgsb.html#optimize-minimize-lbfgsb>`_ documentation for details.
+
+    Parameters
+    ----------
+    max_iter : int, default 20
+        Maximum number of outer iterations.
+
+    tol : float, default 1e-4
+        Tolerance for convergence.
+
+    verbose : bool, default False
+        Amount of verbosity. 0/False is silent.
+    """
 
     def __init__(self, max_iter=50, tol=1e-4, verbose=False):
         self.max_iter = max_iter

From 5e3afdbd579074294ee96b6fe92aae108f288373 Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 13:50:47 +0200
Subject: [PATCH 10/13] cleanups

---
 skglm/solvers/lbfgs.py | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/skglm/solvers/lbfgs.py b/skglm/solvers/lbfgs.py
index a1b8a1bf8..a9dcd3c8c 100644
--- a/skglm/solvers/lbfgs.py
+++ b/skglm/solvers/lbfgs.py
@@ -6,7 +6,6 @@
 from numpy.linalg import norm
 
 from skglm.solvers import BaseSolver
-from skglm.datafits import BaseDatafit
 
 
 class LBFGS(BaseSolver):
@@ -18,7 +17,7 @@ class LBFGS(BaseSolver):
     Parameters
     ----------
     max_iter : int, default 20
-        Maximum number of outer iterations.
+        Maximum number of iterations.
 
     tol : float, default 1e-4
         Tolerance for convergence.
@@ -32,7 +31,7 @@ def __init__(self, max_iter=50, tol=1e-4, verbose=False):
         self.tol = tol
         self.verbose = verbose
 
-    def solve(self, X, y, datafit: BaseDatafit, penalty, w_init=None, Xw_init=None):
+    def solve(self, X, y, datafit, penalty, w_init=None, Xw_init=None):
 
         def objective_function(w):
             Xw = X @ w

From 63b3ad0eef02fbb02aacf636cb0ec7d4f49efdb2 Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 13:54:38 +0200
Subject: [PATCH 11/13] L2 to doc

---
 doc/api.rst                  | 1 +
 skglm/penalties/separable.py | 4 ++--
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/doc/api.rst b/doc/api.rst
index 04ff040b0..0f3aac341 100644
--- a/doc/api.rst
+++ b/doc/api.rst
@@ -36,6 +36,7 @@ Penalties
    IndicatorBox
    L0_5
    L1
+   L2
    L1_plus_L2
    L2_3
    MCPenalty
diff --git a/skglm/penalties/separable.py b/skglm/penalties/separable.py
index a2e6468cb..b1046216e 100644
--- a/skglm/penalties/separable.py
+++ b/skglm/penalties/separable.py
@@ -562,11 +562,11 @@ def generalized_support(self, w):
 class L2(BasePenalty):
     r""":math:`ell_2` penalty.
 
-    The penalty reads::
+    The penalty reads
 
     .. math::
 
-        \alpha xx \lVert w \rVert_2
+        \alpha / 2  ||w||_2^2
     """
 
     def __init__(self, alpha):

From eae3b81504f7452f5afa46bb7b0680cf4287d958 Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 13:56:40 +0200
Subject: [PATCH 12/13] cleanups tests

---
 skglm/tests/test_lbfgs_solver.py | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/skglm/tests/test_lbfgs_solver.py b/skglm/tests/test_lbfgs_solver.py
index a4984c9f8..3146aeab2 100644
--- a/skglm/tests/test_lbfgs_solver.py
+++ b/skglm/tests/test_lbfgs_solver.py
@@ -18,11 +18,10 @@ def test_lbfgs_L2_logreg():
         n_samples, n_features, random_state=0)
     y = np.sign(y)
 
+    # fit L-BFGS
     datafit = compiled_clone(Logistic())
     penalty = compiled_clone(L2(reg))
-    solver = LBFGS()
-
-    w, *_ = solver.solve(X, y, datafit, penalty)
+    w, *_ = LBFGS().solve(X, y, datafit, penalty)
 
     # fit scikit learn
     estimator = LogisticRegression(
@@ -38,5 +37,4 @@ def test_lbfgs_L2_logreg():
 
 
 if __name__ == "__main__":
-    test_lbfgs_L2_logreg()
     pass

From 251865ea20de06d9c290c6029cc1cdb9da729867 Mon Sep 17 00:00:00 2001
From: Badr-MOUFAD <badr.moufad@emines.um6p.ma>
Date: Mon, 12 Jun 2023 14:06:30 +0200
Subject: [PATCH 13/13] lexicographic order

---
 doc/api.rst | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/doc/api.rst b/doc/api.rst
index 0f3aac341..9c3e1d21f 100644
--- a/doc/api.rst
+++ b/doc/api.rst
@@ -36,8 +36,8 @@ Penalties
    IndicatorBox
    L0_5
    L1
-   L2
    L1_plus_L2
+   L2
    L2_3
    MCPenalty
    WeightedL1