Adds Adelie code

randalo/adelie_integration.py

@@ -0,0 +1,68 @@
+import adelie
+import randalo as ra
+import numpy as np
+from dataclasses import dataclass
+
+def curry(f, *args0, **kwargs0):
+    return lambda *args, **kwargs: f(*args0, *args, **kwargs0, **kwargs)
+
+class AdelieState:
+    def __init__(self, state):
+        self.state = state
+        self.ra_lmda = ra.HyperParameter()
+
+    def set_index(self, idx):
+        self.index = idx
+        self.ra_lmda.value = self.state.lmda[idx]
+
+
+def adelie_state_to_Jacobian(state, adelie_state):
+    n, p = state.X.shape
+    G, = state.groups.shape
+    L, = state.lmda_path.shape
+
+    assert p == G, "Group lasso with adelie is not supported."
+    assert state.penalty == None
+
+    assert state.offsets == None
+    assert state.intercept == None
+    ell_1_term = state.alpha * ra.L1Regularizer()
+    ell_2_2_term = (1 - state.alpha) / 2 * ra.SquareRegularizer()
+    reg = adelie_state.ra_lmda * (ell_1_term + ell_2_2_term)
+
+    loss = ra.MSELoss()
+
+    J = ra.Jacobian(lambda: (
+        betas[adelie_state.index], # What is the type of this?
+        screen_set[active_set[active_sizes[:adelie_state.index]]]),
+        loss,
+        reg,
+    )
+
+    return loss, J
+
+def adelie_state_to_randalo(state, adelie_state, loss, J, index, rng):
+    y_hat = state.X @ state.beta[index]
+    adelie_state.set_index(index)
+    randalo = ra.RandALO(
+            loss,
+            J,
+            state.y,
+            y_hat,
+            rng=rng)
+
+    return randalo
+
+def get_alo_for_sweep(state, risk_fun):
+    L, = state.lmda.shape
+    adelie_state = AdelieState(state)
+    loss, J = adelie_state_to_jacobian(state, adelie_state)
+
+    output = np.empty(L)
+
+    for i in range(L):
+        randalo = adelie_state_to_randalo(state, adelie_state, loss, J, i)
+        output[i] = randalo.evaluate(risk_fun)
+
+    return output
+