update test_bm_example1

benches/test_bm_example1.py

@@ -3,13 +3,14 @@
 
 import numpy as np
 
-from ellalgo.cutting_plane import cutting_plane_optim
+from ellalgo.cutting_plane import Options, cutting_plane_optim
 from ellalgo.ell import Ell
 from ellalgo.ell_typing import OracleFeas, OracleOptim
 
 
 class MyOracleFeas(OracleFeas):
     idx = 0
+    gamma = 0.0
 
     # constraint 1: x + y <= 3
     def fn1(self, x, y):
@@ -19,15 +20,21 @@ def fn1(self, x, y):
     def fn2(self, x, y):
         return -x + y + 1
 
+    def fn0(self, x, y):
+        return self.gamma - (x + y)
+
     def grad1(self):
         return np.array([1.0, 1.0])
 
     def grad2(self):
         return np.array([-1.0, 1.0])
 
+    def grad0(self):
+        return np.array([-1.0, -1.0])
+
     def __init__(self):
-        self.fns = (self.fn1, self.fn2)
-        self.grads = (self.grad1, self.grad2)
+        self.fns = (self.fn1, self.fn2, self.fn0)
+        self.grads = (self.grad1, self.grad2, self.grad0)
 
     def assess_feas(self, z):
         """[summary]
@@ -40,9 +47,9 @@ def assess_feas(self, z):
         """
         x, y = z
 
-        for _ in [0, 1]:
+        for _ in range(3):
             self.idx += 1
-            if self.idx == 2:
+            if self.idx == 3:
                 self.idx = 0  # round robin
             if (fj := self.fns[self.idx](x, y)) > 0:
                 return self.grads[self.idx](), fj
@@ -62,20 +69,16 @@ def assess_optim(self, z, gamma: float):
         Returns:
             [type]: [description]
         """
+        self.helper.gamma = gamma
         if cut := self.helper.assess_feas(z):
             return cut, None
         x, y = z
         # objective: maximize x + y
-        f0 = x + y
-        if (fj := gamma - f0) > 0.0:
-            return (-1.0 * np.array([1.0, 1.0]), fj), None
-
-        gamma = f0
-        return (-1.0 * np.array([1.0, 1.0]), 0.0), gamma
+        return (np.array([-1.0, -1.0]), 0.0), x + y
 
 
 class MyOracleFeas2(OracleFeas):
-    idx = 0
+    gamma = 0.0
 
     # constraint 1: x + y <= 3
     def fn1(self, x, y):
@@ -85,15 +88,21 @@ def fn1(self, x, y):
     def fn2(self, x, y):
         return -x + y + 1
 
+    def fn0(self, x, y):
+        return self.gamma - (x + y)
+
     def grad1(self):
         return np.array([1.0, 1.0])
 
     def grad2(self):
         return np.array([-1.0, 1.0])
 
+    def grad0(self):
+        return np.array([-1.0, -1.0])
+
     def __init__(self):
-        self.fns = (self.fn1, self.fn2)
-        self.grads = (self.grad1, self.grad2)
+        self.fns = (self.fn1, self.fn2, self.fn0)
+        self.grads = (self.grad1, self.grad2, self.grad0)
 
     def assess_feas(self, z):
         """[summary]
@@ -105,13 +114,9 @@ def assess_feas(self, z):
             [type]: [description]
         """
         x, y = z
-
-        for _ in [0, 1]:
-            self.idx += 1
-            if self.idx == 2:
-                self.idx = 0  # round robin
-            if (fj := self.fns[self.idx](x, y)) > 0:
-                return self.grads[self.idx](), fj
+        for idx in range(3):
+            if (fj := self.fns[idx](x, y)) > 0:
+                return self.grads[idx](), fj
         return None
 
 
@@ -128,31 +133,29 @@ def assess_optim(self, z, gamma: float):
         Returns:
             [type]: [description]
         """
+        self.helper.gamma = gamma
         if cut := self.helper.assess_feas(z):
             return cut, None
         x, y = z
         # objective: maximize x + y
-        f0 = x + y
-        if (fj := gamma - f0) > 0.0:
-            return (-1.0 * np.array([1.0, 1.0]), fj), None
-
-        gamma = f0
-        return (-1.0 * np.array([1.0, 1.0]), 0.0), gamma
+        return (np.array([-1.0, -1.0]), 0.0), x + y
 
 
 def run_example1(omega):
     xinit = np.array([0.0, 0.0])  # initial xinit
     ellip = Ell(10.0, xinit)
-    xbest, _, num_iters = cutting_plane_optim(omega(), ellip, float("-inf"))
+    options = Options()
+    options.tolerance = 1e-10
+    xbest, _, num_iters = cutting_plane_optim(omega(), ellip, float("-inf"), options)
     assert xbest is not None
     return num_iters
 
 
 def test_bm_with_round_robin(benchmark):
     num_iters = benchmark(run_example1, MyOracle)
-    assert num_iters == 31
+    assert num_iters == 25
 
 
 def test_bm_without_round_robin(benchmark):
     num_iters = benchmark(run_example1, MyOracle2)
-    assert num_iters == 31
+    assert num_iters == 25