2 Unit tests for the :class:`SymmetricLinearGame` class.
5 from unittest
import TestCase
7 from dunshire
.games
import SymmetricLinearGame
8 from dunshire
.matrices
import eigenvalues_re
, inner_product
, norm
9 from dunshire
import options
10 from .randomgen
import (random_icecream_game
, random_ll_icecream_game
,
11 random_ll_orthant_game
, random_nn_scaling
,
12 random_orthant_game
, random_positive_orthant_game
,
16 # Tell pylint to shut up about the large number of methods.
17 class SymmetricLinearGameTest(TestCase
): # pylint: disable=R0904
19 Tests for the SymmetricLinearGame and Solution classes.
21 def assert_within_tol(self
, first
, second
, modifier
=1):
23 Test that ``first`` and ``second`` are equal within a multiple of
24 our default tolerances.
30 The first number to compare.
33 The second number to compare.
36 A scaling factor (default: 1) applied to the default
37 tolerance for this comparison. If you have a poorly-
38 conditioned matrix, for example, you may want to set this
42 self
.assertTrue(abs(first
- second
) < options
.ABS_TOL
*modifier
)
45 def test_solutions_dont_change_orthant(self
):
46 G
= random_orthant_game()
47 self
.assert_solutions_dont_change(G
)
49 def test_solutions_dont_change_icecream(self
):
50 G
= random_icecream_game()
51 self
.assert_solutions_dont_change(G
)
53 def assert_solutions_dont_change(self
, G
):
55 If we solve the same problem twice, we should get
56 the same answer both times.
60 p1_diff
= norm(soln1
.player1_optimal() - soln2
.player1_optimal())
61 p2_diff
= norm(soln1
.player2_optimal() - soln2
.player2_optimal())
62 gv_diff
= abs(soln1
.game_value() - soln2
.game_value())
64 p1_close
= p1_diff
< options
.ABS_TOL
65 p2_close
= p2_diff
< options
.ABS_TOL
66 gv_close
= gv_diff
< options
.ABS_TOL
68 self
.assertTrue(p1_close
and p2_close
and gv_close
)
71 def assert_player1_start_valid(self
, G
):
72 x
= G
.player1_start()['x']
73 s
= G
.player1_start()['s']
74 s1
= s
[0:G
.dimension()]
75 s2
= s
[G
.dimension():]
76 self
.assert_within_tol(norm(G
.A()*x
- G
.b()), 0)
77 self
.assertTrue((s1
,s2
) in G
.C())
80 def test_player1_start_valid_orthant(self
):
82 Ensure that player one's starting point is in the orthant.
84 G
= random_orthant_game()
85 self
.assert_player1_start_valid(G
)
88 def test_player1_start_valid_icecream(self
):
90 Ensure that player one's starting point is in the ice-cream cone.
92 G
= random_icecream_game()
93 self
.assert_player1_start_valid(G
)
96 def test_condition_lower_bound(self
):
98 Ensure that the condition number of a game is greater than or
101 It should be safe to compare these floats directly: we compute
102 the condition number as the ratio of one nonnegative real number
103 to a smaller nonnegative real number.
105 G
= random_orthant_game()
106 self
.assertTrue(G
.condition() >= 1.0)
107 G
= random_icecream_game()
108 self
.assertTrue(G
.condition() >= 1.0)
111 def assert_scaling_works(self
, G
):
113 Test that scaling ``L`` by a nonnegative number scales the value
114 of the game by the same number.
116 (alpha
, H
) = random_nn_scaling(G
)
117 value1
= G
.solution().game_value()
118 value2
= H
.solution().game_value()
119 modifier
= 4*max(abs(alpha
), 1)
120 self
.assert_within_tol(alpha
*value1
, value2
, modifier
)
123 def test_scaling_orthant(self
):
125 Test that scaling ``L`` by a nonnegative number scales the value
126 of the game by the same number over the nonnegative orthant.
128 G
= random_orthant_game()
129 self
.assert_scaling_works(G
)
132 def test_scaling_icecream(self
):
134 The same test as :meth:`test_nonnegative_scaling_orthant`,
135 except over the ice cream cone.
137 G
= random_icecream_game()
138 self
.assert_scaling_works(G
)
141 def assert_translation_works(self
, G
):
143 Check that translating ``L`` by alpha*(e1*e2.trans()) increases
144 the value of the associated game by alpha.
146 # We need to use ``L`` later, so make sure we transpose it
147 # before passing it in as a column-indexed matrix.
149 value1
= soln1
.game_value()
150 x_bar
= soln1
.player1_optimal()
151 y_bar
= soln1
.player2_optimal()
153 # This is the "correct" representation of ``M``, but COLUMN
155 (alpha
, H
) = random_translation(G
)
156 value2
= H
.solution().game_value()
158 modifier
= 4*max(abs(alpha
), 1)
159 self
.assert_within_tol(value1
+ alpha
, value2
, modifier
)
161 # Make sure the same optimal pair works.
162 self
.assert_within_tol(value2
, H
.payoff(x_bar
, y_bar
), modifier
)
165 def test_translation_orthant(self
):
167 Test that translation works over the nonnegative orthant.
169 G
= random_orthant_game()
170 self
.assert_translation_works(G
)
173 def test_translation_icecream(self
):
175 The same as :meth:`test_translation_orthant`, except over the
178 G
= random_icecream_game()
179 self
.assert_translation_works(G
)
182 def assert_opposite_game_works(self
, G
):
184 Check the value of the "opposite" game that gives rise to a
185 value that is the negation of the original game. Comes from
188 # This is the "correct" representation of ``M``, but
192 # so we have to transpose it when we feed it to the constructor.
193 # Note: the condition number of ``H`` should be comparable to ``G``.
194 H
= SymmetricLinearGame(M
.trans(), G
.K(), G
.e2(), G
.e1())
197 x_bar
= soln1
.player1_optimal()
198 y_bar
= soln1
.player2_optimal()
201 # The modifier of 4 is because each could be off by 2*ABS_TOL,
202 # which is how far apart the primal/dual objectives have been
204 self
.assert_within_tol(-soln1
.game_value(), soln2
.game_value(), 4)
206 # Make sure the switched optimal pair works. Since x_bar and
207 # y_bar come from G, we use the same modifier.
208 self
.assert_within_tol(soln2
.game_value(), H
.payoff(y_bar
, x_bar
), 4)
212 def test_opposite_game_orthant(self
):
214 Test the value of the "opposite" game over the nonnegative
217 G
= random_orthant_game()
218 self
.assert_opposite_game_works(G
)
221 def test_opposite_game_icecream(self
):
223 Like :meth:`test_opposite_game_orthant`, except over the
226 G
= random_icecream_game()
227 self
.assert_opposite_game_works(G
)
230 def assert_orthogonality(self
, G
):
232 Two orthogonality relations hold at an optimal solution, and we
236 x_bar
= soln
.player1_optimal()
237 y_bar
= soln
.player2_optimal()
238 value
= soln
.game_value()
240 ip1
= inner_product(y_bar
, G
.L()*x_bar
- value
*G
.e1())
241 ip2
= inner_product(value
*G
.e2() - G
.L().trans()*y_bar
, x_bar
)
243 # Huh.. well, y_bar and x_bar can each be epsilon away, but
244 # x_bar is scaled by L, so that's (norm(L) + 1), and then
245 # value could be off by epsilon, so that's another norm(e1) or
246 # norm(e2). On the other hand, this test seems to pass most of
247 # the time even with a modifier of one. How about.. four?
248 self
.assert_within_tol(ip1
, 0, 4)
249 self
.assert_within_tol(ip2
, 0, 4)
252 def test_orthogonality_orthant(self
):
254 Check the orthgonality relationships that hold for a solution
255 over the nonnegative orthant.
257 G
= random_orthant_game()
258 self
.assert_orthogonality(G
)
261 def test_orthogonality_icecream(self
):
263 Check the orthgonality relationships that hold for a solution
264 over the ice-cream cone.
266 G
= random_icecream_game()
267 self
.assert_orthogonality(G
)
270 def test_positive_operator_value(self
):
272 Test that a positive operator on the nonnegative orthant gives
273 rise to a a game with a nonnegative value.
275 This test theoretically applies to the ice-cream cone as well,
276 but we don't know how to make positive operators on that cone.
278 G
= random_positive_orthant_game()
279 self
.assertTrue(G
.solution().game_value() >= -options
.ABS_TOL
)
282 def assert_lyapunov_works(self
, G
):
284 Check that Lyapunov games act the way we expect.
288 # We only check for positive/negative stability if the game
289 # value is not basically zero. If the value is that close to
290 # zero, we just won't check any assertions.
292 # See :meth:`assert_within_tol` for an explanation of the
294 eigs
= eigenvalues_re(G
.L())
296 if soln
.game_value() > options
.ABS_TOL
:
297 # L should be positive stable
298 positive_stable
= all([eig
> -options
.ABS_TOL
for eig
in eigs
])
299 self
.assertTrue(positive_stable
)
300 elif soln
.game_value() < -options
.ABS_TOL
:
301 # L should be negative stable
302 negative_stable
= all([eig
< options
.ABS_TOL
for eig
in eigs
])
303 self
.assertTrue(negative_stable
)
305 # The dual game's value should always equal the primal's.
306 # The modifier of 4 is because even though the games are dual,
307 # CVXOPT doesn't know that, and each could be off by 2*ABS_TOL.
308 dualsoln
= G
.dual().solution()
309 self
.assert_within_tol(dualsoln
.game_value(), soln
.game_value(), 4)
312 def test_lyapunov_orthant(self
):
314 Test that a Lyapunov game on the nonnegative orthant works.
316 G
= random_ll_orthant_game()
317 self
.assert_lyapunov_works(G
)
320 def test_lyapunov_icecream(self
):
322 Test that a Lyapunov game on the ice-cream cone works.
324 G
= random_ll_icecream_game()
325 self
.assert_lyapunov_works(G
)