|
| 1 | +import common |
| 2 | +import pytest |
| 3 | +import torch |
| 4 | + |
| 5 | +from gflownet.envs.crystals.clattice_parameters import ( |
| 6 | + CUBIC, |
| 7 | + HEXAGONAL, |
| 8 | + MONOCLINIC, |
| 9 | + ORTHORHOMBIC, |
| 10 | + PARAMETER_NAMES, |
| 11 | + RHOMBOHEDRAL, |
| 12 | + TETRAGONAL, |
| 13 | + TRICLINIC, |
| 14 | + CLatticeParametersSingleDimIncrement, |
| 15 | +) |
| 16 | +from gflownet.envs.crystals.lattice_parameters import LATTICE_SYSTEMS |
| 17 | +from gflownet.utils.common import tfloat |
| 18 | + |
| 19 | +N_REPETITIONS = 100 |
| 20 | + |
| 21 | + |
| 22 | +def values_are_source_or_distinct(values, source_value=-1): |
| 23 | + values_not_source = [v for v in values if v != source_value] |
| 24 | + return len({*values_not_source}) == len(values_not_source) |
| 25 | + |
| 26 | + |
| 27 | +@pytest.fixture() |
| 28 | +def env(lattice_system): |
| 29 | + return CLatticeParametersSingleDimIncrement( |
| 30 | + lattice_system=lattice_system, |
| 31 | + min_length=1.0, |
| 32 | + max_length=5.0, |
| 33 | + min_angle=30.0, |
| 34 | + max_angle=150.0, |
| 35 | + ) |
| 36 | + |
| 37 | + |
| 38 | +@pytest.mark.parametrize("lattice_system", LATTICE_SYSTEMS) |
| 39 | +def test__environment__initializes_properly(env, lattice_system): |
| 40 | + pass |
| 41 | + |
| 42 | + |
| 43 | +@pytest.mark.parametrize( |
| 44 | + "lattice_system, expected_params", |
| 45 | + [ |
| 46 | + (CUBIC, [None, None, None, 90, 90, 90]), |
| 47 | + (HEXAGONAL, [None, None, None, 90, 90, 120]), |
| 48 | + (MONOCLINIC, [None, None, None, 90, None, 90]), |
| 49 | + (ORTHORHOMBIC, [None, None, None, 90, 90, 90]), |
| 50 | + (RHOMBOHEDRAL, [None, None, None, None, None, None]), |
| 51 | + (TETRAGONAL, [None, None, None, 90, 90, 90]), |
| 52 | + (TRICLINIC, [None, None, None, None, None, None]), |
| 53 | + ], |
| 54 | +) |
| 55 | +def test__environment__has_expected_fixed_parameters( |
| 56 | + env, lattice_system, expected_params |
| 57 | +): |
| 58 | + for expected_value, param_name in zip(expected_params, PARAMETER_NAMES): |
| 59 | + if expected_value is not None: |
| 60 | + assert getattr(env, param_name) == expected_value |
| 61 | + |
| 62 | + |
| 63 | +@pytest.mark.parametrize( |
| 64 | + "lattice_system", |
| 65 | + [CUBIC], |
| 66 | +) |
| 67 | +@pytest.mark.repeat(N_REPETITIONS) |
| 68 | +def test__cubic__constraints_remain_after_random_actions(env, lattice_system): |
| 69 | + env = env.reset() |
| 70 | + while not env.done: |
| 71 | + (a, b, c), (alpha, beta, gamma) = env._unpack_lengths_angles() |
| 72 | + assert len({a, b, c}) == 1 |
| 73 | + assert len({alpha, beta, gamma, 90.0}) == 1 |
| 74 | + env.step_random() |
| 75 | + |
| 76 | + |
| 77 | +@pytest.mark.parametrize( |
| 78 | + "lattice_system", |
| 79 | + [HEXAGONAL], |
| 80 | +) |
| 81 | +@pytest.mark.repeat(N_REPETITIONS) |
| 82 | +def test__hexagonal__constraints_remain_after_random_actions(env, lattice_system): |
| 83 | + env = env.reset() |
| 84 | + while not env.done: |
| 85 | + env.step_random() |
| 86 | + (a, b, c), (alpha, beta, gamma) = env._unpack_lengths_angles() |
| 87 | + assert a == b |
| 88 | + if c != -1: |
| 89 | + assert c != a |
| 90 | + assert len({alpha, beta, 90.0}) == 1 |
| 91 | + assert gamma == 120.0 |
| 92 | + |
| 93 | + |
| 94 | +@pytest.mark.parametrize( |
| 95 | + "lattice_system", |
| 96 | + [MONOCLINIC], |
| 97 | +) |
| 98 | +@pytest.mark.repeat(N_REPETITIONS) |
| 99 | +def test__monoclinic__constraints_remain_after_random_actions(env, lattice_system): |
| 100 | + env = env.reset() |
| 101 | + while not env.done: |
| 102 | + env.step_random() |
| 103 | + (a, b, c), (alpha, beta, gamma) = env._unpack_lengths_angles() |
| 104 | + values_are_source_or_distinct((a, b, c)) |
| 105 | + assert len({alpha, gamma, 90.0}) == 1 |
| 106 | + assert beta != 90.0 |
| 107 | + |
| 108 | + |
| 109 | +@pytest.mark.parametrize( |
| 110 | + "lattice_system", |
| 111 | + [ORTHORHOMBIC], |
| 112 | +) |
| 113 | +@pytest.mark.repeat(N_REPETITIONS) |
| 114 | +def test__orthorhombic__constraints_remain_after_random_actions(env, lattice_system): |
| 115 | + env = env.reset() |
| 116 | + while not env.done: |
| 117 | + env.step_random() |
| 118 | + (a, b, c), (alpha, beta, gamma) = env._unpack_lengths_angles() |
| 119 | + values_are_source_or_distinct((a, b, c)) |
| 120 | + assert len({alpha, beta, gamma, 90.0}) == 1 |
| 121 | + |
| 122 | + |
| 123 | +@pytest.mark.parametrize( |
| 124 | + "lattice_system", |
| 125 | + [RHOMBOHEDRAL], |
| 126 | +) |
| 127 | +@pytest.mark.repeat(N_REPETITIONS) |
| 128 | +def test__rhombohedral__constraints_remain_after_random_actions(env, lattice_system): |
| 129 | + env = env.reset() |
| 130 | + while not env.done: |
| 131 | + env.step_random() |
| 132 | + (a, b, c), (alpha, beta, gamma) = env._unpack_lengths_angles() |
| 133 | + assert len({a, b, c}) == 1 |
| 134 | + assert len({alpha, beta, gamma}) == 1 |
| 135 | + assert len({alpha, beta, gamma, 90.0}) == 2 |
| 136 | + |
| 137 | + |
| 138 | +@pytest.mark.parametrize( |
| 139 | + "lattice_system", |
| 140 | + [TETRAGONAL], |
| 141 | +) |
| 142 | +@pytest.mark.repeat(N_REPETITIONS) |
| 143 | +def test__tetragonal__constraints_remain_after_random_actions(env, lattice_system): |
| 144 | + env = env.reset() |
| 145 | + while not env.done: |
| 146 | + env.step_random() |
| 147 | + (a, b, c), (alpha, beta, gamma) = env._unpack_lengths_angles() |
| 148 | + values_are_source_or_distinct((a, c)) |
| 149 | + values_are_source_or_distinct((b, c)) |
| 150 | + assert len({alpha, beta, gamma, 90.0}) == 1 |
| 151 | + |
| 152 | + |
| 153 | +@pytest.mark.parametrize( |
| 154 | + "lattice_system", |
| 155 | + [TRICLINIC], |
| 156 | +) |
| 157 | +@pytest.mark.repeat(N_REPETITIONS) |
| 158 | +def test__triclinic__constraints_remain_after_random_actions(env, lattice_system): |
| 159 | + env = env.reset() |
| 160 | + while not env.done: |
| 161 | + env.step_random() |
| 162 | + (a, b, c), (alpha, beta, gamma) = env._unpack_lengths_angles() |
| 163 | + values_are_source_or_distinct((a, b, c)) |
| 164 | + values_are_source_or_distinct((alpha, beta, gamma, 90.0)) |
| 165 | + |
| 166 | + |
| 167 | +@pytest.mark.parametrize( |
| 168 | + "lattice_system, states, states_proxy_expected", |
| 169 | + [ |
| 170 | + ( |
| 171 | + TRICLINIC, |
| 172 | + [ |
| 173 | + [0.0, 0.0, 0.0, 0.0, 0.0, 0.0], |
| 174 | + [0.0, 0.2, 0.5, 0.0, 0.5, 1.0], |
| 175 | + [1.0, 1.0, 1.0, 1.0, 1.0, 1.0], |
| 176 | + ], |
| 177 | + [ |
| 178 | + [1.0, 1.0, 1.0, 30.0, 30.0, 30.0], |
| 179 | + [1.0, 1.8, 3.0, 30.0, 90.0, 150.0], |
| 180 | + [5.0, 5.0, 5.0, 150.0, 150.0, 150.0], |
| 181 | + ], |
| 182 | + ), |
| 183 | + ( |
| 184 | + CUBIC, |
| 185 | + [ |
| 186 | + [0.0, 0.0, 0.0, 0.0, 0.0, 0.0], |
| 187 | + [0.25, 0.5, 0.75, 0.25, 0.5, 0.75], |
| 188 | + [1.0, 1.0, 1.0, 1.0, 1.0, 1.0], |
| 189 | + ], |
| 190 | + [ |
| 191 | + [1.0, 1.0, 1.0, 30.0, 30.0, 30.0], |
| 192 | + [2.0, 3.0, 4.0, 60.0, 90.0, 120.0], |
| 193 | + [5.0, 5.0, 5.0, 150.0, 150.0, 150.0], |
| 194 | + ], |
| 195 | + ), |
| 196 | + ], |
| 197 | +) |
| 198 | +def test__statetorch2proxy__returns_expected( |
| 199 | + env, lattice_system, states, states_proxy_expected |
| 200 | +): |
| 201 | + """ |
| 202 | + Various lattice systems are tried because the conversion should be independent of |
| 203 | + the lattice system, since the states are expected to satisfy the constraints. |
| 204 | + """ |
| 205 | + # Get policy states from the batch of states converted into each subenv |
| 206 | + # Get policy states from env.statetorch2policy |
| 207 | + states_torch = tfloat(states, float_type=env.float, device=env.device) |
| 208 | + states_proxy_expected_torch = tfloat( |
| 209 | + states_proxy_expected, float_type=env.float, device=env.device |
| 210 | + ) |
| 211 | + states_proxy = env.statetorch2proxy(states_torch) |
| 212 | + assert torch.all(torch.eq(states_proxy, states_proxy_expected_torch)) |
| 213 | + states_proxy = env.statebatch2proxy(states_torch) |
| 214 | + assert torch.all(torch.eq(states_proxy, states_proxy_expected_torch)) |
| 215 | + |
| 216 | + |
| 217 | +@pytest.mark.parametrize( |
| 218 | + "lattice_system, states, states_policy_expected", |
| 219 | + [ |
| 220 | + ( |
| 221 | + TRICLINIC, |
| 222 | + [ |
| 223 | + [0.0, 0.0, 0.0, 0.0, 0.0, 0.0], |
| 224 | + [0.0, 0.2, 0.5, 0.0, 0.5, 1.0], |
| 225 | + [1.0, 1.0, 1.0, 1.0, 1.0, 1.0], |
| 226 | + ], |
| 227 | + [ |
| 228 | + [0.0, 0.0, 0.0, 0.0, 0.0, 0.0], |
| 229 | + [0.0, 0.2, 0.5, 0.0, 0.5, 1.0], |
| 230 | + [1.0, 1.0, 1.0, 1.0, 1.0, 1.0], |
| 231 | + ], |
| 232 | + ), |
| 233 | + ( |
| 234 | + CUBIC, |
| 235 | + [ |
| 236 | + [0.0, 0.0, 0.0, 0.0, 0.0, 0.0], |
| 237 | + [0.25, 0.5, 0.75, 0.25, 0.5, 0.75], |
| 238 | + [1.0, 1.0, 1.0, 1.0, 1.0, 1.0], |
| 239 | + ], |
| 240 | + [ |
| 241 | + [0.0, 0.0, 0.0, 0.0, 0.0, 0.0], |
| 242 | + [0.25, 0.5, 0.75, 0.25, 0.5, 0.75], |
| 243 | + [1.0, 1.0, 1.0, 1.0, 1.0, 1.0], |
| 244 | + ], |
| 245 | + ), |
| 246 | + ], |
| 247 | +) |
| 248 | +def test__statetorch2policy__returns_expected( |
| 249 | + env, lattice_system, states, states_policy_expected |
| 250 | +): |
| 251 | + """ |
| 252 | + Various lattice systems are tried because the conversion should be independent of |
| 253 | + the lattice system, since the states are expected to satisfy the constraints. |
| 254 | + """ |
| 255 | + # Get policy states from the batch of states converted into each subenv |
| 256 | + # Get policy states from env.statetorch2policy |
| 257 | + states_torch = tfloat(states, float_type=env.float, device=env.device) |
| 258 | + states_policy_expected_torch = tfloat( |
| 259 | + states_policy_expected, float_type=env.float, device=env.device |
| 260 | + ) |
| 261 | + states_policy = env.statetorch2policy(states_torch) |
| 262 | + assert torch.all(torch.eq(states_policy, states_policy_expected_torch)) |
| 263 | + states_policy = env.statebatch2policy(states_torch) |
| 264 | + assert torch.all(torch.eq(states_policy, states_policy_expected_torch)) |
| 265 | + |
| 266 | + |
| 267 | +@pytest.mark.parametrize( |
| 268 | + "lattice_system, expected_output", |
| 269 | + [ |
| 270 | + (CUBIC, "(1.0, 1.0, 1.0), (90.0, 90.0, 90.0)"), |
| 271 | + (HEXAGONAL, "(1.0, 1.0, 1.0), (90.0, 90.0, 120.0)"), |
| 272 | + (MONOCLINIC, "(1.0, 1.0, 1.0), (90.0, 30.0, 90.0)"), |
| 273 | + (ORTHORHOMBIC, "(1.0, 1.0, 1.0), (90.0, 90.0, 90.0)"), |
| 274 | + (RHOMBOHEDRAL, "(1.0, 1.0, 1.0), (30.0, 30.0, 30.0)"), |
| 275 | + (TETRAGONAL, "(1.0, 1.0, 1.0), (90.0, 90.0, 90.0)"), |
| 276 | + (TRICLINIC, "(1.0, 1.0, 1.0), (30.0, 30.0, 30.0)"), |
| 277 | + ], |
| 278 | +) |
| 279 | +@pytest.mark.skip(reason="skip until it gets updated") |
| 280 | +def test__state2readable__gives_expected_results_for_initial_states( |
| 281 | + env, lattice_system, expected_output |
| 282 | +): |
| 283 | + assert env.state2readable() == expected_output |
| 284 | + |
| 285 | + |
| 286 | +@pytest.mark.parametrize( |
| 287 | + "lattice_system, readable", |
| 288 | + [ |
| 289 | + (CUBIC, "(1.0, 1.0, 1.0), (90.0, 90.0, 90.0)"), |
| 290 | + (HEXAGONAL, "(1.0, 1.0, 1.0), (90.0, 90.0, 120.0)"), |
| 291 | + (MONOCLINIC, "(1.0, 1.0, 1.0), (90.0, 30.0, 90.0)"), |
| 292 | + (ORTHORHOMBIC, "(1.0, 1.0, 1.0), (90.0, 90.0, 90.0)"), |
| 293 | + (RHOMBOHEDRAL, "(1.0, 1.0, 1.0), (30.0, 30.0, 30.0)"), |
| 294 | + (TETRAGONAL, "(1.0, 1.0, 1.0), (90.0, 90.0, 90.0)"), |
| 295 | + (TRICLINIC, "(1.0, 1.0, 1.0), (30.0, 30.0, 30.0)"), |
| 296 | + ], |
| 297 | +) |
| 298 | +@pytest.mark.skip(reason="skip until it gets updated") |
| 299 | +def test__readable2state__gives_expected_results_for_initial_states( |
| 300 | + env, lattice_system, readable |
| 301 | +): |
| 302 | + assert env.readable2state(readable) == env.state |
| 303 | + |
| 304 | + |
| 305 | +@pytest.mark.parametrize( |
| 306 | + "lattice_system", |
| 307 | + [CUBIC, HEXAGONAL, MONOCLINIC, ORTHORHOMBIC, RHOMBOHEDRAL, TETRAGONAL, TRICLINIC], |
| 308 | +) |
| 309 | +def test__continuous_env_common(env, lattice_system): |
| 310 | + return common.test__continuous_env_common(env) |
0 commit comments