Adding test_randomseed to TestLQNash

[hsalberti]

Additional test to guarantee randomseed is working on lqnash based on issue #349

[pep8speaks]

Hello @hsalberti! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file quantecon/_lqnash.py:

Line 70:17: W291 trailing whitespace
Line 71:80: E501 line too long (90 > 79 characters)

Comment last updated at 2023-10-30 11:51:27 UTC

[jstac]

Thanks @hsalberti , much appreciated.

@oyamad , could you please review this when you have time?

[mmcky]

@hsalberti I have approved the CI to run. Thanks for your PR.

[mmcky]

@hsalberti the CI has errored with the following info

=========================== short test summary info ============================
FAILED quantecon/tests/test_lqnash.py::TestLQNash::test_randomseed - assert not True
 +  where True = <function array_equal at 0x7f6779380dc0>(array([[0.67919293, 0.        ]]), array([[0.67919293, 0.        ]]))
 +    where <function array_equal at 0x7f6779380dc0> = np.array_equal
============ 1 failed, 513 passed, 4 warnings in 264.67s (0:04:24) =============

[oyamad]

Looking at the code, randomness at

QuantEcon.py/quantecon/_lqnash.py

Lines 113 to 114 in d1e4d22

	F1 = random_state.standard_normal((k_1, n))
	F2 = random_state.standard_normal((k_2, n))

is almost irrelevant. The new F1 and F2 are deterministically redefined here

QuantEcon.py/quantecon/_lqnash.py

Lines 131 to 132 in d1e4d22

	F1 = solve(F1_left, F1_right)
	F2 = H2.dot(A)+G2.dot(W2.T) - np.dot(H2.dot(B1)+G2.dot(M2.T), F1)

and the old (random) F1 and F2 are used (as F10 and F20) only for the convergence check here

QuantEcon.py/quantecon/_lqnash.py

Line 144 in d1e4d22

dd = np.max(np.abs(F10 - F1)) + np.max(np.abs(F20 - F2))

Replacing these

QuantEcon.py/quantecon/_lqnash.py

Lines 113 to 114 in d1e4d22

	F1 = random_state.standard_normal((k_1, n))
	F2 = random_state.standard_normal((k_2, n))

by deterministic arrays with some large value will suffice.

So randomness in this function should be removed in the first place...

[hsalberti]

Okay, got it.
May I remove "randomstate" from nnash and its documentation?