Test failures on master?

[amueller]

======================================================================
ERROR: test_basic (hyperopt.tests.test_pchoice.TestPChoice)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/andy/checkout/hyperopt/hyperopt/tests/test_pchoice.py", line 21, in test_basic
    nesto = hyperopt.pyll.stochastic.sample(space, rng=rng)
  File "/home/andy/checkout/hyperopt/hyperopt/pyll/stochastic.py", line 196, in sample
    return rec_eval(foo, **kwargs)
  File "/home/andy/checkout/hyperopt/hyperopt/pyll/base.py", line 874, in rec_eval
    rval_var = node.pos_args[switch_i + 1]
TypeError: only integer scalar arrays can be converted to a scalar index

======================================================================
ERROR: test_basic2 (hyperopt.tests.test_pchoice.TestPChoice)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/andy/checkout/hyperopt/hyperopt/tests/test_pchoice.py", line 45, in test_basic2
    nesto = hyperopt.pyll.stochastic.sample(space, rng=rng)
  File "/home/andy/checkout/hyperopt/hyperopt/pyll/stochastic.py", line 196, in sample
    return rec_eval(foo, **kwargs)
  File "/home/andy/checkout/hyperopt/hyperopt/pyll/base.py", line 874, in rec_eval
    rval_var = node.pos_args[switch_i + 1]
TypeError: only integer scalar arrays can be converted to a scalar index

======================================================================
ERROR: test_basic3 (hyperopt.tests.test_pchoice.TestPChoice)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/andy/checkout/hyperopt/hyperopt/tests/test_pchoice.py", line 68, in test_basic3
    nesto = hyperopt.pyll.stochastic.sample(space, rng=rng)
  File "/home/andy/checkout/hyperopt/hyperopt/pyll/stochastic.py", line 196, in sample
    return rec_eval(foo, **kwargs)
  File "/home/andy/checkout/hyperopt/hyperopt/pyll/base.py", line 874, in rec_eval
    rval_var = node.pos_args[switch_i + 1]
TypeError: only integer scalar arrays can be converted to a scalar index

======================================================================
ERROR: test_mu_is_used_correctly (hyperopt.tests.test_tpe.TestGMM1)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/andy/checkout/hyperopt/hyperopt/tests/test_tpe.py", line 79, in test_mu_is_used_correctly
    GMM1([1], [10.0], [0.0000001], rng=self.rng))
  File "/home/andy/checkout/hyperopt/hyperopt/tpe.py", line 79, in GMM1
    active = np.argmax(rng.multinomial(1, weights, (n_samples,)), axis=1)
  File "mtrand.pyx", line 4621, in mtrand.RandomState.multinomial (numpy/random/mtrand/mtrand.c:37893)
TypeError: 'numpy.float64' object cannot be interpreted as an integer

======================================================================
FAIL: test_q1lognormal (hyperopt.tests.test_anneal.TestItAtLeastSortOfWorks)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/andy/checkout/hyperopt/hyperopt/tests/test_domains.py", line 249, in test_q1lognormal
    self.work()
  File "/home/andy/checkout/hyperopt/hyperopt/tests/test_anneal.py", line 114, in work
    assert min(trials.losses()) < thresh
AssertionError

----------------------------------------------------------------------
Ran 191 tests in 51.085s

FAILED (SKIP=2, errors=4, failures=1)

[nouiz]

I think the problem is that your NumPy is to recent for hyperopt. Newer numpy always need fixes in Theano and we for sick type of errors. Le mer. 2 août 2017 14:49, Andreas Mueller <notifications@github.com> a écrit :

…

====================================================================== ERROR: test_basic (hyperopt.tests.test_pchoice.TestPChoice) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/andy/checkout/hyperopt/hyperopt/tests/test_pchoice.py", line 21, in test_basic nesto = hyperopt.pyll.stochastic.sample(space, rng=rng) File "/home/andy/checkout/hyperopt/hyperopt/pyll/stochastic.py", line 196, in sample return rec_eval(foo, **kwargs) File "/home/andy/checkout/hyperopt/hyperopt/pyll/base.py", line 874, in rec_eval rval_var = node.pos_args[switch_i + 1] TypeError: only integer scalar arrays can be converted to a scalar index ====================================================================== ERROR: test_basic2 (hyperopt.tests.test_pchoice.TestPChoice) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/andy/checkout/hyperopt/hyperopt/tests/test_pchoice.py", line 45, in test_basic2 nesto = hyperopt.pyll.stochastic.sample(space, rng=rng) File "/home/andy/checkout/hyperopt/hyperopt/pyll/stochastic.py", line 196, in sample return rec_eval(foo, **kwargs) File "/home/andy/checkout/hyperopt/hyperopt/pyll/base.py", line 874, in rec_eval rval_var = node.pos_args[switch_i + 1] TypeError: only integer scalar arrays can be converted to a scalar index ====================================================================== ERROR: test_basic3 (hyperopt.tests.test_pchoice.TestPChoice) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/andy/checkout/hyperopt/hyperopt/tests/test_pchoice.py", line 68, in test_basic3 nesto = hyperopt.pyll.stochastic.sample(space, rng=rng) File "/home/andy/checkout/hyperopt/hyperopt/pyll/stochastic.py", line 196, in sample return rec_eval(foo, **kwargs) File "/home/andy/checkout/hyperopt/hyperopt/pyll/base.py", line 874, in rec_eval rval_var = node.pos_args[switch_i + 1] TypeError: only integer scalar arrays can be converted to a scalar index ====================================================================== ERROR: test_mu_is_used_correctly (hyperopt.tests.test_tpe.TestGMM1) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/andy/checkout/hyperopt/hyperopt/tests/test_tpe.py", line 79, in test_mu_is_used_correctly GMM1([1], [10.0], [0.0000001], rng=self.rng)) File "/home/andy/checkout/hyperopt/hyperopt/tpe.py", line 79, in GMM1 active = np.argmax(rng.multinomial(1, weights, (n_samples,)), axis=1) File "mtrand.pyx", line 4621, in mtrand.RandomState.multinomial (numpy/random/mtrand/mtrand.c:37893) TypeError: 'numpy.float64' object cannot be interpreted as an integer ====================================================================== FAIL: test_q1lognormal (hyperopt.tests.test_anneal.TestItAtLeastSortOfWorks) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/andy/checkout/hyperopt/hyperopt/tests/test_domains.py", line 249, in test_q1lognormal self.work() File "/home/andy/checkout/hyperopt/hyperopt/tests/test_anneal.py", line 114, in work assert min(trials.losses()) < thresh AssertionError ---------------------------------------------------------------------- Ran 191 tests in 51.085s FAILED (SKIP=2, errors=4, failures=1) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#315>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AALC-2nQGUcHJ0A1l6YLlPHwahtq4U62ks5sUMTDgaJpZM4OrgTo> .

[amueller]

Thanks for your reply. I'm using numpy 1.11.1.
Sorry, the second sentence didn't parse.

The first three errors are because switch_i is a (1, ) array, but I didn't know where a good place would be for the conversion - I could do it where the error happens, but that might hide some issue further up, as I'm not familiar with the code. The 4th seems like a simple casting issues, but the last one seems like an optimization issue, so I have no ideas.

[nouiz]

I'm not able to help more as I don't use hyperopt. I was just mantioning something that could help. Using older numpy could fix some of those problems, as hyperopt was made with other numpy then 1.11. I can't help more.

…

On Wed, Aug 2, 2017 at 2:49 PM Andreas Mueller ***@***.***> wrote: ====================================================================== ERROR: test_basic (hyperopt.tests.test_pchoice.TestPChoice) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/andy/checkout/hyperopt/hyperopt/tests/test_pchoice.py", line 21, in test_basic nesto = hyperopt.pyll.stochastic.sample(space, rng=rng) File "/home/andy/checkout/hyperopt/hyperopt/pyll/stochastic.py", line 196, in sample return rec_eval(foo, **kwargs) File "/home/andy/checkout/hyperopt/hyperopt/pyll/base.py", line 874, in rec_eval rval_var = node.pos_args[switch_i + 1] TypeError: only integer scalar arrays can be converted to a scalar index ====================================================================== ERROR: test_basic2 (hyperopt.tests.test_pchoice.TestPChoice) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/andy/checkout/hyperopt/hyperopt/tests/test_pchoice.py", line 45, in test_basic2 nesto = hyperopt.pyll.stochastic.sample(space, rng=rng) File "/home/andy/checkout/hyperopt/hyperopt/pyll/stochastic.py", line 196, in sample return rec_eval(foo, **kwargs) File "/home/andy/checkout/hyperopt/hyperopt/pyll/base.py", line 874, in rec_eval rval_var = node.pos_args[switch_i + 1] TypeError: only integer scalar arrays can be converted to a scalar index ====================================================================== ERROR: test_basic3 (hyperopt.tests.test_pchoice.TestPChoice) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/andy/checkout/hyperopt/hyperopt/tests/test_pchoice.py", line 68, in test_basic3 nesto = hyperopt.pyll.stochastic.sample(space, rng=rng) File "/home/andy/checkout/hyperopt/hyperopt/pyll/stochastic.py", line 196, in sample return rec_eval(foo, **kwargs) File "/home/andy/checkout/hyperopt/hyperopt/pyll/base.py", line 874, in rec_eval rval_var = node.pos_args[switch_i + 1] TypeError: only integer scalar arrays can be converted to a scalar index ====================================================================== ERROR: test_mu_is_used_correctly (hyperopt.tests.test_tpe.TestGMM1) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/andy/checkout/hyperopt/hyperopt/tests/test_tpe.py", line 79, in test_mu_is_used_correctly GMM1([1], [10.0], [0.0000001], rng=self.rng)) File "/home/andy/checkout/hyperopt/hyperopt/tpe.py", line 79, in GMM1 active = np.argmax(rng.multinomial(1, weights, (n_samples,)), axis=1) File "mtrand.pyx", line 4621, in mtrand.RandomState.multinomial (numpy/random/mtrand/mtrand.c:37893) TypeError: 'numpy.float64' object cannot be interpreted as an integer ====================================================================== FAIL: test_q1lognormal (hyperopt.tests.test_anneal.TestItAtLeastSortOfWorks) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/andy/checkout/hyperopt/hyperopt/tests/test_domains.py", line 249, in test_q1lognormal self.work() File "/home/andy/checkout/hyperopt/hyperopt/tests/test_anneal.py", line 114, in work assert min(trials.losses()) < thresh AssertionError ---------------------------------------------------------------------- Ran 191 tests in 51.085s FAILED (SKIP=2, errors=4, failures=1) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#315>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AALC-2nQGUcHJ0A1l6YLlPHwahtq4U62ks5sUMTDgaJpZM4OrgTo> .

[aldanor]

Any ideas on this? The first four seem to be a lack of a simple conversion or casting (somewhere appropriate) as @amueller mentioned, but what about the last one?

(On a side note, is hyperopt actively maintained by anyone? I could probably try fixing a few things if there was anyone to review and merge.)

[michaelmior]

Closing since this refers to an outdated version of hyperopt. Feel free to reopen if you're still having any issues.