FloatingPointError on ewm().std()

[chrism2671]

I'm not sure if this is a bug, but I'm pretty sure I shouldn't be seeing these warnings. Any clarification would be a real help, as I need to ensure the accuracy of the calculations in all circumstances (including situations where the deviation is effectively zero)

Paste-ready code:

import traceback
import warnings
import sys
import pandas as pd
import numpy as np
np.seterr(all='raise')
def warn_with_traceback(message, category, filename, lineno, file=None, line=None):
    traceback.print_stack()
    log = file if hasattr(file,'write') else sys.stderr
    log.write(warnings.formatwarning(message, category, filename, lineno, line))

warnings.showwarning = warn_with_traceback

pd.Series(np.random.randn(50)).pct_change().ewm(span=35, min_periods=35).std()

Expected output:
pd.Series of exponentially weighted standard deviation.

output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.5.1.final.0
python-bits: 64
OS: Linux
OS-release: 4.2.0-34-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8

pandas: 0.18.1
nose: 1.3.7
pip: 8.1.2
setuptools: 23.0.0
Cython: 0.24
numpy: 1.10.4
scipy: 0.17.1
statsmodels: 0.6.1
xarray: None
IPython: 4.2.0
sphinx: 1.4.1
patsy: 0.4.1
dateutil: 2.5.3
pytz: 2016.4
blosc: None
bottleneck: 1.0.0
tables: 3.2.2
numexpr: 2.5.2
matplotlib: 1.5.1
openpyxl: 2.3.2
xlrd: 1.0.0
xlwt: 1.1.2
xlsxwriter: 0.9.2
lxml: 3.6.0
bs4: 4.4.0
html5lib: 0.999
httplib2: None
apiclient: None
sqlalchemy: 1.0.13
pymysql: None
psycopg2: None
jinja2: 2.8
boto: None
pandas_datareader: 0.2.1

[jreback]

you are overriding pandas by doing this: np.seterr(all='raise')
pandas by definition will catch things like operating on nans and do the right thing. see #13109
so in theory a user could eventually change this. But to be honest its not very useful.

In [3]: np.random.seed(1234)

In [4]: pd.options.display.max_rows=8

In [5]: pd.Series(np.random.randn(50)).pct_change().ewm(span=35, min_periods=35).std()
Out[5]: 
0           NaN
1           NaN
2           NaN
3           NaN
        ...    
46    27.070401
47    26.253511
48    27.533031
49    26.703128
dtype: float64

[chrism2671]

Thanks, that's useful insight. I just wanted to make sure I wasn't seeing something unexpected.

[chrism2671]

One further question. The reason I was investigating this is because this warning is still getting thrown out when using the ThreadPoolExecutor to do this calculation (concurrent.futures), as:

/projects/anaconda3/lib/python3.5/site-packages/pandas/core/ops.py:716: RuntimeWarning: invalid value encountered in less
  result = getattr(x, name)(y)

I do believe the numbers are coming out as expected (I haven't checked meticulously), but did not want to suppress this warning without at least checking where it's coming from.