opensmfs / pycorrelate Goto Github PK

• Pycorrelate version: 0.3
• Python version: 3.7.3
• Operating System: Windows 10

Description

I was trying the examples from documentation from the below URL.
https://pycorrelate.readthedocs.io/en/latest/notebooks/pycorrelate-examples.html#Log-scale-bins-(base-10)

I aplogize for being naive here.

How do I interpret the value G? How does it says that "t" and "u" correlated?

Can I get G value between 0 and 1, I mean if value near to 1 then 't' and 'u' highly correlated.

• Pycorrelate version: 0.3
• Python version: 3.6.4
• Operating System: Windows

Description

When you want to use the correlation algorithms with pythonic lists instead of numpy arrays, the program crashes badly. This could easily be fixed with a check on whether the function arguments are lists, and calling np.array() in that case.

What I Did

from pycorrelate import pcorrelate
X = [1,2]
Y = [3,4]
edges = [1,2,3,4]
pcorrelate(X, Y, edges)

The resulting crash is:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Users\meassi\AppData\Local\Programs\Python\Python36\lib\site-packages\numba\dispatcher.py", line 401, in _compile_for_args
    error_rewrite(e, 'typing')
  File "C:\Users\meassi\AppData\Local\Programs\Python\Python36\lib\site-packages\numba\dispatcher.py", line 344, in error_rewrite
    reraise(type(e), e, None)
  File "C:\Users\meassi\AppData\Local\Programs\Python\Python36\lib\site-packages\numba\six.py", line 668, in reraise
    raise value.with_traceback(tb)
numba.errors.TypingError: Failed in nopython mode pipeline (step: nopython frontend)
�[1m�[1m�[1mInvalid use of Function(<function diff at 0x000001AD75F4C488>) with argument(s) of type(s): (reflected list(int64))
 * parameterized
�[1mIn definition 0:�[0m
�[1m    All templates rejected with literals.�[0m
�[1mIn definition 1:�[0m
�[1m    All templates rejected without literals.�[0m
�[1mThis error is usually caused by passing an argument of a type that is unsupported by the named function.�[0m�[0m
�[0m�[1m[1] During: resolving callee type: Function(<function diff at 0x000001AD75F4C488>)�[0m
�[0m�[1m[2] During: typing of call at C:\Users\meassi\AppData\Local\Programs\Python\Python36\lib\site-packages\pycorrelate\pycorrelate.py (111)
�[0m
�[1m
File "..\..\..\..\AppData\Local\Programs\Python\Python36\lib\site-packages\pycorrelate\pycorrelate.py", line 111:�[0m
�[1mdef pcorrelate(t, u, bins, normalize=False):
    <source elided>
        counts += imax - imin
�[1m    G = counts / np.diff(bins)
�[0m    �[1m^�[0m�[0m

This is not usually a problem with Numba itself but instead often caused by
the use of unsupported features or an issue in resolving types.

To see Python/NumPy features supported by the latest release of Numba visit:
http://numba.pydata.org/numba-doc/latest/reference/pysupported.html
and
http://numba.pydata.org/numba-doc/latest/reference/numpysupported.html

For more information about typing errors and how to debug them visit:
http://numba.pydata.org/numba-doc/latest/user/troubleshoot.html#my-code-doesn-t-compile

If you think your code should work with Numba, please report the error message
and traceback, along with a minimal reproducer at:
https://github.com/numba/numba/issues/new

The documentation for ucorrelate says the following: "The classical text-book linear cross-correlation between two signals defined at uniformly spaced intervals." I know that it is fairly obvious, but it would nevertheless be good to restate as follows: "The classical textbook linear cross-correlation between two signals defined at uniformly-spaced intervals with the same interval size."

Issue moved from https://github.com/tritemio/pycorrelate/issues/7. Original author @Phillip-M-Feldman.

• Pycorrelate version: 0.3
• Python version: 3.6
• Operating System: Windows 10

Description

I wrote a script to compare the runtime perforance of numpy.correlate and pycorrelate.ucorrelate. The input sequences have length 1e5, and the maximum lag of interest is 10,000. Since I can't specify a maximum lag with numpy.correlate, I used the mode='same' option.

Although numpy.correlate does the full cross-correlation, and was thus expected to be much slower, the results showed that it is actually faster (the following test results are based on 10 iterations in each case):

elapsed (wall time) | user CPU | system CPU

      numpy:  37.03 | 141.09   |   6.92
pycorrelate: 106.91 |  55.62   |  51.78

I'm wondering if you have any thoughts re. why the elapse time is worse for pycorrelate.ucorrelate. Also, have you considering using FFTs to speed up the calculations?

• Pycorrelate version: 0.3
• Python version: 3.6.4
• Operating System: Windows

Description

Correlation tends to be a time-consuming task, and it would be nice to have some kind of progress bar to get an estimate for how long it's going to take. This should be rather easy to implement, as we know the length of the arrays. I suggest adding it as a optional parameter to the functions, like

pcorrelate(..., progress=False)

This way, users don't get unexpected output in their terminal, but they can enable this useful feature if they want to. I'm willing to work on this.

In the docs, in the theory section I believe the first formula is lacking $\tau$ on the RHS.