Trying to find python numpy indexes with more than 2 logical conditions

Question

Why does

w = data[np.where(np.logical_and(data['RA']>=(ra_in-0.1), data['RA']<=(ra_in+0.1)))]

work, but

w = data[np.where(np.logical_and(data['RA']>=(ra_in-0.1), data['RA']<=(ra_in+0.1), data['DEC']>=(dec_in-0.1), data['DEC']<=(dec_in+0.1)  ))]

doesn't??? This gives an "ValueError: invalid number of arguments" error. The 'in' values are floats; the data are numpy.ndarrays.

Because logical_and takes two such vectors? All the remaining parameters are specifications on how to process the and. — willeM_ Van Onsem
– willeM_ Van Onsem, Commented Sep 16, 2017 at 19:36

willeM_ Van Onsem · Accepted Answer · 2017-09-16 19:50:04Z

2

The documentation of numpy.logical_and says:

numpy.logical_and(x1, x2, /, out=None, *, where=True, casting='same_kind', order='K', dtype=None, subok=True[, signature, extobj])

As you can see, except for additional parameters on how to perform the logical_and on the two array-like objects x1 and x2, there is no place for additional array-like objects. So in short logical_and only performs the logical and over two arrays.

You can use a cascade of logica_ands to perform the requested operation:

data[np.where(
    np.logical_and(
        np.logical_and(data['RA']>=(ra_in-0.1), data['RA']<=(ra_in+0.1),
        np.logical_and(data['DEC']>=(dec_in-0.1), data['DEC']<=(dec_in+0.1)
    ))]

So here we convert the invalid ~~logical_and(A,B,C,D)~~ into logical_and(logical_and(A,B),logical_and(C,D)).

Furthermore you can use the more elegant logical and &:

data[np.where(
    (data['RA']>=(ra_in-0.1)) & (data['RA']<=(ra_in+0.1)) & (data['DEC']>=(dec_in-0.1)) & (data['DEC']<=(dec_in+0.1))
    )]

Note that since & binds with higher priority, you need to add brackets. Note that however, we still have some cascading, since:

(data['RA']>=(ra_in-0.1)) & (data['RA']<=(ra_in+0.1)) & (data['DEC']>=(dec_in-0.1)) & (data['DEC']<=(dec_in+0.1))

is equivalent to:

(((data['RA']>=(ra_in-0.1)) & (data['RA']<=(ra_in+0.1))) & (data['DEC']>=(dec_in-0.1))) & (data['DEC']<=(dec_in+0.1))

or more abstract: A & B & C & D is equivalent to ((A & B) & C) & D.

edited Sep 16, 2017 at 19:50

answered Sep 16, 2017 at 19:40

willeM_ Van Onsem

482k33 gold badges483 silver badges624 bronze badges

Sign up to request clarification or add additional context in comments.

6 Comments

npross Over a year ago

Hmmm, (a) I could but (b) that code throws a syntax error, you need to close <=(ra_in+0.1)) and (dec_in+0.1)).

npross Over a year ago

but, is this seriously the best way to do numpy logic? Cascades of logical_and's ??!!

willeM_ Van Onsem Over a year ago

@npross: no, I would not use a logical_and at all, unless you need some of the additional options. You can use &.

npross Over a year ago

Nope. I get:: --------------------------------------------------------------------------- TypeError Traceback (most recent call last) <ipython-input-199-8839202fb04c> in <module>() 1 data[np.where( ----> 2 data['RA']>=(ra_in-0.1) & data['RA']<=(ra_in+0.1) & data['DEC']>=(dec_in-0.1) & data['DEC']<=(dec_in+0.1) 3 )] TypeError: ufunc 'bitwise_and' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''

willeM_ Van Onsem Over a year ago

@nposs: yes, but you should add brackets. The & operator has precedence over <=.

|

Collectives™ on Stack Overflow

Trying to find python numpy indexes with more than 2 logical conditions

1 Answer 1

6 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

6 Comments

Your Answer

Sign up or log in

Post as a guest

Related