Vectors

1D-array operations

Create

Action	Code	Details
Undefined, length n	np.empty(n)	Warning: don't use the initial values
Float zeros, length n	np.zeros(n)
Int zeros, length n	np.zeros(n, dtype=np.int64)
Ones, length n	np.ones(n)
True values, length n	np.full(n, fill_value=True)
False values, length n	np.full(n, fill_value=False)
Fill with value v	np.full(n, fill_value=v)
Increasing numbers [0, b-1]	np.arrange(b)
Increasing numbers [a, b]	np.arrange(a, b)
Numbers from a to b with step size s	np.arrange(a, b, s)
Linear range from a to b of length n	np.linspace(a, b, num=n)
From tuple	np.array(tuple)
From list	np.array(list)
From iter	np.fromiter(iter)
From iter (max length n)	np.fromiter(iter, count=n)
From two vectors	np.append(v1, v2)
From two vectors	np.concatenate((v1, v2))

Test

Action	Code	Details
Numpy array is vector	v.ndim == 1
Vectors are equal	np.array_equal(v, v2)
All numerical elements are equal	np.ptp(v) == 0
Logical type	v.dtype == np.bool_
Float type	v.dtype == np.float_
Integer type	v.dtype == np.int_
Contains nan	np.isnan(v).any()
Contains inf	np.isinf(v).any()
Contains value	value in v
Does not contain value	value not in v
All finite	np.isfinite(v).all()
All elements are equal to value	np.all(v == value)
All elements are equal to value	(v == value).all()
All numerical elements are close to value	np.all(np.isclose(v, value))

Extract

Action	Code	Details
Number of elements	len(v)
Hash	hash(v.data.tobytes())
Unique values	numpy.unique(v)
Unique values	set(v)
Number of unique values	len(numpy.unique(v))
Count per unique value	np.unique(v, return_counts=True)
Count non-zero values	np.count_nonzero(v)
Count per positive integer from [0, max(v)]	np.bincount(v)	Elements must be nonnegative ints
Index of first max element	np.argmax(v)	Returns index of NaN if present!
Index of first min element	np.argmin(v)	Returns index of NaN if present!

Aggregate

See the API docs for a complete list

Update

All operations are in-place

Transform

Action	Code	Details
Set first element	v[0] = value
Set last element	v[-1] = value
Set value of the _i_th element	v[i] = value
Fill with constant value	v[:] = value
Fill first n values	v[:n] = value

Order

Action	Code	Details
Sort elements ascending	v.sort()	NaNs are put last
Sort elements descending	v[::-1].sort()	Note that this puts NaNs first!

Derive

Map

Action	Code	Details
Clip (truncate) between [a, b ]	np.clip(v, a_min=a, a_max=b)
Bin index	np.digitize(v, bins)
Piecewise-linear interpolation of coordinate mapping xp -> yp	np.interp(v, xp, yp)	No option for extrapolation!
Piecewise-linear interpolation with extrapolation	f = scipy.interpolate.interp1d(xp, yp, fill_value = 'extrapolate') f(v)	Deprecated without replacement, lol
Ascending order (indices) of elements	np.argsort(v)
Descending order (indices) of elements	np.argsort(-v)

Reorder elements

Action	Code	Details
Reverse elements	v[::-1]
Reverse elements	np.flip(v)
Shift elements forwards, with roll-over	np.roll(v, 1)
Shift elements backwards, with roll-over	np.roll(v, -1)
Sort ascending	np.sort(v)	NaNs are last
Sort descending	v[np.argsort(-v)]	NaNs are last

Shrink

Action	Code	Details
Pairwise difference to next element	np.diff(v)	n-1 elements
Pairwise difference between elements with given lag	np.diff(v, n=lag)

Grow

Action	Code	Details
Pad with value	np.pad(v, pad_width=1, constant_values=v)
Pad with outer elements	np.pad(v, pad_width=1, mode='edge')
Replicate n times	np.repeat(v, n)
Append vector	np.append(v1, v2)
Append vectors	np.concatenate((v1, v2, v3))

Convert

Action	Code	Details
Bytes	v.tobytes()
Tuple	tuple(v)
List	v.tolist()
Set	set(v)