# numpy.poly1d () in Python

| | | | | | | | | | |
Note

This forms part of the old polynomial API. Since version 1.4, the new polynomial API defined in numpy.polynomial is preferred. A summary of the differences can be found in the transition guide.

The `numpy.poly1d()` function allows to define a polynomial function. It therefore makes it straightforward to use natural operations on polynomials.

It is a convenience class, used to encapsulate "natural‚Äù operations on polynomials so that said operations may take on their customary form in code.

Syntax: numpy.poly1d(arr, root, var)
Parameters :
arr : [array_like] The polynomial coefficients are given in decreasing order of powers. If the second parameter (root) is set to True then array values are the roots of the polynomial equation.

root : [bool, optional] True means polynomial roots. Default is False.
var : variable like x, y, z that we need in polynomial [default is x].

Arguments :
c : Polynomial coefficient.
coef : Polynomial coefficient.
coefficients : Polynomial coefficient.
order : Order or degree of polynomial.
o : Order or degree of polynomial.
r : Polynomial root.
roots : Polynomial root.

Return: Polynomial and the operation applied

## Numpy poly1d Examples

### np.poly1d example #1

```def _break_points(num, den):
"""Extract break points over real axis and gains given these locations"""
# type: (np.poly1d, np.poly1d) -> (np.array, np.array)
dnum = num.deriv(m=1)
dden = den.deriv(m=1)
polynom = den * dnum - num * dden
real_break_pts = polynom.r
# don’t care about infinite break points
real_break_pts = real_break_pts[num(real_break_pts) != 0]
k_break = -den(real_break_pts) / num(real_break_pts)
idx = k_break >= 0   # only positives gains
k_break = k_break[idx]
real_break_pts = real_break_pts[idx]
if len(k_break) == 0:
k_break = [0]
real_break_pts = den.roots
return k_break, real_break_pts
```

### np.poly1d example #2

```def test_poly1d_math(self):
# here we use some simple coeffs to make calculations easier
p = np.poly1d([1., 2, 4])
q = np.poly1d([4., 2, 1])
assert_equal(p/q, (np.poly1d([0.25]), np.poly1d([1.5, 3.75])))
assert_equal(p.integ(), np.poly1d([1/3, 1., 4., 0.]))
assert_equal(p.integ(1), np.poly1d([1/3, 1., 4., 0.]))

p = np.poly1d([1., 2, 3])
q = np.poly1d([3., 2, 1])
assert_equal(p * q, np.poly1d([3., 8., 14., 8., 3.]))
assert_equal(p + q, np.poly1d([4., 4., 4.]))
assert_equal(p - q, np.poly1d([-2., 0., 2.]))
assert_equal(p ** 4, np.poly1d([1., 8., 36., 104., 214., 312., 324., 216., 81.]))
assert_equal(p(q), np.poly1d([9., 12., 16., 8., 6.]))
assert_equal(q(p), np.poly1d([3., 12., 32., 40., 34.]))
assert_equal(p.deriv(), np.poly1d([2., 2.]))
assert_equal(p.deriv(2), np.poly1d([2.]))
assert_equal(np.polydiv(np.poly1d([1, 0, -1]), np.poly1d([1, 1])),
(np.poly1d([1., -1.]), np.poly1d([0.])))
```

### np.poly1d example #3

```def test_poly1d_str_and_repr(self):
p = np.poly1d([1., 2, 3])
assert_equal(repr(p), ’poly1d([1., 2., 3.])’)
assert_equal(str(p),
’   2
’
’1 x + 2 x + 3’)

q = np.poly1d([3., 2, 1])
assert_equal(repr(q), ’poly1d([3., 2., 1.])’)
assert_equal(str(q),
’   2
’
’3 x + 2 x + 1’)

r = np.poly1d([1.89999 + 2j, -3j, -5.12345678, 2 + 1j])
assert_equal(str(r),
’            3      2
’
’(1.9 + 2j) x - 3j x - 5.123 x + (2 + 1j)’)

assert_equal(str(np.poly1d([-3, -2, -1])),
’    2
’
’-3 x - 2 x - 1’)
```

### np.poly1d example #4

```def data_analysis(e_ph, flux, method="least"):

if method == "least":
coeffs = np.polyfit(x=e_ph, y=flux, deg=11)
polynom = np.poly1d(coeffs)

x = np.linspace(e_ph[0], e_ph[-1], num=100)
pd = np.polyder(polynom, m=1)
indx = np.argmax(np.abs(pd(x)))
eph_c = x[indx]

pd2 = np.polyder(polynom, m=2)
p2_roots = np.roots(pd2)
p2_roots = p2_roots[p2_roots[:].imag == 0]
p2_roots = np.real(p2_roots)
Eph_fin = find_nearest(p2_roots,eph_c)
return Eph_fin, polynom

elif method == "new method":
pass

#plt.plot(Etotal, total, "ro")
#plt.plot(x, polynom(x))
```

### np.poly1d example #5

```def _systopoly1d(sys):
"""Extract numerator and denominator polynomails for a system"""
# Allow inputs from the signal processing toolbox
if (isinstance(sys, scipy.signal.lti)):
nump = sys.num
denp = sys.den

else:
# Convert to a transfer function, if needed
sys = _convert_to_transfer_function(sys)

# Make sure we have a SISO system
if (sys.inputs > 1 or sys.outputs > 1):
raise ControlMIMONotImplemented()

# Start by extracting the numerator and denominator from system object
nump = sys.num[0][0]
denp = sys.den[0][0]

# Check to see if num, den are already polynomials; otherwise convert
if (not isinstance(nump, poly1d)):
nump = poly1d(nump)

if (not isinstance(denp, poly1d)):
denp = poly1d(denp)

return (nump, denp)
```

### np.poly1d example #6

```def quadraticInterpolation(valueList2d, numDegrees, n,
startTime=None, endTime=None):
’’’
Generates a series of points on a smooth curve that cross the given points

numDegrees - the degrees of the fitted polynomial
- the curve gets weird if this value is too high for the input
n - number of points to output
startTime/endTime/n - n points will be generated at evenly spaced
intervals between startTime and endTime
’’’
_numpyCheck()

x, y = zip(*valueList2d)

if startTime is None:
startTime = x[0]
if endTime is None:
endTime = x[-1]

polyFunc = np.poly1d(np.polyfit(x, y, numDegrees))

newX = np.linspace(startTime, endTime, n)

retList = [(n, polyFunc(n)) for n in newX]

return retList
```

### np.poly1d example #7

```def fit_y(self, X, Y, x1, x2):
len(X) != 0
# if X only include one point, the function will get line y=Y[0]
if np.sum(X == X[0]) == len(X):
return Y[0], Y[0]
p = np.poly1d(np.polyfit(X, Y, 1))
return p(x1), p(x2)
```

### np.poly1d example #8

```def remove_linear_BG_XAS_preedge(
xmcd_data, scanparams, process_parameters=None, process_number=-1
):
"""Should remove a linear bg based on the preedge average"""
preedge_spectrum = get_preedge_spectrum(process_parameters, xmcd_data)

preedge_poly = np.poly1d(
np.polyfit(preedge_spectrum["Energy"], preedge_spectrum["XAS"], 1)
)

xas_bg = preedge_poly(xmcd_data["Energy"])

for xas in ["XAS+", "XAS-", "XAS"]:
xmcd_data[xas] -= xas_bg

return (xmcd_data, {"xas_bg_poly_coeffs": " ".join(map(str, preedge_poly.coeffs))})
```

### np.poly1d example #9

```def fit_y(self, X, Y, x1, x2):
len(X) != 0
# if X only include one point, the function will get line y=Y[0]
if np.sum(X == X[0]) == len(X):
return Y[0], Y[0]
p = np.poly1d(np.polyfit(X, Y, 1))
return p(x1), p(x2)
```

### np.poly1d example #10

```def __init__(self, roots, weights=None, hn=1.0, kn=1.0, wfunc=None, limits=None, monic=0,eval_func=None):
np.poly1d.__init__(self, roots, r=1)
equiv_weights = [weights[k] / wfunc(roots[k]) for k in range(len(roots))]
self.__dict__[’weights’] = np.array(list(zip(roots,weights,equiv_weights)))
self.__dict__[’weight_func’] = wfunc
self.__dict__[’limits’] = limits
mu = sqrt(hn)
if monic:
evf = eval_func
if evf:
eval_func = lambda x: evf(x)/kn
mu = mu / abs(kn)
kn = 1.0
self.__dict__[’normcoef’] = mu
self.__dict__[’coeffs’] *= kn

# Note: eval_func will be discarded on arithmetic
self.__dict__[’_eval_func’] = eval_func
```

## How do I merge two dictionaries in a single expression (taking union of dictionaries)?

### Question by Carl Meyer

I have two Python dictionaries, and I want to write a single expression that returns these two dictionaries, merged (i.e. taking the union). The `update()` method would be what I need, if it returned its result instead of modifying a dictionary in-place.

``````>>> x = {"a": 1, "b": 2}
>>> y = {"b": 10, "c": 11}
>>> z = x.update(y)
>>> print(z)
None
>>> x
{"a": 1, "b": 10, "c": 11}
``````

How can I get that final merged dictionary in `z`, not `x`?

(To be extra-clear, the last-one-wins conflict-handling of `dict.update()` is what I"m looking for as well.)

## How can I merge two Python dictionaries in a single expression?

For dictionaries `x` and `y`, `z` becomes a shallowly-merged dictionary with values from `y` replacing those from `x`.

• In Python 3.9.0 or greater (released 17 October 2020): PEP-584, discussed here, was implemented and provides the simplest method:

``````z = x | y          # NOTE: 3.9+ ONLY
``````
• In Python 3.5 or greater:

``````z = {**x, **y}
``````
• In Python 2, (or 3.4 or lower) write a function:

``````def merge_two_dicts(x, y):
z.update(y)    # modifies z with keys and values of y
return z
``````

and now:

``````z = merge_two_dicts(x, y)
``````

### Explanation

Say you have two dictionaries and you want to merge them into a new dictionary without altering the original dictionaries:

``````x = {"a": 1, "b": 2}
y = {"b": 3, "c": 4}
``````

The desired result is to get a new dictionary (`z`) with the values merged, and the second dictionary"s values overwriting those from the first.

``````>>> z
{"a": 1, "b": 3, "c": 4}
``````

A new syntax for this, proposed in PEP 448 and available as of Python 3.5, is

``````z = {**x, **y}
``````

And it is indeed a single expression.

Note that we can merge in with literal notation as well:

``````z = {**x, "foo": 1, "bar": 2, **y}
``````

and now:

``````>>> z
{"a": 1, "b": 3, "foo": 1, "bar": 2, "c": 4}
``````

It is now showing as implemented in the release schedule for 3.5, PEP 478, and it has now made its way into the What"s New in Python 3.5 document.

However, since many organizations are still on Python 2, you may wish to do this in a backward-compatible way. The classically Pythonic way, available in Python 2 and Python 3.0-3.4, is to do this as a two-step process:

``````z = x.copy()
z.update(y) # which returns None since it mutates z
``````

In both approaches, `y` will come second and its values will replace `x`"s values, thus `b` will point to `3` in our final result.

## Not yet on Python 3.5, but want a single expression

If you are not yet on Python 3.5 or need to write backward-compatible code, and you want this in a single expression, the most performant while the correct approach is to put it in a function:

``````def merge_two_dicts(x, y):
"""Given two dictionaries, merge them into a new dict as a shallow copy."""
z = x.copy()
z.update(y)
return z
``````

and then you have a single expression:

``````z = merge_two_dicts(x, y)
``````

You can also make a function to merge an arbitrary number of dictionaries, from zero to a very large number:

``````def merge_dicts(*dict_args):
"""
Given any number of dictionaries, shallow copy and merge into a new dict,
precedence goes to key-value pairs in latter dictionaries.
"""
result = {}
for dictionary in dict_args:
result.update(dictionary)
return result
``````

This function will work in Python 2 and 3 for all dictionaries. e.g. given dictionaries `a` to `g`:

``````z = merge_dicts(a, b, c, d, e, f, g)
``````

and key-value pairs in `g` will take precedence over dictionaries `a` to `f`, and so on.

Don"t use what you see in the formerly accepted answer:

``````z = dict(x.items() + y.items())
``````

In Python 2, you create two lists in memory for each dict, create a third list in memory with length equal to the length of the first two put together, and then discard all three lists to create the dict. In Python 3, this will fail because you"re adding two `dict_items` objects together, not two lists -

``````>>> c = dict(a.items() + b.items())
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
TypeError: unsupported operand type(s) for +: "dict_items" and "dict_items"
``````

and you would have to explicitly create them as lists, e.g. `z = dict(list(x.items()) + list(y.items()))`. This is a waste of resources and computation power.

Similarly, taking the union of `items()` in Python 3 (`viewitems()` in Python 2.7) will also fail when values are unhashable objects (like lists, for example). Even if your values are hashable, since sets are semantically unordered, the behavior is undefined in regards to precedence. So don"t do this:

``````>>> c = dict(a.items() | b.items())
``````

This example demonstrates what happens when values are unhashable:

``````>>> x = {"a": []}
>>> y = {"b": []}
>>> dict(x.items() | y.items())
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
TypeError: unhashable type: "list"
``````

Here"s an example where `y` should have precedence, but instead the value from `x` is retained due to the arbitrary order of sets:

``````>>> x = {"a": 2}
>>> y = {"a": 1}
>>> dict(x.items() | y.items())
{"a": 2}
``````

Another hack you should not use:

``````z = dict(x, **y)
``````

This uses the `dict` constructor and is very fast and memory-efficient (even slightly more so than our two-step process) but unless you know precisely what is happening here (that is, the second dict is being passed as keyword arguments to the dict constructor), it"s difficult to read, it"s not the intended usage, and so it is not Pythonic.

Here"s an example of the usage being remediated in django.

Dictionaries are intended to take hashable keys (e.g. `frozenset`s or tuples), but this method fails in Python 3 when keys are not strings.

``````>>> c = dict(a, **b)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
TypeError: keyword arguments must be strings
``````

From the mailing list, Guido van Rossum, the creator of the language, wrote:

I am fine with declaring dict({}, **{1:3}) illegal, since after all it is abuse of the ** mechanism.

and

Apparently dict(x, **y) is going around as "cool hack" for "call x.update(y) and return x". Personally, I find it more despicable than cool.

It is my understanding (as well as the understanding of the creator of the language) that the intended usage for `dict(**y)` is for creating dictionaries for readability purposes, e.g.:

``````dict(a=1, b=10, c=11)
``````

``````{"a": 1, "b": 10, "c": 11}
``````

Despite what Guido says, `dict(x, **y)` is in line with the dict specification, which btw. works for both Python 2 and 3. The fact that this only works for string keys is a direct consequence of how keyword parameters work and not a short-coming of dict. Nor is using the ** operator in this place an abuse of the mechanism, in fact, ** was designed precisely to pass dictionaries as keywords.

Again, it doesn"t work for 3 when keys are not strings. The implicit calling contract is that namespaces take ordinary dictionaries, while users must only pass keyword arguments that are strings. All other callables enforced it. `dict` broke this consistency in Python 2:

``````>>> foo(**{("a", "b"): None})
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
TypeError: foo() keywords must be strings
>>> dict(**{("a", "b"): None})
{("a", "b"): None}
``````

This inconsistency was bad given other implementations of Python (PyPy, Jython, IronPython). Thus it was fixed in Python 3, as this usage could be a breaking change.

I submit to you that it is malicious incompetence to intentionally write code that only works in one version of a language or that only works given certain arbitrary constraints.

`dict(x.items() + y.items())` is still the most readable solution for Python 2. Readability counts.

My response: `merge_two_dicts(x, y)` actually seems much clearer to me, if we"re actually concerned about readability. And it is not forward compatible, as Python 2 is increasingly deprecated.

`{**x, **y}` does not seem to handle nested dictionaries. the contents of nested keys are simply overwritten, not merged [...] I ended up being burnt by these answers that do not merge recursively and I was surprised no one mentioned it. In my interpretation of the word "merging" these answers describe "updating one dict with another", and not merging.

Yes. I must refer you back to the question, which is asking for a shallow merge of two dictionaries, with the first"s values being overwritten by the second"s - in a single expression.

Assuming two dictionaries of dictionaries, one might recursively merge them in a single function, but you should be careful not to modify the dictionaries from either source, and the surest way to avoid that is to make a copy when assigning values. As keys must be hashable and are usually therefore immutable, it is pointless to copy them:

``````from copy import deepcopy

def dict_of_dicts_merge(x, y):
z = {}
overlapping_keys = x.keys() & y.keys()
for key in overlapping_keys:
z[key] = dict_of_dicts_merge(x[key], y[key])
for key in x.keys() - overlapping_keys:
z[key] = deepcopy(x[key])
for key in y.keys() - overlapping_keys:
z[key] = deepcopy(y[key])
return z
``````

Usage:

``````>>> x = {"a":{1:{}}, "b": {2:{}}}
>>> y = {"b":{10:{}}, "c": {11:{}}}
>>> dict_of_dicts_merge(x, y)
{"b": {2: {}, 10: {}}, "a": {1: {}}, "c": {11: {}}}
``````

Coming up with contingencies for other value types is far beyond the scope of this question, so I will point you at my answer to the canonical question on a "Dictionaries of dictionaries merge".

## Less Performant But Correct Ad-hocs

These approaches are less performant, but they will provide correct behavior. They will be much less performant than `copy` and `update` or the new unpacking because they iterate through each key-value pair at a higher level of abstraction, but they do respect the order of precedence (latter dictionaries have precedence)

You can also chain the dictionaries manually inside a dict comprehension:

``````{k: v for d in dicts for k, v in d.items()} # iteritems in Python 2.7
``````

or in Python 2.6 (and perhaps as early as 2.4 when generator expressions were introduced):

``````dict((k, v) for d in dicts for k, v in d.items()) # iteritems in Python 2
``````

`itertools.chain` will chain the iterators over the key-value pairs in the correct order:

``````from itertools import chain
z = dict(chain(x.items(), y.items())) # iteritems in Python 2
``````

## Performance Analysis

I"m only going to do the performance analysis of the usages known to behave correctly. (Self-contained so you can copy and paste yourself.)

``````from timeit import repeat
from itertools import chain

x = dict.fromkeys("abcdefg")
y = dict.fromkeys("efghijk")

def merge_two_dicts(x, y):
z = x.copy()
z.update(y)
return z

min(repeat(lambda: {**x, **y}))
min(repeat(lambda: merge_two_dicts(x, y)))
min(repeat(lambda: {k: v for d in (x, y) for k, v in d.items()}))
min(repeat(lambda: dict(chain(x.items(), y.items()))))
min(repeat(lambda: dict(item for d in (x, y) for item in d.items())))
``````

In Python 3.8.1, NixOS:

``````>>> min(repeat(lambda: {**x, **y}))
1.0804965235292912
>>> min(repeat(lambda: merge_two_dicts(x, y)))
1.636518670246005
>>> min(repeat(lambda: {k: v for d in (x, y) for k, v in d.items()}))
3.1779992282390594
>>> min(repeat(lambda: dict(chain(x.items(), y.items()))))
2.740647904574871
>>> min(repeat(lambda: dict(item for d in (x, y) for item in d.items())))
4.266070580109954
``````
``````\$ uname -a
Linux nixos 4.19.113 #1-NixOS SMP Wed Mar 25 07:06:15 UTC 2020 x86_64 GNU/Linux
``````

## Resources on Dictionaries

In your case, what you can do is:

``````z = dict(list(x.items()) + list(y.items()))
``````

This will, as you want it, put the final dict in `z`, and make the value for key `b` be properly overridden by the second (`y`) dict"s value:

``````>>> x = {"a":1, "b": 2}
>>> y = {"b":10, "c": 11}
>>> z = dict(list(x.items()) + list(y.items()))
>>> z
{"a": 1, "c": 11, "b": 10}

``````

If you use Python 2, you can even remove the `list()` calls. To create z:

``````>>> z = dict(x.items() + y.items())
>>> z
{"a": 1, "c": 11, "b": 10}
``````

If you use Python version 3.9.0a4 or greater, then you can directly use:

``````x = {"a":1, "b": 2}
y = {"b":10, "c": 11}
z = x | y
print(z)
``````
``````{"a": 1, "c": 11, "b": 10}
``````

An alternative:

``````z = x.copy()
z.update(y)
``````

Another, more concise, option:

``````z = dict(x, **y)
``````

Note: this has become a popular answer, but it is important to point out that if `y` has any non-string keys, the fact that this works at all is an abuse of a CPython implementation detail, and it does not work in Python 3, or in PyPy, IronPython, or Jython. Also, Guido is not a fan. So I can"t recommend this technique for forward-compatible or cross-implementation portable code, which really means it should be avoided entirely.

This probably won"t be a popular answer, but you almost certainly do not want to do this. If you want a copy that"s a merge, then use copy (or deepcopy, depending on what you want) and then update. The two lines of code are much more readable - more Pythonic - than the single line creation with .items() + .items(). Explicit is better than implicit.

In addition, when you use .items() (pre Python 3.0), you"re creating a new list that contains the items from the dict. If your dictionaries are large, then that is quite a lot of overhead (two large lists that will be thrown away as soon as the merged dict is created). update() can work more efficiently, because it can run through the second dict item-by-item.

In terms of time:

``````>>> timeit.Timer("dict(x, **y)", "x = dict(zip(range(1000), range(1000)))
y=dict(zip(range(1000,2000), range(1000,2000)))").timeit(100000)
15.52571702003479
>>> timeit.Timer("temp = x.copy()
temp.update(y)", "x = dict(zip(range(1000), range(1000)))
y=dict(zip(range(1000,2000), range(1000,2000)))").timeit(100000)
15.694622993469238
>>> timeit.Timer("dict(x.items() + y.items())", "x = dict(zip(range(1000), range(1000)))
y=dict(zip(range(1000,2000), range(1000,2000)))").timeit(100000)
41.484580039978027
``````

IMO the tiny slowdown between the first two is worth it for the readability. In addition, keyword arguments for dictionary creation was only added in Python 2.3, whereas copy() and update() will work in older versions.

## How do I copy a string to the clipboard?

### Question by Dancrew32

I"m trying to make a basic Windows application that builds a string out of user input and then adds it to the clipboard. How do I copy a string to the clipboard using Python?

Actually, `pywin32` and `ctypes` seem to be an overkill for this simple task. `Tkinter` is a cross-platform GUI framework, which ships with Python by default and has clipboard accessing methods along with other cool stuff.

If all you need is to put some text to system clipboard, this will do it:

``````from Tkinter import Tk
r = Tk()
r.withdraw()
r.clipboard_clear()
r.clipboard_append("i can has clipboardz?")
r.update() # now it stays on the clipboard after the window is closed
r.destroy()
``````

And that"s all, no need to mess around with platform-specific third-party libraries.

If you are using Python 3, replace `TKinter` with `tkinter`.

## Python script to copy text to clipboard

I just need a python script that copies text to the clipboard.

After the script gets executed i need the output of the text to be pasted to another source. Is it possible to write a python script that does this job?

See Pyperclip. Example (taken from Pyperclip site):

``````import pyperclip
pyperclip.copy("The text to be copied to the clipboard.")
spam = pyperclip.paste()
``````

Also, see Xerox. But it appears to have more dependencies.

## How to apply gradient clipping in TensorFlow?

Considering the example code.

I would like to know How to apply gradient clipping on this network on the RNN where there is a possibility of exploding gradients.

``````tf.clip_by_value(t, clip_value_min, clip_value_max, name=None)
``````

This is an example that could be used but where do I introduce this ? In the def of RNN

``````    lstm_cell = rnn_cell.BasicLSTMCell(n_hidden, forget_bias=1.0)
# Split data because rnn cell needs a list of inputs for the RNN inner loop
_X = tf.split(0, n_steps, _X) # n_steps
tf.clip_by_value(_X, -1, 1, name=None)
``````

But this doesn"t make sense as the tensor _X is the input and not the grad what is to be clipped?

Do I have to define my own Optimizer for this or is there a simpler option?

Gradient clipping needs to happen after computing the gradients, but before applying them to update the model"s parameters. In your example, both of those things are handled by the `AdamOptimizer.minimize()` method.

In order to clip your gradients you"ll need to explicitly compute, clip, and apply them as described in this section in TensorFlow"s API documentation. Specifically you"ll need to substitute the call to the `minimize()` method with something like the following:

``````optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate)
``````

Despite what seems to be popular, you probably want to clip the whole gradient by its global norm:

``````optimizer = tf.train.AdamOptimizer(1e-3)
``````

Clipping each gradient matrix individually changes their relative scale but is also possible:

``````optimizer = tf.train.AdamOptimizer(1e-3)
``````

In TensorFlow 2, a tape computes the gradients, the optimizers come from Keras, and we don"t need to store the update op because it runs automatically without passing it to a session:

``````optimizer = tf.keras.optimizers.Adam(1e-3)
# ...
loss = ...
variables = ...
``````

## Finding the index of an item in a list

Given a list `["foo", "bar", "baz"]` and an item in the list `"bar"`, how do I get its index (`1`) in Python?

``````>>> ["foo", "bar", "baz"].index("bar")
1
``````

Reference: Data Structures > More on Lists

# Caveats follow

Note that while this is perhaps the cleanest way to answer the question as asked, `index` is a rather weak component of the `list` API, and I can"t remember the last time I used it in anger. It"s been pointed out to me in the comments that because this answer is heavily referenced, it should be made more complete. Some caveats about `list.index` follow. It is probably worth initially taking a look at the documentation for it:

``````list.index(x[, start[, end]])
``````

Return zero-based index in the list of the first item whose value is equal to x. Raises a `ValueError` if there is no such item.

The optional arguments start and end are interpreted as in the slice notation and are used to limit the search to a particular subsequence of the list. The returned index is computed relative to the beginning of the full sequence rather than the start argument.

## Linear time-complexity in list length

An `index` call checks every element of the list in order, until it finds a match. If your list is long, and you don"t know roughly where in the list it occurs, this search could become a bottleneck. In that case, you should consider a different data structure. Note that if you know roughly where to find the match, you can give `index` a hint. For instance, in this snippet, `l.index(999_999, 999_990, 1_000_000)` is roughly five orders of magnitude faster than straight `l.index(999_999)`, because the former only has to search 10 entries, while the latter searches a million:

``````>>> import timeit
>>> timeit.timeit("l.index(999_999)", setup="l = list(range(0, 1_000_000))", number=1000)
9.356267921015387
>>> timeit.timeit("l.index(999_999, 999_990, 1_000_000)", setup="l = list(range(0, 1_000_000))", number=1000)
0.0004404920036904514

``````

## Only returns the index of the first match to its argument

A call to `index` searches through the list in order until it finds a match, and stops there. If you expect to need indices of more matches, you should use a list comprehension, or generator expression.

``````>>> [1, 1].index(1)
0
>>> [i for i, e in enumerate([1, 2, 1]) if e == 1]
[0, 2]
>>> g = (i for i, e in enumerate([1, 2, 1]) if e == 1)
>>> next(g)
0
>>> next(g)
2
``````

Most places where I once would have used `index`, I now use a list comprehension or generator expression because they"re more generalizable. So if you"re considering reaching for `index`, take a look at these excellent Python features.

## Throws if element not present in list

A call to `index` results in a `ValueError` if the item"s not present.

``````>>> [1, 1].index(2)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: 2 is not in list
``````

If the item might not be present in the list, you should either

1. Check for it first with `item in my_list` (clean, readable approach), or
2. Wrap the `index` call in a `try/except` block which catches `ValueError` (probably faster, at least when the list to search is long, and the item is usually present.)

One thing that is really helpful in learning Python is to use the interactive help function:

``````>>> help(["foo", "bar", "baz"])
Help on list object:

class list(object)
...

|
|  index(...)
|      L.index(value, [start, [stop]]) -> integer -- return first index of value
|
``````

which will often lead you to the method you are looking for.

The majority of answers explain how to find a single index, but their methods do not return multiple indexes if the item is in the list multiple times. Use `enumerate()`:

``````for i, j in enumerate(["foo", "bar", "baz"]):
if j == "bar":
print(i)
``````

The `index()` function only returns the first occurrence, while `enumerate()` returns all occurrences.

As a list comprehension:

``````[i for i, j in enumerate(["foo", "bar", "baz"]) if j == "bar"]
``````

Here"s also another small solution with `itertools.count()` (which is pretty much the same approach as enumerate):

``````from itertools import izip as zip, count # izip for maximum efficiency
[i for i, j in zip(count(), ["foo", "bar", "baz"]) if j == "bar"]
``````

This is more efficient for larger lists than using `enumerate()`:

``````\$ python -m timeit -s "from itertools import izip as zip, count" "[i for i, j in zip(count(), ["foo", "bar", "baz"]*500) if j == "bar"]"
10000 loops, best of 3: 174 usec per loop
\$ python -m timeit "[i for i, j in enumerate(["foo", "bar", "baz"]*500) if j == "bar"]"
10000 loops, best of 3: 196 usec per loop
``````

To get all indexes:

``````indexes = [i for i,x in enumerate(xs) if x == "foo"]
``````

`index()` returns the first index of value!

| index(...)
| L.index(value, [start, [stop]]) -> integer -- return first index of value

``````def all_indices(value, qlist):
indices = []
idx = -1
while True:
try:
idx = qlist.index(value, idx+1)
indices.append(idx)
except ValueError:
break
return indices

all_indices("foo", ["foo";"bar";"baz";"foo"])
``````

## Why is it string.join(list) instead of list.join(string)?

### Question by Evan Fosmark

This has always confused me. It seems like this would be nicer:

``````my_list = ["Hello", "world"]
print(my_list.join("-"))
# Produce: "Hello-world"
``````

Than this:

``````my_list = ["Hello", "world"]
print("-".join(my_list))
# Produce: "Hello-world"
``````

Is there a specific reason it is like this?

It"s because any iterable can be joined (e.g, list, tuple, dict, set), but its contents and the "joiner" must be strings.

For example:

``````"_".join(["welcome", "to", "stack", "overflow"])
"_".join(("welcome", "to", "stack", "overflow"))
``````
``````"welcome_to_stack_overflow"
``````

Using something other than strings will raise the following error:

``````TypeError: sequence item 0: expected str instance, int found
``````

This was discussed in the String methods... finally thread in the Python-Dev achive, and was accepted by Guido. This thread began in Jun 1999, and `str.join` was included in Python 1.6 which was released in Sep 2000 (and supported Unicode). Python 2.0 (supported `str` methods including `join`) was released in Oct 2000.

• There were four options proposed in this thread:
• `str.join(seq)`
• `seq.join(str)`
• `seq.reduce(str)`
• `join` as a built-in function
• Guido wanted to support not only `list`s and `tuple`s, but all sequences/iterables.
• `seq.reduce(str)` is difficult for newcomers.
• `seq.join(str)` introduces unexpected dependency from sequences to str/unicode.
• `join()` as a built-in function would support only specific data types. So using a built-in namespace is not good. If `join()` supports many datatypes, creating an optimized implementation would be difficult, if implemented using the `__add__` method then it would ve `O(n¬≤)`.
• The separator string (`sep`) should not be omitted. Explicit is better than implicit.

Here are some additional thoughts (my own, and my friend"s):

• Unicode support was coming, but it was not final. At that time UTF-8 was the most likely about to replace UCS2/4. To calculate total buffer length of UTF-8 strings it needs to know character coding rule.
• At that time, Python had already decided on a common sequence interface rule where a user could create a sequence-like (iterable) class. But Python didn"t support extending built-in types until 2.2. At that time it was difficult to provide basic `iterable` class (which is mentioned in another comment).

Guido"s decision is recorded in a historical mail, deciding on `str.join(seq)`:

Funny, but it does seem right! Barry, go for it...
Guido van Rossum

Because the `join()` method is in the string class, instead of the list class?

I agree it looks funny.

Historical note. When I first learned Python, I expected join to be a method of a list, which would take the delimiter as an argument. Lots of people feel the same way, and there‚Äôs a story behind the join method. Prior to Python 1.6, strings didn‚Äôt have all these useful methods. There was a separate string module which contained all the string functions; each function took a string as its first argument. The functions were deemed important enough to put onto the strings themselves, which made sense for functions like lower, upper, and split. But many hard-core Python programmers objected to the new join method, arguing that it should be a method of the list instead, or that it shouldn‚Äôt move at all but simply stay a part of the old string module (which still has lots of useful stuff in it). I use the new join method exclusively, but you will see code written either way, and if it really bothers you, you can use the old string.join function instead.

--- Mark Pilgrim, Dive into Python

## join list of lists in python

### Question by Kozyarchuk

Is the a short syntax for joining a list of lists into a single list( or iterator) in python?

For example I have a list as follows and I want to iterate over a,b and c.

``````x = [["a";"b"], ["c"]]
``````

The best I can come up with is as follows.

``````result = []
[ result.extend(el) for el in x]

for el in result:
print el
``````

``````import itertools
a = [["a","b"], ["c"]]
print(list(itertools.chain.from_iterable(a)))
``````

``````x = [["a";"b"], ["c"]]

result = sum(x, [])
``````

## Meaning of @classmethod and @staticmethod for beginner?

### Question by user1632861

Could someone explain to me the meaning of `@classmethod` and `@staticmethod` in python? I need to know the difference and the meaning.

As far as I understand, `@classmethod` tells a class that it"s a method which should be inherited into subclasses, or... something. However, what"s the point of that? Why not just define the class method without adding `@classmethod` or `@staticmethod` or any `@` definitions?

tl;dr: when should I use them, why should I use them, and how should I use them?

Though `classmethod` and `staticmethod` are quite similar, there"s a slight difference in usage for both entities: `classmethod` must have a reference to a class object as the first parameter, whereas `staticmethod` can have no parameters at all.

## Example

``````class Date(object):

def __init__(self, day=0, month=0, year=0):
self.day = day
self.month = month
self.year = year

@classmethod
def from_string(cls, date_as_string):
day, month, year = map(int, date_as_string.split("-"))
date1 = cls(day, month, year)
return date1

@staticmethod
def is_date_valid(date_as_string):
day, month, year = map(int, date_as_string.split("-"))
return day <= 31 and month <= 12 and year <= 3999

date2 = Date.from_string("11-09-2012")
is_date = Date.is_date_valid("11-09-2012")
``````

## Explanation

Let"s assume an example of a class, dealing with date information (this will be our boilerplate):

``````class Date(object):

def __init__(self, day=0, month=0, year=0):
self.day = day
self.month = month
self.year = year
``````

This class obviously could be used to store information about certain dates (without timezone information; let"s assume all dates are presented in UTC).

Here we have `__init__`, a typical initializer of Python class instances, which receives arguments as a typical `instancemethod`, having the first non-optional argument (`self`) that holds a reference to a newly created instance.

Class Method

We have some tasks that can be nicely done using `classmethod`s.

Let"s assume that we want to create a lot of `Date` class instances having date information coming from an outer source encoded as a string with format "dd-mm-yyyy". Suppose we have to do this in different places in the source code of our project.

So what we must do here is:

1. Parse a string to receive day, month and year as three integer variables or a 3-item tuple consisting of that variable.
2. Instantiate `Date` by passing those values to the initialization call.

This will look like:

``````day, month, year = map(int, string_date.split("-"))
date1 = Date(day, month, year)
``````

For this purpose, C++ can implement such a feature with overloading, but Python lacks this overloading. Instead, we can use `classmethod`. Let"s create another "constructor".

``````    @classmethod
def from_string(cls, date_as_string):
day, month, year = map(int, date_as_string.split("-"))
date1 = cls(day, month, year)
return date1

date2 = Date.from_string("11-09-2012")
``````

Let"s look more carefully at the above implementation, and review what advantages we have here:

1. We"ve implemented date string parsing in one place and it"s reusable now.
2. Encapsulation works fine here (if you think that you could implement string parsing as a single function elsewhere, this solution fits the OOP paradigm far better).
3. `cls` is an object that holds the class itself, not an instance of the class. It"s pretty cool because if we inherit our `Date` class, all children will have `from_string` defined also.

Static method

What about `staticmethod`? It"s pretty similar to `classmethod` but doesn"t take any obligatory parameters (like a class method or instance method does).

Let"s look at the next use case.

We have a date string that we want to validate somehow. This task is also logically bound to the `Date` class we"ve used so far, but doesn"t require instantiation of it.

Here is where `staticmethod` can be useful. Let"s look at the next piece of code:

``````    @staticmethod
def is_date_valid(date_as_string):
day, month, year = map(int, date_as_string.split("-"))
return day <= 31 and month <= 12 and year <= 3999

# usage:
is_date = Date.is_date_valid("11-09-2012")
``````

So, as we can see from usage of `staticmethod`, we don"t have any access to what the class is---it"s basically just a function, called syntactically like a method, but without access to the object and its internals (fields and another methods), while classmethod does.

Rostyslav Dzinko"s answer is very appropriate. I thought I could highlight one other reason you should choose `@classmethod` over `@staticmethod` when you are creating an additional constructor.

In the example above, Rostyslav used the `@classmethod` `from_string` as a Factory to create `Date` objects from otherwise unacceptable parameters. The same can be done with `@staticmethod` as is shown in the code below:

``````class Date:
def __init__(self, month, day, year):
self.month = month
self.day   = day
self.year  = year

def display(self):
return "{0}-{1}-{2}".format(self.month, self.day, self.year)

@staticmethod
def millenium(month, day):
return Date(month, day, 2000)

new_year = Date(1, 1, 2013)               # Creates a new Date object
millenium_new_year = Date.millenium(1, 1) # also creates a Date object.

# Proof:
new_year.display()           # "1-1-2013"
millenium_new_year.display() # "1-1-2000"

isinstance(new_year, Date) # True
isinstance(millenium_new_year, Date) # True
``````

Thus both `new_year` and `millenium_new_year` are instances of the `Date` class.

But, if you observe closely, the Factory process is hard-coded to create `Date` objects no matter what. What this means is that even if the `Date` class is subclassed, the subclasses will still create plain `Date` objects (without any properties of the subclass). See that in the example below:

``````class DateTime(Date):
def display(self):
return "{0}-{1}-{2} - 00:00:00PM".format(self.month, self.day, self.year)

datetime1 = DateTime(10, 10, 1990)
datetime2 = DateTime.millenium(10, 10)

isinstance(datetime1, DateTime) # True
isinstance(datetime2, DateTime) # False

datetime1.display() # returns "10-10-1990 - 00:00:00PM"
datetime2.display() # returns "10-10-2000" because it"s not a DateTime object but a Date object. Check the implementation of the millenium method on the Date class for more details.
``````

`datetime2` is not an instance of `DateTime`? WTF? Well, that"s because of the `@staticmethod` decorator used.

In most cases, this is undesired. If what you want is a Factory method that is aware of the class that called it, then `@classmethod` is what you need.

Rewriting `Date.millenium` as (that"s the only part of the above code that changes):

``````@classmethod
def millenium(cls, month, day):
return cls(month, day, 2000)
``````

ensures that the `class` is not hard-coded but rather learnt. `cls` can be any subclass. The resulting `object` will rightly be an instance of `cls`.
Let"s test that out:

``````datetime1 = DateTime(10, 10, 1990)
datetime2 = DateTime.millenium(10, 10)

isinstance(datetime1, DateTime) # True
isinstance(datetime2, DateTime) # True

datetime1.display() # "10-10-1990 - 00:00:00PM"
datetime2.display() # "10-10-2000 - 00:00:00PM"
``````

The reason is, as you know by now, that `@classmethod` was used instead of `@staticmethod`

`@classmethod` means: when this method is called, we pass the class as the first argument instead of the instance of that class (as we normally do with methods). This means you can use the class and its properties inside that method rather than a particular instance.

`@staticmethod` means: when this method is called, we don"t pass an instance of the class to it (as we normally do with methods). This means you can put a function inside a class but you can"t access the instance of that class (this is useful when your method does not use the instance).

## What is the meaning of single and double underscore before an object name?

Can someone please explain the exact meaning of having single and double leading underscores before an object"s name in Python, and the difference between both?

Also, does that meaning stay the same regardless of whether the object in question is a variable, a function, a method, etc.?

## Single Underscore

Names, in a class, with a leading underscore are simply to indicate to other programmers that the attribute or method is intended to be private. However, nothing special is done with the name itself.

To quote PEP-8:

_single_leading_underscore: weak "internal use" indicator. E.g. `from M import *` does not import objects whose name starts with an underscore.

## Double Underscore (Name Mangling)

From the Python docs:

Any identifier of the form `__spam` (at least two leading underscores, at most one trailing underscore) is textually replaced with `_classname__spam`, where `classname` is the current class name with leading underscore(s) stripped. This mangling is done without regard to the syntactic position of the identifier, so it can be used to define class-private instance and class variables, methods, variables stored in globals, and even variables stored in instances. private to this class on instances of other classes.

And a warning from the same page:

Name mangling is intended to give classes an easy way to define ‚Äúprivate‚Äù instance variables and methods, without having to worry about instance variables defined by derived classes, or mucking with instance variables by code outside the class. Note that the mangling rules are designed mostly to avoid accidents; it still is possible for a determined soul to access or modify a variable that is considered private.

## Example

``````>>> class MyClass():
...     def __init__(self):
...             self.__superprivate = "Hello"
...             self._semiprivate = ", world!"
...
>>> mc = MyClass()
>>> print mc.__superprivate
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: myClass instance has no attribute "__superprivate"
>>> print mc._semiprivate
, world!
>>> print mc.__dict__
{"_MyClass__superprivate": "Hello", "_semiprivate": ", world!"}
``````

`__foo__`: this is just a convention, a way for the Python system to use names that won"t conflict with user names.

`_foo`: this is just a convention, a way for the programmer to indicate that the variable is private (whatever that means in Python).

`__foo`: this has real meaning: the interpreter replaces this name with `_classname__foo` as a way to ensure that the name will not overlap with a similar name in another class.

No other form of underscores have meaning in the Python world.

There"s no difference between class, variable, global, etc in these conventions.

## How to get all subsets of a set? (powerset)

Given a set

``````{0, 1, 2, 3}
``````

How can I produce the subsets:

``````[set(),
{0},
{1},
{2},
{3},
{0, 1},
{0, 2},
{0, 3},
{1, 2},
{1, 3},
{2, 3},
{0, 1, 2},
{0, 1, 3},
{0, 2, 3},
{1, 2, 3},
{0, 1, 2, 3}]
``````

The Python `itertools` page has exactly a `powerset` recipe for this:

``````from itertools import chain, combinations

def powerset(iterable):
"powerset([1,2,3]) --> () (1,) (2,) (3,) (1,2) (1,3) (2,3) (1,2,3)"
s = list(iterable)
return chain.from_iterable(combinations(s, r) for r in range(len(s)+1))
``````

Output:

``````>>> list(powerset("abcd"))
[(), ("a",), ("b",), ("c",), ("d",), ("a", "b"), ("a", "c"), ("a", "d"), ("b", "c"), ("b", "d"), ("c", "d"), ("a", "b", "c"), ("a", "b", "d"), ("a", "c", "d"), ("b", "c", "d"), ("a", "b", "c", "d")]
``````

If you don"t like that empty tuple at the beginning, you can just change the `range` statement to `range(1, len(s)+1)` to avoid a 0-length combination.

## How do I merge two dictionaries in a single expression (taking union of dictionaries)?

### Question by Carl Meyer

I have two Python dictionaries, and I want to write a single expression that returns these two dictionaries, merged (i.e. taking the union). The `update()` method would be what I need, if it returned its result instead of modifying a dictionary in-place.

``````>>> x = {"a": 1, "b": 2}
>>> y = {"b": 10, "c": 11}
>>> z = x.update(y)
>>> print(z)
None
>>> x
{"a": 1, "b": 10, "c": 11}
``````

How can I get that final merged dictionary in `z`, not `x`?

(To be extra-clear, the last-one-wins conflict-handling of `dict.update()` is what I"m looking for as well.)

## How can I merge two Python dictionaries in a single expression?

For dictionaries `x` and `y`, `z` becomes a shallowly-merged dictionary with values from `y` replacing those from `x`.

• In Python 3.9.0 or greater (released 17 October 2020): PEP-584, discussed here, was implemented and provides the simplest method:

``````z = x | y          # NOTE: 3.9+ ONLY
``````
• In Python 3.5 or greater:

``````z = {**x, **y}
``````
• In Python 2, (or 3.4 or lower) write a function:

``````def merge_two_dicts(x, y):
z.update(y)    # modifies z with keys and values of y
return z
``````

and now:

``````z = merge_two_dicts(x, y)
``````

### Explanation

Say you have two dictionaries and you want to merge them into a new dictionary without altering the original dictionaries:

``````x = {"a": 1, "b": 2}
y = {"b": 3, "c": 4}
``````

The desired result is to get a new dictionary (`z`) with the values merged, and the second dictionary"s values overwriting those from the first.

``````>>> z
{"a": 1, "b": 3, "c": 4}
``````

A new syntax for this, proposed in PEP 448 and available as of Python 3.5, is

``````z = {**x, **y}
``````

And it is indeed a single expression.

Note that we can merge in with literal notation as well:

``````z = {**x, "foo": 1, "bar": 2, **y}
``````

and now:

``````>>> z
{"a": 1, "b": 3, "foo": 1, "bar": 2, "c": 4}
``````

It is now showing as implemented in the release schedule for 3.5, PEP 478, and it has now made its way into the What"s New in Python 3.5 document.

However, since many organizations are still on Python 2, you may wish to do this in a backward-compatible way. The classically Pythonic way, available in Python 2 and Python 3.0-3.4, is to do this as a two-step process:

``````z = x.copy()
z.update(y) # which returns None since it mutates z
``````

In both approaches, `y` will come second and its values will replace `x`"s values, thus `b` will point to `3` in our final result.

## Not yet on Python 3.5, but want a single expression

If you are not yet on Python 3.5 or need to write backward-compatible code, and you want this in a single expression, the most performant while the correct approach is to put it in a function:

``````def merge_two_dicts(x, y):
"""Given two dictionaries, merge them into a new dict as a shallow copy."""
z = x.copy()
z.update(y)
return z
``````

and then you have a single expression:

``````z = merge_two_dicts(x, y)
``````

You can also make a function to merge an arbitrary number of dictionaries, from zero to a very large number:

``````def merge_dicts(*dict_args):
"""
Given any number of dictionaries, shallow copy and merge into a new dict,
precedence goes to key-value pairs in latter dictionaries.
"""
result = {}
for dictionary in dict_args:
result.update(dictionary)
return result
``````

This function will work in Python 2 and 3 for all dictionaries. e.g. given dictionaries `a` to `g`:

``````z = merge_dicts(a, b, c, d, e, f, g)
``````

and key-value pairs in `g` will take precedence over dictionaries `a` to `f`, and so on.

Don"t use what you see in the formerly accepted answer:

``````z = dict(x.items() + y.items())
``````

In Python 2, you create two lists in memory for each dict, create a third list in memory with length equal to the length of the first two put together, and then discard all three lists to create the dict. In Python 3, this will fail because you"re adding two `dict_items` objects together, not two lists -

``````>>> c = dict(a.items() + b.items())
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
TypeError: unsupported operand type(s) for +: "dict_items" and "dict_items"
``````

and you would have to explicitly create them as lists, e.g. `z = dict(list(x.items()) + list(y.items()))`. This is a waste of resources and computation power.

Similarly, taking the union of `items()` in Python 3 (`viewitems()` in Python 2.7) will also fail when values are unhashable objects (like lists, for example). Even if your values are hashable, since sets are semantically unordered, the behavior is undefined in regards to precedence. So don"t do this:

``````>>> c = dict(a.items() | b.items())
``````

This example demonstrates what happens when values are unhashable:

``````>>> x = {"a": []}
>>> y = {"b": []}
>>> dict(x.items() | y.items())
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
TypeError: unhashable type: "list"
``````

Here"s an example where `y` should have precedence, but instead the value from `x` is retained due to the arbitrary order of sets:

``````>>> x = {"a": 2}
>>> y = {"a": 1}
>>> dict(x.items() | y.items())
{"a": 2}
``````

Another hack you should not use:

``````z = dict(x, **y)
``````

This uses the `dict` constructor and is very fast and memory-efficient (even slightly more so than our two-step process) but unless you know precisely what is happening here (that is, the second dict is being passed as keyword arguments to the dict constructor), it"s difficult to read, it"s not the intended usage, and so it is not Pythonic.

Here"s an example of the usage being remediated in django.

Dictionaries are intended to take hashable keys (e.g. `frozenset`s or tuples), but this method fails in Python 3 when keys are not strings.

``````>>> c = dict(a, **b)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
TypeError: keyword arguments must be strings
``````

From the mailing list, Guido van Rossum, the creator of the language, wrote:

I am fine with declaring dict({}, **{1:3}) illegal, since after all it is abuse of the ** mechanism.

and

Apparently dict(x, **y) is going around as "cool hack" for "call x.update(y) and return x". Personally, I find it more despicable than cool.

It is my understanding (as well as the understanding of the creator of the language) that the intended usage for `dict(**y)` is for creating dictionaries for readability purposes, e.g.:

``````dict(a=1, b=10, c=11)
``````

``````{"a": 1, "b": 10, "c": 11}
``````

Despite what Guido says, `dict(x, **y)` is in line with the dict specification, which btw. works for both Python 2 and 3. The fact that this only works for string keys is a direct consequence of how keyword parameters work and not a short-coming of dict. Nor is using the ** operator in this place an abuse of the mechanism, in fact, ** was designed precisely to pass dictionaries as keywords.

Again, it doesn"t work for 3 when keys are not strings. The implicit calling contract is that namespaces take ordinary dictionaries, while users must only pass keyword arguments that are strings. All other callables enforced it. `dict` broke this consistency in Python 2:

``````>>> foo(**{("a", "b"): None})
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
TypeError: foo() keywords must be strings
>>> dict(**{("a", "b"): None})
{("a", "b"): None}
``````

This inconsistency was bad given other implementations of Python (PyPy, Jython, IronPython). Thus it was fixed in Python 3, as this usage could be a breaking change.

I submit to you that it is malicious incompetence to intentionally write code that only works in one version of a language or that only works given certain arbitrary constraints.

`dict(x.items() + y.items())` is still the most readable solution for Python 2. Readability counts.

My response: `merge_two_dicts(x, y)` actually seems much clearer to me, if we"re actually concerned about readability. And it is not forward compatible, as Python 2 is increasingly deprecated.

`{**x, **y}` does not seem to handle nested dictionaries. the contents of nested keys are simply overwritten, not merged [...] I ended up being burnt by these answers that do not merge recursively and I was surprised no one mentioned it. In my interpretation of the word "merging" these answers describe "updating one dict with another", and not merging.

Yes. I must refer you back to the question, which is asking for a shallow merge of two dictionaries, with the first"s values being overwritten by the second"s - in a single expression.

Assuming two dictionaries of dictionaries, one might recursively merge them in a single function, but you should be careful not to modify the dictionaries from either source, and the surest way to avoid that is to make a copy when assigning values. As keys must be hashable and are usually therefore immutable, it is pointless to copy them:

``````from copy import deepcopy

def dict_of_dicts_merge(x, y):
z = {}
overlapping_keys = x.keys() & y.keys()
for key in overlapping_keys:
z[key] = dict_of_dicts_merge(x[key], y[key])
for key in x.keys() - overlapping_keys:
z[key] = deepcopy(x[key])
for key in y.keys() - overlapping_keys:
z[key] = deepcopy(y[key])
return z
``````

Usage:

``````>>> x = {"a":{1:{}}, "b": {2:{}}}
>>> y = {"b":{10:{}}, "c": {11:{}}}
>>> dict_of_dicts_merge(x, y)
{"b": {2: {}, 10: {}}, "a": {1: {}}, "c": {11: {}}}
``````

Coming up with contingencies for other value types is far beyond the scope of this question, so I will point you at my answer to the canonical question on a "Dictionaries of dictionaries merge".

## Less Performant But Correct Ad-hocs

These approaches are less performant, but they will provide correct behavior. They will be much less performant than `copy` and `update` or the new unpacking because they iterate through each key-value pair at a higher level of abstraction, but they do respect the order of precedence (latter dictionaries have precedence)

You can also chain the dictionaries manually inside a dict comprehension:

``````{k: v for d in dicts for k, v in d.items()} # iteritems in Python 2.7
``````

or in Python 2.6 (and perhaps as early as 2.4 when generator expressions were introduced):

``````dict((k, v) for d in dicts for k, v in d.items()) # iteritems in Python 2
``````

`itertools.chain` will chain the iterators over the key-value pairs in the correct order:

``````from itertools import chain
z = dict(chain(x.items(), y.items())) # iteritems in Python 2
``````

## Performance Analysis

I"m only going to do the performance analysis of the usages known to behave correctly. (Self-contained so you can copy and paste yourself.)

``````from timeit import repeat
from itertools import chain

x = dict.fromkeys("abcdefg")
y = dict.fromkeys("efghijk")

def merge_two_dicts(x, y):
z = x.copy()
z.update(y)
return z

min(repeat(lambda: {**x, **y}))
min(repeat(lambda: merge_two_dicts(x, y)))
min(repeat(lambda: {k: v for d in (x, y) for k, v in d.items()}))
min(repeat(lambda: dict(chain(x.items(), y.items()))))
min(repeat(lambda: dict(item for d in (x, y) for item in d.items())))
``````

In Python 3.8.1, NixOS:

``````>>> min(repeat(lambda: {**x, **y}))
1.0804965235292912
>>> min(repeat(lambda: merge_two_dicts(x, y)))
1.636518670246005
>>> min(repeat(lambda: {k: v for d in (x, y) for k, v in d.items()}))
3.1779992282390594
>>> min(repeat(lambda: dict(chain(x.items(), y.items()))))
2.740647904574871
>>> min(repeat(lambda: dict(item for d in (x, y) for item in d.items())))
4.266070580109954
``````
``````\$ uname -a
Linux nixos 4.19.113 #1-NixOS SMP Wed Mar 25 07:06:15 UTC 2020 x86_64 GNU/Linux
``````

## Resources on Dictionaries

In your case, what you can do is:

``````z = dict(list(x.items()) + list(y.items()))
``````

This will, as you want it, put the final dict in `z`, and make the value for key `b` be properly overridden by the second (`y`) dict"s value:

``````>>> x = {"a":1, "b": 2}
>>> y = {"b":10, "c": 11}
>>> z = dict(list(x.items()) + list(y.items()))
>>> z
{"a": 1, "c": 11, "b": 10}

``````

If you use Python 2, you can even remove the `list()` calls. To create z:

``````>>> z = dict(x.items() + y.items())
>>> z
{"a": 1, "c": 11, "b": 10}
``````

If you use Python version 3.9.0a4 or greater, then you can directly use:

``````x = {"a":1, "b": 2}
y = {"b":10, "c": 11}
z = x | y
print(z)
``````
``````{"a": 1, "c": 11, "b": 10}
``````

An alternative:

``````z = x.copy()
z.update(y)
``````

Another, more concise, option:

``````z = dict(x, **y)
``````

Note: this has become a popular answer, but it is important to point out that if `y` has any non-string keys, the fact that this works at all is an abuse of a CPython implementation detail, and it does not work in Python 3, or in PyPy, IronPython, or Jython. Also, Guido is not a fan. So I can"t recommend this technique for forward-compatible or cross-implementation portable code, which really means it should be avoided entirely.

This probably won"t be a popular answer, but you almost certainly do not want to do this. If you want a copy that"s a merge, then use copy (or deepcopy, depending on what you want) and then update. The two lines of code are much more readable - more Pythonic - than the single line creation with .items() + .items(). Explicit is better than implicit.

In addition, when you use .items() (pre Python 3.0), you"re creating a new list that contains the items from the dict. If your dictionaries are large, then that is quite a lot of overhead (two large lists that will be thrown away as soon as the merged dict is created). update() can work more efficiently, because it can run through the second dict item-by-item.

In terms of time:

``````>>> timeit.Timer("dict(x, **y)", "x = dict(zip(range(1000), range(1000)))
y=dict(zip(range(1000,2000), range(1000,2000)))").timeit(100000)
15.52571702003479
>>> timeit.Timer("temp = x.copy()
temp.update(y)", "x = dict(zip(range(1000), range(1000)))
y=dict(zip(range(1000,2000), range(1000,2000)))").timeit(100000)
15.694622993469238
>>> timeit.Timer("dict(x.items() + y.items())", "x = dict(zip(range(1000), range(1000)))
y=dict(zip(range(1000,2000), range(1000,2000)))").timeit(100000)
41.484580039978027
``````

IMO the tiny slowdown between the first two is worth it for the readability. In addition, keyword arguments for dictionary creation was only added in Python 2.3, whereas copy() and update() will work in older versions.

## How to execute a program or call a system command?

### Question by alan lai

How do you call an external command (as if I"d typed it at the Unix shell or Windows command prompt) from within a Python script?

Use the `subprocess` module in the standard library:

``````import subprocess
subprocess.run(["ls", "-l"])
``````

The advantage of `subprocess.run` over `os.system` is that it is more flexible (you can get the `stdout`, `stderr`, the "real" status code, better error handling, etc...).

Even the documentation for `os.system` recommends using `subprocess` instead:

The `subprocess` module provides more powerful facilities for spawning new processes and retrieving their results; using that module is preferable to using this function. See the Replacing Older Functions with the subprocess Module section in the `subprocess` documentation for some helpful recipes.

On Python 3.4 and earlier, use `subprocess.call` instead of `.run`:

``````subprocess.call(["ls", "-l"])
``````

Here"s a summary of the ways to call external programs and the advantages and disadvantages of each:

1. `os.system("some_command with args")` passes the command and arguments to your system"s shell. This is nice because you can actually run multiple commands at once in this manner and set up pipes and input/output redirection. For example:

``````os.system("some_command < input_file | another_command > output_file")
``````

However, while this is convenient, you have to manually handle the escaping of shell characters such as spaces, et cetera. On the other hand, this also lets you run commands which are simply shell commands and not actually external programs. See the documentation.

2. `stream = os.popen("some_command with args")` will do the same thing as `os.system` except that it gives you a file-like object that you can use to access standard input/output for that process. There are 3 other variants of popen that all handle the i/o slightly differently. If you pass everything as a string, then your command is passed to the shell; if you pass them as a list then you don"t need to worry about escaping anything. See the documentation.

3. The `Popen` class of the `subprocess` module. This is intended as a replacement for `os.popen`, but has the downside of being slightly more complicated by virtue of being so comprehensive. For example, you"d say:

``````print subprocess.Popen("echo Hello World", shell=True, stdout=subprocess.PIPE).stdout.read()
``````

``````print os.popen("echo Hello World").read()
``````

but it is nice to have all of the options there in one unified class instead of 4 different popen functions. See the documentation.

4. The `call` function from the `subprocess` module. This is basically just like the `Popen` class and takes all of the same arguments, but it simply waits until the command completes and gives you the return code. For example:

``````return_code = subprocess.call("echo Hello World", shell=True)
``````
5. If you"re on Python 3.5 or later, you can use the new `subprocess.run` function, which is a lot like the above but even more flexible and returns a `CompletedProcess` object when the command finishes executing.

6. The `os` module also has all of the fork/exec/spawn functions that you"d have in a C program, but I don"t recommend using them directly.

The `subprocess` module should probably be what you use.

Finally, please be aware that for all methods where you pass the final command to be executed by the shell as a string and you are responsible for escaping it. There are serious security implications if any part of the string that you pass can not be fully trusted. For example, if a user is entering some/any part of the string. If you are unsure, only use these methods with constants. To give you a hint of the implications consider this code:

``````print subprocess.Popen("echo %s " % user_input, stdout=PIPE).stdout.read()
``````

and imagine that the user enters something "`my mama didnt love me && rm -rf /`" which could erase the whole filesystem.

Typical implementation:

``````import subprocess

p = subprocess.Popen("ls", shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
print line,
retval = p.wait()
``````

You are free to do what you want with the `stdout` data in the pipe. In fact, you can simply omit those parameters (`stdout=` and `stderr=`) and it"ll behave like `os.system()`.

Some hints on detaching the child process from the calling one (starting the child process in background).

Suppose you want to start a long task from a CGI script. That is, the child process should live longer than the CGI script execution process.

The classical example from the subprocess module documentation is:

``````import subprocess
import sys

# Some code here

pid = subprocess.Popen([sys.executable, "longtask.py"]) # Call subprocess

# Some more code here
``````

The idea here is that you do not want to wait in the line "call subprocess" until the longtask.py is finished. But it is not clear what happens after the line "some more code here" from the example.

My target platform was FreeBSD, but the development was on Windows, so I faced the problem on Windows first.

On Windows (Windows XP), the parent process will not finish until the longtask.py has finished its work. It is not what you want in a CGI script. The problem is not specific to Python; in the PHP community the problems are the same.

The solution is to pass DETACHED_PROCESS Process Creation Flag to the underlying CreateProcess function in Windows API. If you happen to have installed pywin32, you can import the flag from the win32process module, otherwise you should define it yourself:

``````DETACHED_PROCESS = 0x00000008

creationflags=DETACHED_PROCESS).pid
``````

/* UPD 2015.10.27 @eryksun in a comment below notes, that the semantically correct flag is CREATE_NEW_CONSOLE (0x00000010) */

On FreeBSD we have another problem: when the parent process is finished, it finishes the child processes as well. And that is not what you want in a CGI script either. Some experiments showed that the problem seemed to be in sharing sys.stdout. And the working solution was the following:

``````pid = subprocess.Popen([sys.executable, "longtask.py"], stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
``````

I have not checked the code on other platforms and do not know the reasons of the behaviour on FreeBSD. If anyone knows, please share your ideas. Googling on starting background processes in Python does not shed any light yet.

``````import os
``````

Note that this is dangerous, since the command isn"t cleaned. I leave it up to you to google for the relevant documentation on the "os" and "sys" modules. There are a bunch of functions (exec* and spawn*) that will do similar things.

## Shop

Best laptop for Sims 4

\$

Best laptop for Zoom

\$499

Best laptop for Minecraft

\$590

Best laptop for engineering student

\$

Best laptop for development

\$

Best laptop for Cricut Maker

\$

Best laptop for hacking

\$890

Best laptop for Machine Learning

\$950

Latest questions

NUMPYNUMPY

psycopg2: insert multiple rows with one query

NUMPYNUMPY

How to convert Nonetype to int or string?

NUMPYNUMPY

How to specify multiple return types using type-hints

NUMPYNUMPY

Javascript Error: IPython is not defined in JupyterLab

## Wiki

Python OpenCV | cv2.putText () method

numpy.arctan2 () in Python

Python | os.path.realpath () method

Python OpenCV | cv2.circle () method

Python OpenCV cv2.cvtColor () method

Python - Move item to the end of the list

time.perf_counter () function in Python

Check if one list is a subset of another in Python

Python os.path.join () method