It seems numpy's einsum function does not work with scipy.sparse matrices. Are there alternatives to do the sorts of things einsum can do with sparse matrices?
In response to #eickenberg's answer: The particular einsum I'm wanting to is numpy.einsum("ki,kj->ij",A,A) - the sum of the outer products of the rows.
A restriction of scipy.sparse matrices is that they represent linear operators and are thus kept two dimensional, which leads to the question: Which operation are you seeking to do?
All einsum operations on a pair of 2D matrices are very easy to write without einsum using dot, transpose and pointwise operations, provided that the result does not exceed two dimensions.
So if you need a specific operation on a number of sparse matrices, it is probable that you can write it without einsum.
UPDATE: A specific way to implement np.einsum("ki, kj -> ij", A, A) is A.T.dot(A). In order to convince yourself, please try the following example:
import numpy as np
rng = np.random.RandomState(42)
a = rng.randn(3, 3)
b = rng.randn(3, 3)
the_einsum_ab = np.einsum("ki, kj -> ij", a, b)
the_a_transpose_times_b = a.T.dot(b)
# We write a test in order to assert equality
from numpy.testing import assert_array_equal
assert_array_equal(the_einsum_ab, the_a_transpose_times_b) # This passes, so equality
This result is slightly more general. Now if you use b = a you obtain your specific result.
einsum translates the index string into a calculation using the C version of np.nditer. http://docs.scipy.org/doc/numpy/reference/arrays.nditer.html is a nice introduction to nditer. Note especially the Cython example at the end.
https://github.com/hpaulj/numpy-einsum/blob/master/einsum_py.py is a Python simulation of the einsum.
scipy.sparse has its own code (ultimately in C) to perform the basic operations, summation, matrix multiplication, etc. Sparse matricies have their own data structures. They can be lists, dictionaries, or a set of numpy arrays. Numpy notation can be used because sparse has the appropriate __xxx__ methods.
A sparse matrix is a matrix, a 2d array object. A sparse einsum could be written, but it would end up using the sparse matrix multiplication, not nditer. So at best it would be a notational convenience.
Sparse csr_matrix.dot is:
def dot(self, other):
"""Ordinary dot product
...
"""
return self * other
A=sparse.csr_matrix([[1,2],[3,4]])
A.dot(A.T).A
(A*A.T).A
A.__rmul__(A.T).A
A.__mul__(A.T).A
np.einsum('ij,kj',A.A,A.A)
# array([[ 5, 11],
# [11, 25]])
Related
I have a 2d array s and I want to calculate differences elementwise, i.e.:
Since it cannot be written as a single matrix multiplication, I was wondering what is the proper way to vectorize it?
You can use broadcasting for that: d = s[:, None, :] - s[None, :, :]. Note the None enable you to create a new dimension. Numpy implicitly perform the broadcasting operation between the two arrays.
In Python, I need to create an NxM matrix in which the ij entry has value of i^2 + j^2.
I'm currently constructing it using two for loops, but the array is quite big and the computation time is long and I need to perform it several times. Is there a more efficient way of constructing such matrix using maybe Numpy ?
You can use broadcasting in numpy. You may refer to the official documentation. For example,
import numpy as np
N = 3; M = 4 #whatever values you'd like
a = (np.arange(N)**2).reshape((-1,1)) #make it to column vector
b = np.arange(M)**2
print(a+b) #broadcasting applied
Instead of using np.arange(), you can use np.array([...some array...]) for customizing it.
I want to implement an operation on two matrices that is similar to matrix multiplication in that it each element of the resulting matrix is a function of the ith row of the first matrix and the jth column of the second matrix.
I would like to be able to do this using numpy and/or pandas using vectorized computations.
In other words:
How do I implemenent $A \bigotimes B = C$
where $C_{ij} = sum_k f(a_{ik}, b_{kj})$ in numpy and/or pandas?
So first off, I think what I'm trying to achieve is some sort of Cartesian product but elementwise, across the columns only.
What I'm trying to do is, if you have multiple 2D arrays of size [ (N,D1), (N,D2), (N,D3)...(N,Dn) ]
The result is thus to be a combinatorial product across axis=1 such that the final result will then be of shape (N, D) where D=D1*D2*D3*...Dn
e.g.
A = np.array([[1,2],
[3,4]])
B = np.array([[10,20,30],
[5,6,7]])
cartesian_product( [A,B], axis=1 )
>> np.array([[ 1*10, 1*20, 1*30, 2*10, 2*20, 2*30 ]
[ 3*5, 3*6, 3*7, 4*5, 4*6, 4*7 ]])
and extendable to cartesian_product([A,B,C,D...], axis=1)
e.g.
A = np.array([[1,2],
[3,4]])
B = np.array([[10,20],
[5,6]])
C = np.array([[50, 0],
[60, 8]])
cartesian_product( [A,B,C], axis=1 )
>> np.array([[ 1*10*50, 1*10*0, 1*20*50, 1*20*0, 2*10*50, 2*10*0, 2*20*50, 2*20*0]
[ 3*5*60, 3*5*8, 3*6*60, 3*6*8, 4*5*60, 4*5*8, 4*6*60, 4*6*8]])
I have a working solution that essentially creates an empty (N,D) matrix and then broadcasting a vector columnwise product for each column within nested for loops for each matrix in the provided list. Clearly is horrible once the arrays get larger!
Is there an existing solution within numpy or tensorflow for this? Potentially one that is efficiently paralleizable (A tensorflow solution would be wonderful but a numpy is ok and as long as the vector logic is clear then it shouldn't be hard to make a tf equivalent)
I'm not sure if I need to use einsum, tensordot, meshgrid or some combination thereof to achieve this. I have a solution but only for single-dimension vectors from https://stackoverflow.com/a/11146645/2123721 even though that solution says to work for arbitrary dimensions array (which appears to mean vectors). With that one i can do a .prod(axis=1), but again this is only valid for vectors.
thanks!
Here's one approach to do this iteratively in an accumulating manner making use of broadcasting after extending dimensions for each pair from the list of arrays for elmentwise multiplications -
L = [A,B,C] # list of arrays
n = L[0].shape[0]
out = (L[1][:,None]*L[0][:,:,None]).reshape(n,-1)
for i in L[2:]:
out = (i[:,None]*out[:,:,None]).reshape(n,-1)
I have a vector and wish to make another vector of the same length whose k-th component is
The question is: how can we vectorize this for speed? NumPy vectorize() is actually a for loop, so it doesn't count.
Veedrac pointed out that "There is no way to apply a pure Python function to every element of a NumPy array without calling it that many times". Since I'm using NumPy functions rather than "pure Python" ones, I suppose it's possible to vectorize, but I don't know how.
import numpy as np
from scipy.integrate import quad
ws = 2 * np.random.random(10) - 1
n = len(ws)
integrals = np.empty(n)
def f(x, w):
if w < 0: return np.abs(x * w)
else: return np.exp(x) * w
def temp(x): return np.array([f(x, w) for w in ws]).sum()
def integrand(x, w): return f(x, w) * np.log(temp(x))
## Python for loop
for k in range(n):
integrals[k] = quad(integrand, -1, 1, args = ws[k])[0]
## NumPy vectorize
integrals = np.vectorize(quad)(integrand, -1, 1, args = ws)[0]
On a side note, is a Cython for loop always faster than NumPy vectorization?
The function quad executes an adaptive algorithm, which means the computations it performs depend on the specific thing being integrated. This cannot be vectorized in principle.
In your case, a for loop of length 10 is a non-issue. If the program takes long, it's because integration takes long, not because you have a for loop.
When you absolutely need to vectorize integration (not in the example above), use a non-adaptive method, with the understanding that precision may suffer. These can be directly applied to a 2D NumPy array obtained by evaluating all of your functions on some regularly spaced 1D array (a linspace). You'll have to choose the linspace yourself since the methods aren't adaptive.
numpy.trapz is the simplest and least precise
scipy.integrate.simps is equally easy to use and more precise (Simpson's rule requires an odd number of samples, but the method works around having an even number, too).
scipy.integrate.romb is in principle of higher accuracy than Simpson (for smooth data) but it requires the number of samples to be 2**n+1 for some integer n.
#zaq's answer focusing on quad is spot on. So I'll look at some other aspects of the problem.
In recent https://stackoverflow.com/a/41205930/901925 I argue that vectorize is of most value when you need to apply the full broadcasting mechanism to a function that only takes scalar values. Your quad qualifies as taking scalar inputs. But you are only iterating on one array, ws. The x that is passed on to your functions is generated by quad itself. quad and integrand are still Python functions, even if they use numpy operations.
cython improves low level iteration, stuff that it can convert to C code. Your primary iteration is at a high level, calling an imported function, quad. Cython can't touch or rewrite that.
You might be able to speed up integrand (and on down) with cython, but first focus on getting the most speed from that with regular numpy code.
def f(x, w):
if w < 0: return np.abs(x * w)
else: return np.exp(x) * w
With if w<0 w must be scalar. Can it be written so it works with an array w? If so, then
np.array([f(x, w) for w in ws]).sum()
could be rewritten as
fn(x, ws).sum()
Alternatively, since both x and w are scalar, you might get a bit of speed improvement by using math.exp etc instead of np.exp. Same for log and abs.
I'd try to write f(x,w) so it takes arrays for both x and w, returning a 2d result. If so, then temp and integrand would also work with arrays. Since quad feeds a scalar x, that may not help here, but with other integrators it could make a big difference.
If f(x,w) can be evaluated on a regular nx10 grid of x=np.linspace(-1,1,n) and ws, then an integral (of sorts) just requires a couple of summations over that space.
You can use quadpy for fully vectorized computation. You'll have to adapt your function to allow for vector inputs first, but that is done rather easily:
import numpy as np
import quadpy
np.random.seed(0)
ws = 2 * np.random.random(10) - 1
def f(x):
out = np.empty((len(ws), *x.shape))
out0 = np.abs(np.multiply.outer(ws, x))
out1 = np.multiply.outer(ws, np.exp(x))
out[ws < 0] = out0[ws < 0]
out[ws >= 0] = out1[ws >= 0]
return out
def integrand(x):
return f(x) * np.log(np.sum(f(x), axis=0))
val, err = quadpy.quad(integrand, -1, +1, epsabs=1.0e-10)
print(val)
[0.3266534 1.44001826 0.68767868 0.30035222 0.18011948 0.97630376
0.14724906 2.62169217 3.10276876 0.27499376]