In [1]:

    

from scipy.sparse import csr_matrix, csc_matrix, lil_matrix, dok_matrix, eye


    

In [2]:

    

n = 1000


    

In [3]:

    

csr = eye(n, format='csr')
csc = eye(n, format='csc')
lil = eye(n, format='lil')
dok = eye(n, format='dok')


    

In [4]:

    

%%timeit
csr.getrow(0)


    

    




48.3 µs ± 2.71 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [5]:

    

%%timeit
csc.getrow(0)


    

    




139 µs ± 11.6 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [6]:

    

%%timeit
lil.getrow(0)


    

    




18.1 µs ± 522 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [7]:

    

%%timeit
dok.getrow(0)


    

    




700 µs ± 60.7 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)

In [8]:

    

%%timeit
csr[0]


    

    




78.3 µs ± 5.64 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [9]:

    

%%timeit
csc[0]


    

    




74.5 µs ± 715 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [10]:

    

%%timeit
lil[0]


    

    




43.9 µs ± 4.06 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [11]:

    

%%timeit
dok[0]


    

    




462 µs ± 16.7 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)

In [12]:

    

%%timeit
csr.getcol(0)


    

    




67.5 µs ± 7.8 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [13]:

    

%%timeit
csc.getcol(0)


    

    




49.4 µs ± 2.21 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [14]:

    

%%timeit
lil.getcol(0)


    

    




955 µs ± 15.5 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)

In [15]:

    

%%timeit
dok.getcol(0)


    

    




12.3 ms ± 154 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

In [16]:

    

%%timeit
csr[:, 0]


    

    




86.7 µs ± 7.77 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [17]:

    

%%timeit
csc[:, 0]


    

    




77 µs ± 6.75 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [18]:

    

%%timeit
lil[:, 0]


    

    




442 µs ± 31.9 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)

In [19]:

    

%%timeit
dok[:, 0]


    

    




929 µs ± 36.2 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)

In [20]:

    

%%timeit
csr[0, 0]


    

    




31.8 µs ± 454 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [21]:

    

%%timeit
csc[0, 0]


    

    




104 µs ± 5.93 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [22]:

    

%%timeit
lil[0, 0]


    

    




3.37 µs ± 123 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

In [23]:

    

%%timeit
dok[0, 0]


    

    




18.1 µs ± 134 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

In [24]:

    

%%timeit
csr[:10]


    

    




71.4 µs ± 929 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [25]:

    

%%timeit
csc[:10]


    

    




184 µs ± 8.37 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [26]:

    

%%timeit
lil[:10]


    

    




47.2 µs ± 1.57 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [27]:

    

%%timeit
dok[:10]


    

    




632 µs ± 7.13 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)

In [28]:

    

%%timeit
csr[:, :10]


    

    




77.5 µs ± 1.32 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [29]:

    

%%timeit
csc[:, :10]


    

    




74.9 µs ± 5.32 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [30]:

    

%%timeit
lil[:, :10]


    

    




420 µs ± 14.9 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)

In [31]:

    

%%timeit
dok[:, :10]


    

    




905 µs ± 6.62 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)

In [32]:

    

%%timeit
csr[:10, :10]


    

    




73.9 µs ± 1.63 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [33]:

    

%%timeit
csc[:10, :10]


    

    




178 µs ± 1.79 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [34]:

    

%%timeit
lil[:10, :10]


    

    




47.5 µs ± 925 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)

In [35]:

    

%%timeit
dok[:10, :10]


    

    




210 µs ± 6.29 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)