I suppose you might be using f = lambdify(x, sin(x), 'scipy'). This
produces a function that calls scipy.sin instead of numpy.sin. However,
scipy.sin is just a deprecated wrapper for np.sin, which would explain why
it is a little slower.
I'm a little unclear why it doesn't raise a deprecation
What is the difference between "numpy" and "scipy" in your graph? I
wouldn't expect scipy to make any difference for sin or cos which are NumPy
functions.
Aaron Meurer
On Wed, Sep 15, 2021 at 7:19 AM Davide Sandona'
wrote:
> Thanks Oscar for the wonderful clarification!
>
> I rerun my code
Thanks Oscar for the wonderful clarification!
I rerun my code with SYMPY_USE_CACHE=no, now the results make much more
sense.
[image: lambdify-no-cache.png]
Davide.
Il giorno mer 15 set 2021 alle ore 15:14 Oscar Benjamin <
oscar.j.benja...@gmail.com> ha scritto:
>
> On Wed, 15 Sept 2021 at
On Wed, 15 Sept 2021 at 13:41, Oscar Benjamin
wrote:
> On Tue, 14 Sept 2021 at 23:12, sandona...@gmail.com <
> sandona.dav...@gmail.com> wrote:
>
>> Hello,
>> let's say I'd like to numerically evaluate a single sympy function over
>> an array using sympy as the module. Curiously, passing in
On Tue, 14 Sept 2021 at 23:12, sandona...@gmail.com <
sandona.dav...@gmail.com> wrote:
> Hello,
> let's say I'd like to numerically evaluate a single sympy function over an
> array using sympy as the module. Curiously, passing in regular Python's
> float numbers makes the evaluation much faster
On Wed, Sep 15, 2021 at 2:36 AM Davide Sandona'
wrote:
> What are the results you get from these? For me the last one is
>> slightly faster, which makes sense because the inputs don't have to be
>> converted to SymPy Floats first.
>>
>
> This chart summarizes my findings. Each dot represents the
>
> What are the results you get from these? For me the last one is
> slightly faster, which makes sense because the inputs don't have to be
> converted to SymPy Floats first.
>
This chart summarizes my findings. Each dot represents the evaluation time
with the specified module + data type.
I
On Tue, Sep 14, 2021 at 4:13 PM sandona...@gmail.com
wrote:
>
> Hello,
> let's say I'd like to numerically evaluate a single sympy function over an
> array using sympy as the module. Curiously, passing in regular Python's float
> numbers makes the evaluation much faster then passing in Sympy's
Hello,
let's say I'd like to numerically evaluate a single sympy function over an
array using sympy as the module. Curiously, passing in regular Python's
float numbers makes the evaluation much faster then passing in Sympy's
Float instances. I tried several sympy functions, they tend to follow