The goal is to learn how to memoize a function in IPython with no additional effort in a way that the cache is saved after we close and reopen the notebook.
Step 1
Assign a specific directory that will keep the cache for your functions.
from joblib import Memory
cachedir = "my-cache"
memory = Memory(cachedir, verbose = 0)
Copy
Python
The memory variable will me used as a decorator, and in principle, you can have several different caches for different functions if you need.
Step 2
A classical way to check the caching is to have a print statement inside.
The function will print its output only on the first run.
Starting from the second run, its value will be taken from a dictionary.
@memory.cache
def f(x):
print('Running f(%s)...' % x)
return x
Copy
Python
>>> first = f(1) # First run
Running f(1)...
>>> second = f(1) # Second run
>>> print(first)
1
>>> print(second)
1
Copy
Python
Step 3
You can clean after yourself if necessary.
## We can clear the cache of a specific function
>>> f.clear()
## And now, the values need to be recomputed again
>>> f(1)
Running f(1)...
1
Copy
Python
Step 4
When the function is changed during code development, the cache is cleared.
@memory.cache
def f(x):
print('The function has changed f(%s)' % x)
return x
>>> f(1)
The function has changed f(1)
1
Copy
Python
Step 5
Finally, you can clear the entire cache associated with variable memory
>>> memory.clear()
Copy
Python