2005-08-17

Python lacks sparse arrays

Last days I'm writing many python script to postprocess some calculation results and format them into a form suitable for importing into Mathematica.

In Perl I could write $array[$x]=0 and not worry if index $x actually exists in the array. If not, then the "hole" between the last valid index and $x is automatically filled with undef.

Not so in Python. There you get an exception if you try to access non-existant index in the list. It will not grow the list for you automatically. Of course, you can write the code to test whether the index you're trying to access is larger than the current list size and manually extend the list. This kind of coding defeats the purpose of having a high-level language. It is the same way I would code it in C using realloc().

Another possibility is to use a dictionary with integer indices as keys. However, access time to individual elements might not be constant and keys are traversed in an undefined order when iterating over the dictionary. Unlike Perl, you do not get None value for non-existing keys because the set of all possible keys is open and not restricted to numeric values up to the last key with value not equal to None as it would be with an automatically-growing list.

I have found several workarounds, but the mere fact of having to work around the problem annoys me. Not enough to go back to Perl, but it is annoying having to repeat the same hack over and over again.

3 comments:

Anonymouz said...
This comment has been removed by the author.
Anonymouz said...

Take 2:

Blogger seems to munge the code pretty badly, so here is the same thing on dpaste:

http://dpaste.com/hold/36691/

I think this is what you are after.

Ron Burk said...

http://dpaste.com/hold/36691/

is now a defunct web page. Boy, technology is cool! :-)