I am interested in numerical analysis and have recently began a love affair with Common LISP. I have found my thought processeses strongly influenced by years of more imperative style programming, so I am on a quest to develop a more lispy approach. A common theme for those on the LISP path, I presume.
Anyway, reading through CLTL2 it is stated that vectors can be accessed in constant time, whereas accessing a list is done in linear time (sec. 2.5.1). One can then (setf (apply #'aref *some-vector* idx) new-value) to set one place of a vector. My two-part question is:
a. Is there a way to get several values of an array by giving a list indices, without having to loop through the list? E.g., something akin to a correct version of (map 'array #'aref *some-array* idx-list).
b. Is there a way to set the values of some elements of an array in the same spirit as above? It would be ideal to be able to apply a function (via an adequately defined map, or to reduce it, etc.) to a list of values extracted from arrays for, say, sampling and interpolation.
I would also appreciate any comments regarding array operations in general and numerical computing in particular (pointers to a tutorial or book, some mailing list, common pitfalls, when to use lists or other data structures instead, etc.).
Cheers.
a.
You can map to get to get the values by index like this:
(map 'vector #'(lambda (idx) (aref *some-array* idx)) idx-list)
Note that this works only for one-dimensional arrays. A more-than-one-dimensional version would look like this:
(map 'vector #'(lambda (idx) (apply #'aref *some-array* idx)) idx-list)