I've implemented the following two versions of the classic "Max Sub-Array" problem in Clojure, using the Kadane algorithm.

First with loop / recur

(defn max-sub-array [A]
  (loop [x (first A)
         a (rest A)
         max-ending-here 0
         max-so-far 0]
    (if (seq a)
      (recur (first a) (rest a) (max x, (+ max-ending-here x)) (max max-so-far, max-ending-here))
      max-so-far)))

Then with reduce

(defn max-sub-array-reduction [A]
  (letfn [(find-max-sub-array [[max-ending-here max-so-far] x]
             [(max x (+ max-ending-here x)) (max max-so-far max-ending-here)])]
    (second (reduce find-max-sub-array [0 0] A))))

Is there a more compact implementation, perhaps using filter or merely by making the reduce version more "idiomatic" somehow?

I've implemented the following two versions of the classic "Max Sub-Array" problem in Clojure, using the Kadane algorithm.

First with loop / recur

(defn max-sub-array [A]
  (loop [x (first A)
         a (rest A)
         max-ending-here 0
         max-so-far 0]
    (if (seq a)
      (recur (first a) (rest a) (max x, (+ max-ending-here x)) (max max-so-far, max-ending-here))
      max-so-far)))

Then with reduce

(defn max-sub-array-reduction [A]
  (letfn [(find-max-sub-array [[max-ending-here max-so-far] x]
             [(max x (+ max-ending-here x)) (max max-so-far max-ending-here)])]
    (second (reduce find-max-sub-array [0 0] A))))

Is there a more concise implementation, perhaps using filter or merely by making the reduce version more "idiomatic" somehow?