I'm converting some C++ code to Clojure, and I want
to return a graph g
with a bunch of edges added to it.
I pass in the the number of vertices, the graph, and
the test predicate (eg, a function that could depend on i, j, randomness, ...) something like this:
(defn addSomeEdges [v g test-p]
(doseq [i (range v)]
(doseq [j (range (dec i))]
(if test-p
(add-edges g [i j] )
)))
g)
the problem, of course, is that (add-edges)
returns a new g
. How can I capture this updated graph using best practices Clojure, please? It seems so simple and natural in C++.
Iterativly accumulating information looks like a reducing function if you split it into two parts:
Which can be written using reduce
user> (defn add-edge [g i j]
(assoc g i j))
#'user/add-edge
user> (add-edge {1 2} 2 1)
{1 2, 2 1}
user> (defn addSomeEdges [v g test-p]
(reduce (fn [graph [i j]] ;; this takes the current graph, the points,
(if (test-p graph i j) ;; decides if the edge should be created.
(add-edge graph i j) ;; and returns the next graph
graph)) ;; or returns the graph unchanged.
g ;; This is the initial graph
(for [i (range v)
j (range (dec i))]
[i j]))) ;; this generates the candidate edges to check.
#'user/addSomeEdges
and let's run it!
user> (addSomeEdges 4 {1 2} (fn [g i j] (rand-nth [true false])))
{1 2, 2 0}
user> (addSomeEdges 4 {1 2} (fn [g i j] (rand-nth [true false])))
{1 2, 3 0}
user> (addSomeEdges 4 {1 2} (fn [g i j] (rand-nth [true false])))
{1 2, 2 0, 3 1}
When you think of other tests you can thread these calls together:
user> (as-> {1 2} g
(addSomeEdges 4 g (fn [g i j] (rand-nth [true false])))
(addSomeEdges 7 g (fn [g i j] (< i j)))
(addSomeEdges 9 g (fn [g i j] (contains? (set (keys g)) j))))
{1 2, 3 1, 4 1, 5 3, 6 4, 7 5, 8 6}