From 45b3956421bf25bd2ad79e3b27beaaec65c82cda Mon Sep 17 00:00:00 2001
From: Simon Cruanes <simon.cruanes.2007@m4x.org>
Date: Mon, 5 Apr 2021 00:13:33 -0400
Subject: [PATCH] improve test hash

---
 src/core/tests/dune         |  2 +-
 src/core/tests/test_hash.ml | 71 ++++++++++---------------------------
 2 files changed, 19 insertions(+), 54 deletions(-)

diff --git a/src/core/tests/dune b/src/core/tests/dune
index 4d56aece..ef6300eb 100644
--- a/src/core/tests/dune
+++ b/src/core/tests/dune
@@ -10,7 +10,7 @@
   (name test_hash)
   (modules test_hash)
   (flags :standard -warn-error -a+8)
-  (libraries containers))
+  (libraries containers iter))
 
 (executable
   (name test_random)
diff --git a/src/core/tests/test_hash.ml b/src/core/tests/test_hash.ml
index 4ebb50ea..d261309f 100644
--- a/src/core/tests/test_hash.ml
+++ b/src/core/tests/test_hash.ml
@@ -13,6 +13,9 @@ module Hist = struct
   let add self x =
     Hashtbl.replace self.tbl x (1 + try Hashtbl.find self.tbl x with _ ->0);
     self.n_samples <- 1 + self.n_samples
+  let add_n self x n =
+    Hashtbl.replace self.tbl x (n + try Hashtbl.find self.tbl x with _ ->0);
+    self.n_samples <- n + self.n_samples
 
   let pp out (self:t) : unit =
     let max = Hashtbl.fold (fun k _ n -> max k n) self.tbl 0 in
@@ -24,69 +27,31 @@ module Hist = struct
     done
 end
 
-(* how long does it take to get back to [n0]?
-   @param max if provided, stop when reaching it *)
-let orbit_size ?(max=max_int) n0 : int =
-  let size = ref 1 in
-  let n = ref (H.int n0) in
-  while !n <> n0 && !size < max do
-    n := H.int !n;
-    incr size;
-  done;
-  !size
-
 let reset_line = "\x1b[2K\r"
 
 let t_int n1 n2 =
   Printf.printf "test hash_int on %d--%d\n" n1 n2;
-  let hist = Hist.create() in
+  let count = Hashtbl.create 128 in
   for i=n1 to n2 do
-    Printf.printf "%sorbit for %d…%!" reset_line i;
-    let orb = orbit_size ~max:30_000 i in
-    Hist.add hist orb;
+    Printf.printf "%shash %d…%!" reset_line i;
+    let h = H.int i in
+    Hashtbl.replace count h (1 + CCHashtbl.get_or count h ~default:0);
+    if i mod 1024*1024*1024 = 0 then Gc.major();
   done;
   Printf.printf "%s%!" reset_line;
+  (* reverse table *)
+  let by_count =
+    CCHashtbl.to_iter count
+    |> Iter.map (fun (_h,n) -> n)
+    |> Iter.count ~hash:H.int
+  in
+  let hist = Hist.create() in
+  by_count (fun (n,i) -> Hist.add_n hist n i);
   Format.printf "histogram:@.%a@." Hist.pp hist;
   (*assert (Hist.check_uniform hist);*)
   ()
 
-(* how long does it take to get back to [n0] by
-   hashing [n] with [m0] ?
-   @param max if provided, stop when reaching it *)
-let left_orbit_size ?(max=max_int) n0 m0 : int =
-  let size = ref 1 in
-  let n = ref (H.combine2 (H.int n0) (H.int m0)) in
-  while !n <> n0 && !size < max do
-    n := H.combine2 (H.int !n) (H.int m0);
-    incr size;
-  done;
-  !size
-
-let t_int2 n1 n2 n3 n4 =
-  Printf.printf "test hash combine2 on %d--%d x %d--%d\n" n1 n2 n3 n4;
-  let hist = Hist.create() in
-  for i=n1 to n2 do
-    for j=n3 to n4 do
-      Printf.printf "%sleft-orbit for %d x %d…%!" reset_line i j;
-      let orb = left_orbit_size ~max:20_000 i j in
-      Hist.add hist orb;
-    done;
-  done;
-  Printf.printf "%s%!" reset_line;
-  Format.printf "histogram:@.%a@." Hist.pp hist;
-  ()
-
-let help =
-{|Test the hash function on integers, by computing orbits for inputs.
-An orbit is a cycle [n -> h(n) -> h(h(n)) -> h^3(n) -> … h^k(n)]
-where [h^k(n) == n]. For a good hash functions, orbits should be as long as
-possible.
-|}
-
 let () =
-  print_endline help;
-  t_int 0 80_000;
-  t_int 3_000_000 3_040_000;
-  t_int (-40_000) (-17_000);
-  t_int2 10 300 600 800;
+  t_int 0 2_000_000;
+  t_int (-4_000_000) (-3_500_000);
   ()