I want to make an experiment where I create a list of many lists of randomly generated sequences that all contain every digit 0 to 9 inclusive, that is, the generation function is to generate random numbers and place them in a list of integers while there is at least 1 digit not found in the list.
The intention for the experiment is to try to make some generalizations about things like expected number # of digits in such a function, how long can a sequence get(can my program loop indefinitely and never find that last digit?), and other interesting things(for me).
I am using PERL for the experiment.
The idea seemed simple at first, I sat down, created a list, and figured I can just make a loop that runs an arbitrary amount of times (I decided to choose 100 times), which calls a function generate_sequence(input: none, output: list of numbers that contains at least 1 of every digit) and adds it to the list.
I quickly realized that I struggle cleanly specifying what it means, pragmatically, to generate a list of numbers that contains one of every digit.
My original attempt was to make a list of digits(0..9), and as I generate numbers, I would search the list for that digit if it is in the list, and remove it. This way, it would generate numbers until the list of digits "still needed" is empty. This approach seems unappealing and can involve a lot of redundant tasks such as checking whether the digit generated is in the list of digits needed every single time a number is generated...
Is there a more elegant solution to such a problem? I am really unhappy with the way I am approaching the function.
In general, I need a function F that accepts nothing, and returns a list of randomly generated numbers that contains every digit 1..9, that is, it stops as soon as every digit from 1 to 9 inclusive is generated.
Thanks ahead of time.
Well, the problem is if you 'roll randomly' you don't actually know how many iterations you're going to need - in theory it could be infinite.
If you're doing it in perl you're probably much better off using the List::Util module and shuffle - feed it a list of elements you want to shuffle.
E.g.
#!/usr/bin/env perl
use strict;
use warnings;
use List::Util qw( shuffle );
my #shuffled = shuffle ( 0..9 );
print #shuffled;
You could reproduce this quite easily, but why bother when List::Util is core as of 5.7.3
However it does sound like you're trying to generate a list, that might contain repeats, until you hit a terminate condition.
I'm not entirely sure why, but that would be best done using a hash, and counting occurences. (And terminate when your 'keys' is complete).
E.g.:
#!/usr/bin/env perl
use strict;
use warnings;
my %seen;
my #list_of_numbers;
while ( keys %seen < 10 ) {
my $gen = int rand ( 10 );
$seen{$gen}++;
push ( #list_of_numbers, $gen );
}
print #list_of_numbers;
Note - there's actually an extremely small chance of this rolling extremely long sequences, because of the nature of 'random' - it means in theory you might have a very long 'streak' of not rolling a 6.
For bonus points in %seen you have a frequency spread of your generated numbers.
A python implementation:
from random import randint
s = set(range(10))
def f():
result = []
t = set()
while 1:
n = randint(0, 9)
result.append(n)
t.add(n)
if t == s:
return result
For example:
for i in range(10):
print(len(f()))
....:
20
34
69
22
23
25
20
29
30
32
This should work (python):
import random
nums = []
while any([ i not in set(nums) for i in range(1,11)]):
nums.append(random.randrange(1, 11, 1))
or more specific to what you are trying to do:
import random
lengths = []
for i in range(1000):
nums = []
while any([ i not in set(nums) for i in set(range(1,11))]):
nums.append(random.randrange(1, 11, 1))
lengths.append(len(nums))
This approach counts the iterations needed to fill a dictionary of digits:
import random
c = 0
d = dict()
while len(d.keys()) <10:
d[random.randint(0,9)] = 1
c += 1
print c
Wrote this before you switched to just Perl...
from random import randrange
def F():
todo = set(range(10))
nums = []
while todo:
r = randrange(10)
nums.append(r)
todo.discard(r)
return nums
>>> F()
[8, 2, 2, 3, 1, 0, 3, 9, 3, 4, 7, 4, 7, 5, 0, 9, 5, 5, 6]
Another:
def F():
done = 0
nums = []
while done < 1023:
r = randrange(10)
nums.append(r)
done |= 1 << r
return nums
In Clojure, I am keeping track of both the random list and the existing values, thus avoiding a search on the growing list.
(defn random-list [ up-to ]
(loop [ n [] tries [] ]
(if (> (count n) (dec up-to))
tries
(let [i (rand-int up-to) n-tries (conj tries i)]
(if (some #{i} n )
(recur n n-tries)
(recur (conj n i) n-tries))))))
We can define similar functions:
(defn random-list-to-10 []
(random-list 10))
(random-list-to-10)
; [3 6 9 0 8 0 5 7 3 8 1 8 4 3 4 2]
We can also take only a few random elements:
(take 5 (random-list 10))
; (6 1 0 9 5)
Here is a possible Perl implementation that counts the iterations needed to fill a hash with the 10 digts:
#!/usr/bin/perl
my $count = 0;
my %dict = ();
while (scalar keys %dict < 10) {
$dict{int(rand(10))} = 1;
$count ++;
}
print $count;
(see online demo)
In clojure (though probably not the most elegant):
(loop [n []
s (set (range 0 10))]
(if (= s (set n))
n
(recur (conj n (rand-int 10)) s)))
Sample output:
user=> (loop [n [] s (set (range 0 10))] (if (= s (set n)) n (recur (conj n (rand-int 10)) s)))
[0 6 2 8 5 2 0 0 9 3 0 3 0 1 7 5 0 4]
user=> (loop [n [] s (set (range 0 10))] (if (= s (set n)) n (recur (conj n (rand-int 10)) s)))
[2 1 7 7 3 2 8 8 4 7 5 0 1 3 0 3 0 4 0 0 3 7 3 4 5 8 1 3 8 5 3 5 5 9 4 0 2 1 2 7 8 3 9 7 8 6]
user=> (loop [n [] s (set (range 0 10))] (if (= s (set n)) n (recur (conj n (rand-int 10)) s)))
[7 1 8 3 1 1 0 6 8 4 9 7 0 0 2 7 4 0 1 1 8 8 4 3 9 8 4 2 8 3 2 8 4 6 0 9 9 7 2 3 0 3 0 4 2 4 0 5]
user=> (loop [n [] s (set (range 0 10))] (if (= s (set n)) n (recur (conj n (rand-int 10)) s)))
[9 1 9 0 9 5 3 0 3 8 4 0 1 6 3 0 1 8 0 3 8 3 5 4 3 9 8 8 8 8 2 2 8 9 9 3 9 2 5 1 1 3 4 6 3 1 4 0 2 6 7]
user=> (loop [n [] s (set (range 0 10))] (if (= s (set n)) n (recur (conj n (rand-int 10)) s)))
[4 1 5 5 5 5 2 2 5 5 3 1 5 3 5 1 4 2 4 2 3 1 4 7 1 9 3 8 0 8 4 0 9 3 4 9 9 1 8 8 0 6]
user=> (loop [n [] s (set (range 0 10))] (if (= s (set n)) n (recur (conj n (rand-int 10)) s)))
[0 4 0 9 1 8 4 8 6 6 6 9 8 4 9 0 9 3 3 7 6 1 4 3 8 1 1 4 9 5 1 4 1 2]
user=>
Related
I'm doing an algorithm that randomizes a TSP (array of citys) based on 1 TSP.
(do ((i 0 (+ i 1)))
((= i n-population))
(setf (aref population i) (shuffle TSP 100))
)
And as far as I know im filling up i positions of the array population with (shuffle TSP 100) that is beeing called each iteration, but the algorithm is setting all array positions and not just i position.
[Note. An earlier version of this answer contained a mistake which would badly alter the statistics of the shuffling: please check below for the corrected version and a note as to what the problem was.]
Given your code, slightly elaborated to turn it into a function:
(defun fill-array-with-something (population n-population TSP)
(do ((i 0 (+ i 1)))
((= i n-population))
(setf (aref population i) (shuffle TSP 100))))
Then each element of population from 0 to (1- n-population) will be set to the result of (shuffle TSP 100). There are then two possibilities:
(shuffle TSP 100) returns a fresh object from each call;
(shuffle TSP 100) returns the same object – probably TSP – from each call.
In the first case, each element of the array will have a distinct value. In the second case, all elements below n-population will have the same value.
Without knowing what your shuffle function does, here is an example of one which will give the latter behaviour:
(defun shuffle (vec n)
;; shuffle pairs of elts of VEC, N times.
(loop with max = (length vec)
repeat n
do (rotatef (aref vec (random max))
(aref vec (random max)))
finally (return vec)))
And we can test this:
> (let ((pop (make-array 10))
(tsp (vector 0 1 2 3 4 5 6 7 8 9 )))
(fill-array-with-something pop (length pop) tsp)
pop)
#(#(2 8 7 1 3 9 5 4 0 6) #(2 8 7 1 3 9 5 4 0 6) #(2 8 7 1 3 9 5 4 0 6)
#(2 8 7 1 3 9 5 4 0 6) #(2 8 7 1 3 9 5 4 0 6) #(2 8 7 1 3 9 5 4 0 6)
#(2 8 7 1 3 9 5 4 0 6) #(2 8 7 1 3 9 5 4 0 6) #(2 8 7 1 3 9 5 4 0 6)
#(2 8 7 1 3 9 5 4 0 6))
As you can see all the elements are mysteriously the same thing, which is because my shuffle simply returned its first argument, having modified it in place.
You can check this by either explicitly checking the result of shuffle, or by, for instance, using *print-circle* to see the sharing. The latter approach is pretty neat:
> (let ((*print-circle* t)
(pop (make-array 10))
(tsp (vector 0 1 2 3 4 5 6 7 8 9 )))
(fill-array-with-something pop (length pop) tsp)
(print pop)
(values))
#(#1=#(4 6 7 0 1 2 5 9 3 8) #1# #1# #1# #1# #1# #1# #1# #1# #1#)
And now it's immediately apparent what the problem is.
The solution is to make sure either that shuffle returns a fresh object, or to copy its result. With my shuffle this can be done like this:
(defun fill-array-with-something (population n-population tsp)
(do ((i 0 (+ i 1)))
((= i n-population))
(setf (aref population i) (shuffle (copy-seq TSP) 100))))
Note that a previous version of this answer had (copy-seq (shuffle TSP 100)): with my version of shuffle this is a serious mistake, as it means that the elements in population are related to each other but get increasingly shuffled as you go along. With (shuffle (copy-seq TSP) 100) each element gets the same amount of shuffling, independently.
And now
> (let ((*print-circle* t)
(pop (make-array 10))
(tsp (vector 0 1 2 3 4 5 6 7 8 9 )))
(fill-array-with-something pop (length pop) tsp)
(print pop)
(values))
#(#(8 3 4 1 6 9 2 5 0 7) #(8 6 5 1 3 0 4 2 9 7) #(5 0 4 7 1 6 9 3 2 8)
#(3 0 7 6 2 9 4 5 1 8) #(8 2 5 1 7 3 9 0 4 6) #(0 5 6 3 8 7 2 1 4 9)
#(4 1 3 7 8 0 5 2 9 6) #(6 9 1 5 0 7 4 2 3 8) #(2 7 5 8 0 9 6 3 4 1)
#(5 4 8 9 6 7 2 0 1 3))
I suspect that the problem is in OP function SHUFFLE which has not yet been shared; my suspicion is that SHUFFLE is shuffling the *TSP* array itself in place instead of creating a shuffled copy of that array. The POPULATION values are then all referencing the same shuffled *TSP* array.
To solve this problem, SHUFFLE should return a shuffled array instead of shuffling the array in place. Here is a function that performs a Fisher-Yates shuffle on a vector:
(defun shuffle-vector (vect)
"Takes a vector argument VECT and returns a shuffled vector."
(let ((result (make-array (length vect) :fill-pointer 0)))
(labels ((shuffle (v)
(if (zerop (length v))
result
(let* ((i (random (length v)))
(x (elt v i)))
(vector-push x result)
(shuffle (concatenate 'vector
(subseq v 0 i)
(subseq v (1+ i))))))))
(shuffle vect))))
Testing in the REPL:
CL-USER> (defvar *TSP* #("Village" "Town" "City" "Metropolis" "Megalopolis"))
*TSP*
CL-USER> (defvar *n-population* 5)
*N-POPULATION*
CL-USER> (defvar *population* (make-array *n-population*))
*POPULATION*
CL-USER> (dotimes (i *n-population*)
(setf (aref *population* i) (shuffle-vector *TSP*)))
NIL
CL-USER> *population*
#(#("Megalopolis" "City" "Metropolis" "Town" "Village")
#("Megalopolis" "Metropolis" "Town" "City" "Village")
#("City" "Megalopolis" "Town" "Village" "Metropolis")
#("City" "Megalopolis" "Village" "Metropolis" "Town")
#("Megalopolis" "Town" "Metropolis" "City" "Village"))
I'm struggling to understand this q code programming idiom from the kx cookbook:
q)swin:{[f;w;s] f each { 1_x,y }\[w#0;s]}
q)swin[avg; 3; til 10]
0 0.33333333 1 2 3 4 5 6 7 8
The notation is confusing. Is there an easy way to break it down as a beginner?
I get that the compact notation for the function is probably equivalent to this
swin:{[f;w;s] f each {[x; y] 1_x, y }\[w#0;s]}
w#0 means repeat 0 w times (w is some filler for the first couple of observations?), and 1_x, y means join x, after dropping the first observation, to y. But I don't understand how this then plays out with f = avg applied with each. Is there a way to understand this easily?
http://code.kx.com/q/ref/adverbs/#converge-iterate
Scan (\) on a binary (two-param) function takes the first argument as the seed value - in this case 3#0 - and iterates through each of the items in the second list - in this case til 10 - applying the function (append new value, drop first).
q){1_x,y}\[3#0;til 10]
0 0 0
0 0 1
0 1 2
1 2 3
2 3 4
3 4 5
4 5 6
5 6 7
6 7 8
7 8 9
So now you have ten lists and you can apply a function to each list - in this case avg but it could be any other function that applies to a list
q)med each {1_x,y}\[3#0;til 10]
0 0 1 2 3 4 5 6 7 8f
q)
q)first each {1_x,y}\[3#0;til 10]
0 0 0 1 2 3 4 5 6 7
q)
q)last each {1_x,y}\[3#0;til 10]
0 1 2 3 4 5 6 7 8 9
I'd like to insert the result of an evaluated Clojure expression directly in my Emacs buffer, in pretty-printed form.
For example, with something like:
;; [emacs lisp]
(insert (nrepl-dict-get (nrepl-sync-request:eval "(range 30)") "value"))
I get, in the buffer of interest,
;;=>
(0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29)
In the past, I've let Clojure pretty-print things for me, as so:
(nrepl-dict-get
(nrepl-sync-request:eval
(format "(clojure.core/let [x %s] (with-out-str (clojure.pprint/pprint x)))"
"(range 30)"))
"value")
;;=>
"(0\n 1\n 2\n 3\n 4\n 5\n 6\n 7\n 8\n 9\n 10\n 11\n 12\n 13\n 14\n 15\n 16\n 17\n 18\n 19\n 20\n 21\n 22\n 23\n 24\n 25\n 26\n 27\n 28\n 29)\n"
However, the " and \n are being inserted escaped; I want them to be inserted unescaped. In other words, I want the pretty-printed result to be inserted directly without escaping quotes or newlines. This used to work in earlier versions of Cider and cider-nrepl.
Wrapping:
(nrepl-dict-get
(nrepl-sync-request:eval
(format "(clojure.core/let [x %s] (with-out-str (clojure.pprint/pprint x)))"
"(range 30)"))
"value")
in read should solve this.
I've just added this feature to lispy (it's a Paredit-style
package that uses Cider for Clojure eval):
2E will to a pretty-printed eval-and-insert, while
E will keep doing a plain one.
Here's an example (| represents point):
|(for [x (range 8)] (range x))
After E:
|(for [x (range 8)] (range x))
(() (0) (0 1) (0 1 2) (0 1 2 3) (0 1 2 3 4) (0 1 2 3 4 5) (0 1 2 3 4 5 6))
After 2E:
|(for [x (range 8)] (range x))
(()
(0)
(0 1)
(0 1 2)
(0 1 2 3)
(0 1 2 3 4)
(0 1 2 3 4 5)
(0 1 2 3 4 5 6))
Of course you can still do EjM to accomplish the same thing:
(for [x (range 8)] (range x))
|(()
(0)
(0 1)
(0 1 2)
(0 1 2 3)
(0 1 2 3 4)
(0 1 2 3 4 5)
(0 1 2 3 4 5 6))
I have:
(def data [[1 3 4 7 9] [7 6 3 2 7] [1 9 8 6 2]])
I want to average these (element-wise to get):
[3 6 5 5 6]
Like you would in MATLAB:
mean([1 3 4 7 9; 7 6 3 2 7; 1 9 8 6 2])
With Incanter I can do:
(map #(/ % (count m)) (apply plus data))
If data is rather large (and I have lots of them) is there a better way to do this?
Does it help to calculate the (count m) beforehand?
Does it help to defn the #(/ % (count m)) beforehand?
Here's a pretty clean and simple way to do it:
(def data [[1 3 4 7 9] [7 6 3 2 7] [1 9 8 6 2]])
(defn average [coll]
(/ (reduce + coll) (count coll)))
(defn transpose [coll]
(apply map vector coll))
(map average (transpose data))
=> (3 6 5 5 6)
As of 2013, my recommendation would be to just use core.matrix.stats to import all of this functionality:
(mean [[1 3 4 7 9] [7 6 3 2 7] [1 9 8 6 2]])
=> [3.0 6.0 5.0 5.0 6.0]
core.matrix.stats builds on the core.matrix API, so it will also work on other more optimised implementations of vectors and matrices - this is likely to be a better option if you are doing a lot of heavy matrix processing.
Without knowing how to use any of incanter, here's how you could do this "from scratch".
(let [data [[1 3 4 7 9] [7 6 3 2 7] [1 9 8 6 2]]
num (count data)]
(apply map (fn [& items]
(/ (apply + items) num))
data))
;=> (3 6 5 5 6)
If I had a N lists each of length M, how could I write a nice clean function to return a single list of length M, where each element is the sum of the corresponding elements in the N lists?
(starting to learn lisp - go easy!)
This is a job for the map and apply functions. Here is a way to do it, with an EDIT suggested by Nathan Sanders:
(define (add-lists . more)
(apply map + more))
For a more matlab like syntax:
(define (piecewise func)
(lambda more
(apply map func more)))
(define pw piecewise)
((pw +) '(1 2 3 4 5) '(6 7 8 9 0))
((pw -) '(1 2 3 4 5) '(6 7 8 9 0))
((pw *) '(1 2 3 4 5) '(6 7 8 9 0))
((pw /) '(1 2 3 4 5) '(6 7 8 9 0.1))
outputs:
(7 9 11 13 5)
(-5 -5 -5 -5 5)
(6 14 24 36 0)
(1/6 2/7 3/8 4/9 50.0)
Just this works in MIT scheme.
(map + '(1 2 3) '(4 5 6) '(7 8 9))
;Value 28: (12 15 18)