Question

A Brute Force approach is not intended to solve to question but aid in its research. I am working on a Project Euler problem that has me finding all the numbers from X to one less than Y that have have exactly one "substring" divisible by the number of digits in a number.

These are called one-child numbers. 104 is a one-child number. Of its substrings, [1, 0, 4, 10, 04, 104] only 0 is divisible by 3. The question asks to find the amount of one-child numbers that occur less then 10*17. A brute force method is not the correct approach; however, I have a theory that requires me to know the amount of one child numbers occuring before 10*11.

I haven't been successful in finding this number even after leaving my laptop on for half a day. I tried Cython, put i am a novice programmer who knows nothing about C. The result was really bad. I even tried cloud computing, but my ssh pipe always breaks before the process is complete.

If someone could help me pinpoint some different approaches to or optimization for preforming a BRUTE FORCE method for this problem up to 10**11, it would be greatly appreciated.

PLEASE DO NOT...

lend me advice on number theory or your answers to this problem, as I have been working on it for a good deal of time, and I really wish to come to the conclusion on my own.

## a one child number has only one "substring" divisable by the
## number of digits in the number. Example: 104 is a one child number as 0
## is the only substring which 3 may divide, of the set [1,0,4,10,04,104]

## FYI one-child numbers are positive, so the number 0 is not one-child


from multiprocessing import Pool
import os.path

def OneChild(numRange): # hopefully(10*11,1)
    OneChild = []
    start = numRange[0]
    number = numRange[1]

    ## top loop handles one number at a time
    ## loop ends when start become larger then end
    while number >= start:

        ## preparing to analayze one number
        ## for exactly one divisableSubstrings
        numberString = str(start)
        numDigits = len(numberString)
        divisableSubstrings = 0
        ticker1,ticker2 = 0, numDigits

        ## ticker1 starts at 0 and ends at number of digits - 1
        ## ticker2 starts at number of digits and ends +1 from ticker1
        ## an example for a three digit number: (0,3) (0,2) (0,1) (1,3) (1,2) (2,3)
        while ticker1 <= numDigits+1:
            while ticker2 > ticker1:
                if int(numberString[ticker1:ticker2]) % numDigits == 0:
                    divisableSubstrings += 1
                    if divisableSubstrings == 2:
                        ticker1 = numDigits+1
                        ticker2 = ticker1

                ##Counters    
                ticker2 -= 1
            ticker1 += 1
            ticker2 = numDigits             
        if divisableSubstrings == 1: ## One-Child Bouncer 
            OneChild.append(start) ## inefficient but I want the specifics
        start += 1 
    return (OneChild)

## Speed seems improve with more pool arguments, labeled here as cores
## Im guessing this is due to pypy preforming best when task is neither
## to large nor small
def MultiProcList(numRange,start = 1,cores = 100): # multiprocessing
    print "Asked to use %i cores between %i numbers: From %s to %s" % (cores,numRange-start, start,numRange)
    cores = adjustCores(numRange,start,cores)
    print "Using %i cores" % (cores)

    chunk = (numRange+1-start)/cores
    end = chunk+start -1 
    total, argsList= 0, []
    for i in range(cores):
        # print start,end-1
        argsList.append((start,end-1))
        start, end = end , end + chunk
    pool = Pool(processes=cores)
    data = pool.map(OneChild,argsList)
    for d in data:
        total += len(d)
    print total

##    f = open("Result.txt", "w+")
##    f.write(str(total))
##    f.close()

def adjustCores(numRange,start,cores):
    if start == 1:
        start = 0
    else:
        pass
    while (numRange-start)%cores != 0:
        cores -= 1
    return cores

#MultiProcList(10**7)
from timeit import Timer
t = Timer(lambda: MultiProcList(10**6))
print t.timeit(number=1)
Was it helpful?

Solution

This is my fastest brute force code. It use cython to speedup the calculation. Instead of check all the numbers, it finds all the One-Child numbers by recursion.

%%cython
cdef int _one_child_number(int s, int child_count, int digits_count):
    cdef int start, count, c, child_count2, s2, part, i
    if s >= 10**(digits_count-1):
        return child_count
    else:
        if s == 0:
            start = 1
        else:
            start = 0
        count = 0
        for c in range(start, 10):
            s2 = s*10 + c
            child_count2 = child_count
            i = 10
            while True:
                part = s2 % i
                if part % digits_count == 0:
                    child_count2 += 1
                    if child_count2 > 1:
                        break
                if part == s2:
                    break
                i *= 10

            if child_count2 <= 1:
                count += _one_child_number(s2, child_count2, digits_count)
        return count 

def one_child_number(int digits_count):
    return _one_child_number(0, 0, digits_count)

To find the number of F(10**7), it takes about 100ms to get the result 277674.

print sum(one_child_number(i) for i in xrange(8))

You need 64bit integer to calculate large results.

EDIT: I added some comment, but my English is not good, so I convert the code to pure python code, and add some print to help you figure out how it works.

The _one_child_number adds digit from left to s recursively, child_count is the child count in s, digits_count is the final digits of s.

def _one_child_number(s, child_count, digits_count):
    print s, child_count
    if s >= 10**(digits_count-1): # if the length of s is digits_count
        return child_count # child_count is 0 or 1 here, 1 means we found one one-child-number.
    else:
        if s == 0: 
            start = 1 #if the length of s is 0, we choose from 123456789 for the most left digit.
        else:
            start = 0 #otherwise we choose from 0123456789 
        count = 0 # init the one-child-number count
        for c in range(start, 10): # loop for every digit
            s2 = s*10 + c  # add digit c to the right of s

            # following code calculates the child count of s2
            child_count2 = child_count 
            i = 10
            while True:
                part = s2 % i
                if part % digits_count == 0:
                    child_count2 += 1
                    if child_count2 > 1: # when child count > 1, it's not a one-child-number, break
                        break
                if part == s2:
                    break
                i *= 10

            # if the child count by far is less than or equal 1, 
            # call _one_child_number recursively to add next digit.
            if child_count2 <= 1: 
                count += _one_child_number(s2, child_count2, digits_count)
        return count 

Here is he ouput of _one_child_number(0, 0, 3), and the count of one-child-number of 3 digits is the sum of the second column that the first column is a 3 digits number.

0 0
1 0
10 1
101 1
104 1
107 1
11 0
110 1
111 1
112 1
113 1
114 1
115 1
116 1
117 1
118 1
119 1
12 1
122 1
125 1
128 1
13 1
131 1
134 1
137 1
14 0
140 1
141 1
142 1
143 1
144 1
145 1
146 1
147 1
148 1
149 1
15 1
152 1
155 1
158 1
16 1
161 1
164 1
167 1
17 0
170 1
171 1
172 1
173 1
174 1
175 1
176 1
177 1
178 1
179 1
18 1
182 1
185 1
188 1
19 1
191 1
194 1
197 1
2 0
20 1
202 1
205 1
208 1
21 1
211 1
214 1
217 1
22 0
220 1
221 1
222 1
223 1
224 1
225 1
226 1
227 1
228 1
229 1
23 1
232 1
235 1
238 1
24 1
241 1
244 1
247 1
25 0
250 1
251 1
252 1
253 1
254 1
255 1
256 1
257 1
258 1
259 1
26 1
262 1
265 1
268 1
27 1
271 1
274 1
277 1
28 0
280 1
281 1
282 1
283 1
284 1
285 1
286 1
287 1
288 1
289 1
29 1
292 1
295 1
298 1
3 1
31 1
311 1
314 1
317 1
32 1
322 1
325 1
328 1
34 1
341 1
344 1
347 1
35 1
352 1
355 1
358 1
37 1
371 1
374 1
377 1
38 1
382 1
385 1
388 1
4 0
40 1
401 1
404 1
407 1
41 0
410 1
411 1
412 1
413 1
414 1
415 1
416 1
417 1
418 1
419 1
42 1
422 1
425 1
428 1
43 1
431 1
434 1
437 1
44 0
440 1
441 1
442 1
443 1
444 1
445 1
446 1
447 1
448 1
449 1
45 1
452 1
455 1
458 1
46 1
461 1
464 1
467 1
47 0
470 1
471 1
472 1
473 1
474 1
475 1
476 1
477 1
478 1
479 1
48 1
482 1
485 1
488 1
49 1
491 1
494 1
497 1
5 0
50 1
502 1
505 1
508 1
51 1
511 1
514 1
517 1
52 0
520 1
521 1
522 1
523 1
524 1
525 1
526 1
527 1
528 1
529 1
53 1
532 1
535 1
538 1
54 1
541 1
544 1
547 1
55 0
550 1
551 1
552 1
553 1
554 1
555 1
556 1
557 1
558 1
559 1
56 1
562 1
565 1
568 1
57 1
571 1
574 1
577 1
58 0
580 1
581 1
582 1
583 1
584 1
585 1
586 1
587 1
588 1
589 1
59 1
592 1
595 1
598 1
6 1
61 1
611 1
614 1
617 1
62 1
622 1
625 1
628 1
64 1
641 1
644 1
647 1
65 1
652 1
655 1
658 1
67 1
671 1
674 1
677 1
68 1
682 1
685 1
688 1
7 0
70 1
701 1
704 1
707 1
71 0
710 1
711 1
712 1
713 1
714 1
715 1
716 1
717 1
718 1
719 1
72 1
722 1
725 1
728 1
73 1
731 1
734 1
737 1
74 0
740 1
741 1
742 1
743 1
744 1
745 1
746 1
747 1
748 1
749 1
75 1
752 1
755 1
758 1
76 1
761 1
764 1
767 1
77 0
770 1
771 1
772 1
773 1
774 1
775 1
776 1
777 1
778 1
779 1
78 1
782 1
785 1
788 1
79 1
791 1
794 1
797 1
8 0
80 1
802 1
805 1
808 1
81 1
811 1
814 1
817 1
82 0
820 1
821 1
822 1
823 1
824 1
825 1
826 1
827 1
828 1
829 1
83 1
832 1
835 1
838 1
84 1
841 1
844 1
847 1
85 0
850 1
851 1
852 1
853 1
854 1
855 1
856 1
857 1
858 1
859 1
86 1
862 1
865 1
868 1
87 1
871 1
874 1
877 1
88 0
880 1
881 1
882 1
883 1
884 1
885 1
886 1
887 1
888 1
889 1
89 1
892 1
895 1
898 1
9 1
91 1
911 1
914 1
917 1
92 1
922 1
925 1
928 1
94 1
941 1
944 1
947 1
95 1
952 1
955 1
958 1
97 1
971 1
974 1
977 1
98 1
982 1
985 1
988 1
Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top