Problems & Puzzles:
Puzzles
Puzzle 45. Prime
Solitaire (a puzzle based on a game
proposed by Enoch Haga)
Enoch Haga (18/03/99) proposed the following
game:
“Starting from an
arbitrary prime of D digits, construct a sequence of
distinct primes such that every one is formed by the
(D1) rightmost digits of the previous one, plus any odd
digit (except 5)”
He gives the following example of a sequence of 14
members of primes of 3 digits (D=3):
{131, 311, 113, 137, 373, 733, 331, 313, 139, 397,
971, 719, 191, 911}
Obviously this kind of sequences are finite because
reaching at some point you have not any other prime to
select. Less evident is to construct the largest possible
sequence of primes of D digits, for every D=2, 3,….
Question: Get the
largest sequence of this kind, for D = 2, 3, 4, 5, 6
& 7
Hints (by C. Rivera):
Let’s define L(D) as the quantity of members of a
sequence of D digits. Let’s define also Podd(D) as
the quantity of primes of D digits such that all the
digits of these primes are odd (except 5). It is not so
hard to recognize that:
L(D) = Podd(D) +(D – 1)
With the help of a simple code you can calculate
Podd(D). This are the results:
Digits
in the primes (D) 
Quantity
of primes composed of odd digits, except 5 
Podd(D)
+(D1) 
Found 
2 
10 
11 
11 
3 
30 
32 
26 
4 
63 
66 
31 
5 
249 
253 
47 
6 
757 
762 
29 
7 
2709 
2715 
23 
So, the third column is the limit of
the length of this kind of sequences for each D.
For primes of two digits (D=2) it’s a kind of
easy to get the largest sequence:
D = 2: 41, 13, 31, 19,
97, 71, 11, 17, 73, 37, 79; L(D=2)
= 11
In this case we can conclude that there is not any
other larger sequence than the before.
For the case D=3, the largest sequence I have produced
is this:
D = 3: 241, 419, 191,
919, 193, 937, 373, 733, 331, 317, 179, 797, 977, 773,
739, 397, 971, 719, 199, 991, 911, 113, 131, 313, 137,
379; L(D=3) = 26 < 32
So for this case it can exists a larger sequence.
Solution
At 9/5/99 Wilfred Whiteside wrote:
"This morning I worked on Puzzle 45 (Prime
Solitaire). I wrote a recursive subroutine that
works through all possible games. Below are the
results.
For 3
digits there are 265,655 games, 1820 of which result in 26 primes!
None have more than 26 primes!
For 4 digits
there are 44,713 games, 39 of which result in 31 primes.
None have more than 31 primes.
For 5
digits there are 377,133 games, 29 of which result in 47 primes! None have more
than 47 primes.
For 6
digits there are 92,955 games, 48 of which result in 29 primes.
None have more than 29 primes.
For 7
digits there are 488,102 games, 60 of which result in 23 primes.
None have more than 23 primes".
Example best games:
3 Digits:
941 419 199 991 919 193 937 379 797 977 773 739 397 971
719
191 911 113 131 313 137 373 733 331 317 179. There were 26
primes in the sequence
4 digits:
1451 4519 5197 1979 9791 7919 9199 1997 9973 9733 7333
3331 3313 3137 1373 3739 7393 3931 9311 3119 1193 1933
9337 3371 3719 7193 1931 9319 3191 1913 9133. There were 31
primes in the sequence
5 digits:
11483 14831 48313 83137 31379 13799 37993 79939 99397
93979 39791 97919 79193 91939 19391 93911 39113 91139
11399 13997 39979 99793 97931 79319 93199 31991 19913
99139 91393 13931 39313 93133 31333 13331 33317 33179
31793 17939 79393 93937 39371 93719 37199 71999 19997
99971 99713 .There were 47 primes in the
sequence!
6 digits:
158747 587473 874739 747391 473911 739117 391177 911777
117779 177791 777911 779111 791117 911173 111731 117319
173191 731911 319117 191173 911737 117371 173713 737131
371311 713117 131171 311711 117119. There were 29
primes in the sequence
7 digits:
1187587 1875877 8758777 7587779 5877797 8777971 7779719
7797191 7971911 9719113 7191131 1911311 9113113 1131131
1311311 3113111 1131113 1311131 3111313 1113137 1131379
1313797 3137977. There were 23 primes in
the sequence.
***
On November, 2005, Luke Pebody wrote:
8 digits 25 primes (max. possible)
73938743
39387433
93874337
38743379
87433799
74337997
43379977
33799771
33799771
37997717
79977173
99771731
97717313
77173139
71731391
17313911
73139111
31391119
13911193
39111931
91119319
11193199
11931991
19319917
93199177
31991779
9 digits  26 primes (max. possible)
352769419
527694199
276941999
769419997
694199971
941999711
419997113
199971133
999711337
997113379
971133799
711337999
113379997
133799977
337999771
379997713
799977137
999771379
997713793
977137937
771379379
713793797
137937971
379379713
793797131
937971317
10 digits  26 primes (max possible)
4779418529
7794185291
7941852917
9418529179
4185291797
1852917971
8529179717
5291797177
2917971773
9179717731
1797177311
7971773111
9717731117
7177311173
1773111737
7731117371
7311173711
3111737119
1117371197
1173711971
1737119717
7371197179
3711971797
7119717971
1197179713
1971797131
Luke's Question: What happens for very large
number of digits.
***
Wildred Whiteside wrote (June, 2007):
11 digits  23 primes (max possible of all
1,547,288,309 games)
24182791183
41827911833
18279118331
82791183317
27911833171
79118331713
91183317131
11833171313
18331713137
83317131377
33171313771
31713137719
17131377191
71313771913
13137719131
31377191311
13771913117
37719131179
77191311797
71913117977
19131179777
91311797771
13117977713
12 digits  27 primes (max possible of all 12,612,606,670 games)
152153734847
521537348477
215373484771
153734847719
537348477191
373484771911
734847719119
348477191191
484771911917
847719119173
477191191733
771911917333
719119173337
191191733377
911917333771
119173337719
191733377191
917333771911
173337719117
733377191177
333771911777
337719117779
377191177799
771911777999
719117779993
191177799931
911777999311
13 digits  26 primes (max possible of all 102,463,733,414 games)
1362682965781
3626829657811
6268296578119
2682965781191
6829657811917
8296578119173
2965781191733
9657811917337
6578119173377
5781191733773
7811917337737
8119173377377
1191733773773
1917337737737
9173377377379
1733773773793
7337737737937
3377377379371
3773773793711
7737737937113
7377379371137
3773793711371
7737937113719
7379371137197
3793711371977
7937113719779
Luke asked what happens for large numbers of digits. By looking at
histograms of the number of N digit games making it to M moves, can
see an exponential decay with the number of moves. Below is a crude
approximation that predicts the max. number of moves M as a function
of the number of digits N.
For large number of digits N, the number of starting primes (those N
digit numbers with no 0's that are prime) is approximately
10*9^(N1) / ln(10^(N0.33))
Example: For the case if N=9, plugging in gives 21562806 predicted
starting primes, but actual number is 21545970.
The fraction of random N digit numbers ending in a 1,3,7, or 9 (nums
not divis by 2 or 5) is approx. 2.5/ln(10^(N0.33)). So on average,
the fraction of N1 digit numbers that can generate an N digit prime
by adding a 1,3,7, or 9 is approximately 10/ln(10^(N0.33)). This is
overly simplistic, but is similar to what the histograms show.
Solving for how many moves would lead to only one game left requires
that [10*9^(N1)/ln(10^(N0.33))] * [10/ln(10^(N0.33))]^(M1) = 1
This gives that M is approximately [(N1)*ln(9)]/ln((N0.33)*ln(10)/10).
Spreadsheeting this shows a prediction that the max number of moves
reaches a minimum for N=12 and then slowly increases for larger N.
for extremely large N, this limits crudely to approx. ln(9)*[N/ln(N)]
which slowly grows without bound.
***
