January 2022

[Max0526]

Originally Posted by Walter

Just to add on to my previous response: the number of primes that can be generated with the max and min wheel in my solution corresponds to the same values that Zoozie posted (741 primes for the min wheel and 1992 primes for the max wheel). Now, to be fair, I see multiple wheels that generate the same number of primes, but it gives me some additional confidence that my solution is right.

If I skip 5 and 8 in my circle: PRIMES: 1860, CMAX = 21258 > 19690.
Greater number of primes doesn't necessarily mean higher score of the circle.
Also, below 8300 for the min so far.

[Walter]

Oops. I had hard coded something for the n=7/d=5 case. After changing this, I am getting the following values:

min: 8265
max: 23209

Anyone else getting those values?

[Zoozie]

I avoided posting the scores since this could make people skip exhaustive search to find the solution.

[Max0526]

Originally Posted by Walter

Oops. I had hard coded something for the n=7/d=5 case. After changing this, I am getting the following values:

min: 8265
max: 23209

Anyone else getting those values?

Got the same min/max!

Originally Posted by Zoozie

It took about 2 minutes to solve in java for n=7,d=5. I did not really optimize except first generating cache of small primes. The n=8, d=6 was solved in some hours, did not time it.

For n=8, d=6:
wheeel for maximum solution had 1992 primes
wheel for minimum solution had 741 primes.

And the same number of primes!

[Max0526]

Just for the fun of it, anybody wants to try coding n = 9, d = 7 and n = 10, d = 8 cases?

[Walter]

Originally Posted by Max0526

Just for the fun of it, anybody wants to try coding n = 9, d = 7 and n = 10, d = 8 cases?

With n = 9, d = 7 I get the following values:

min: 117390
max: 194304

And with n = 10, d = 8:

min: 1747537
max: 1772281

The runtime was about 13000 seconds for the n = 9 case. For the n = 10 case, I used 4 threads, which resulted in a runtime of 12400 seconds.

[Max0526]

With n = 9, d = 7 I get the following values:

min: 117390
max: 194304

And with n = 10, d = 8:

min: 1747537
max: 1772281

Massive amount of computation is done here!
I need to rewrite Python into Java to be able to compete with your run times.
Will post my results when it's done.
Send these min/max results to IBM too. Sometimes they award **.

[SmartMersenne]

Originally Posted by Max0526

Send these min/max results to IBM too. Sometimes they award **.

Not with the new puzzlemaster. He is lazy to do anything extra or even the minimum expected things on time.

[Zoozie]

Originally Posted by SmartMersenne

Not with the new puzzlemaster. He is lazy to do anything extra or even the minimum expected things on time.

Agree! He updates scores only 2 times a month. We can still not see who has solved this month.


Also solution to December challenge still gives a 404.

[Max0526]

Originally Posted by Zoozie

Agree! He updates scores only 2 times a month. We can still not see who has solved this month.


Also solution to December challenge still gives a 404.

Both are available as of this morning.

[dg211]

I get the same answers for n = 9, d = 7 and n = 10, d = 8. It is possible to do it with a lot less computation - my single threaded C++ code took about 3.5s for n = 9, d = 7 and about 32s for n = 10, d = 8. Almost all of the time for n = 10 (like 31.9s) was spent generating a list of 8 digit primes.

Originally Posted by Walter

With n = 9, d = 7 I get the following values:

min: 117390
max: 194304

And with n = 10, d = 8:

min: 1747537
max: 1772281

The runtime was about 13000 seconds for the n = 9 case. For the n = 10 case, I used 4 threads, which resulted in a runtime of 12400 seconds.