Skip to main content
Code Golf

Return to Answer

added 9 characters in body
Source Link
Tim S.
Tim S.
  • 615
  • 4
  • 10

PFGW, 6067 digits, {5956}7{110}

Run PFGW with the following input file and -f100 to prefactor the numbers. In about 2-3 CPU minutes on my computer (i5 Haswell), it finds the PRP (10^(6073-6)-1)/9+6*10^110, or {5956}7{110}. I chose 6000 digits as the starting point as a nothing-up-my-sleeve number that's a little higher than all previous submissions.

ABC2 $a-$b & (10^($a-$b)-1)/9+$b*10^$c
a: primes from 6000 to 6200
b: in { 2 4 6 }
c: from 0 to 5990

Based on how quickly I was able to find this one, I could easily bump up the number of digits and still find a PRP within an hour. With how the rules are written, I might even just find the size where my CPU, running on all 4 cores, can finish one PRP test in an hour, take a long time to find a PRP, and have my "search" consist solely of the one PRP.

P.S. In some senses, this isn't a "code" solution because I didn't write anything besides the input file...but then, many one-line Mathematica solutions to mathematical problems could be described in the same way, as could using a library that does the hard work for you. In reality, I think it's hard to draw a good line between the two. If you like, I could write a script that creates the PFGW input file and calls PFGW. The script could even search in parallel, to use all 4 cores and speed up the search by ~4 times (on my CPU).

P.P.S. I think LLR can do the PRP tests for these numbers, and I'd expect it to be far faster than PFGW. A dedicated sieving program could also be better at factoring these numbers than PFGW's one-at-a-time. If you combined these, I'm sure you could push the bounds much higher than current solutions.

PFGW, 6067 digits, {5956}7{110}

Run PFGW with the following input file and -f100 to prefactor the numbers. In about 2-3 CPU minutes on my computer (i5 Haswell), it finds the PRP (10^(6073-6)-1)/9+6*10^110, or {5956}7{110}. I chose 6000 digits as the starting point as a nothing-up-my-sleeve number that's a little higher than all previous submissions.

ABC2 $a-$b & (10^($a-$b)-1)/9+$b*10^$c
a: primes from 6000 to 6200
b: in { 2 4 6 }
c: from 0 to 5990

Based on how quickly I was able to find this one, I could easily bump up the number of digits and still find a PRP within an hour. With how the rules are written, I might even just find the size where my CPU, running on all 4 cores, can finish one PRP test in an hour, take a long time to find a PRP, and have my "search" consist solely of the one PRP.

P.S. In some senses, this isn't a "code" solution because I didn't write anything besides the input file...but then, many Mathematica solutions to mathematical problems could be described in the same way, as could using a library that does the hard work for you. In reality, I think it's hard to draw a good line between the two. If you like, I could write a script that creates the PFGW input file and calls PFGW. The script could even search in parallel, to use all 4 cores and speed up the search by ~4 times (on my CPU).

P.P.S. I think LLR can do the PRP tests for these numbers, and I'd expect it to be far faster than PFGW. A dedicated sieving program could also be better at factoring these numbers than PFGW's one-at-a-time. If you combined these, I'm sure you could push the bounds much higher than current solutions.

PFGW, 6067 digits, {5956}7{110}

Run PFGW with the following input file and -f100 to prefactor the numbers. In about 2-3 CPU minutes on my computer (i5 Haswell), it finds the PRP (10^(6073-6)-1)/9+6*10^110, or {5956}7{110}. I chose 6000 digits as the starting point as a nothing-up-my-sleeve number that's a little higher than all previous submissions.

ABC2 $a-$b & (10^($a-$b)-1)/9+$b*10^$c
a: primes from 6000 to 6200
b: in { 2 4 6 }
c: from 0 to 5990

Based on how quickly I was able to find this one, I could easily bump up the number of digits and still find a PRP within an hour. With how the rules are written, I might even just find the size where my CPU, running on all 4 cores, can finish one PRP test in an hour, take a long time to find a PRP, and have my "search" consist solely of the one PRP.

P.S. In some senses, this isn't a "code" solution because I didn't write anything besides the input file...but then, many one-line Mathematica solutions to mathematical problems could be described in the same way, as could using a library that does the hard work for you. In reality, I think it's hard to draw a good line between the two. If you like, I could write a script that creates the PFGW input file and calls PFGW. The script could even search in parallel, to use all 4 cores and speed up the search by ~4 times (on my CPU).

P.P.S. I think LLR can do the PRP tests for these numbers, and I'd expect it to be far faster than PFGW. A dedicated sieving program could also be better at factoring these numbers than PFGW's one-at-a-time. If you combined these, I'm sure you could push the bounds much higher than current solutions.

added 173 characters in body
Source Link
Tim S.
Tim S.
  • 615
  • 4
  • 10

PFGW, 6067 digits, {5956}7{110}

Run PFGW with the following input file and -f100, in to prefactor the numbers. In about 2-3 CPU minutes on my computer (i5 Haswell), it finds the PRP (10^(6073-6)-1)/9+6*10^110, or {5956}7{110}. I chose 6000 digits as the starting point as a nothing-up-my-sleeve number that's a little higher than all previous submissions.

ABC2 $a-$b & (10^($a-$b)-1)/9+$b*10^$c
a: primes from 6000 to 6200
b: in { 2 4 6 }
c: from 0 to 5990

Based on how quickly I was able to find this one, I could easily bump up the number of digits and still find a PRP within an hour. With how the rules are written, I might even just find the size where my CPU, running on all 4 cores, can finish one PRP test in an hour, take a long time to find a PRP, and have my "search" consist solely of the one PRP.

P.S. In some senses, this isn't a "code" solution because I didn't write anything besides the input file...but then, many Mathematica solutions to mathematical problems could be described in the same way, as could using a library that does the hard work for you. In reality, I think it's hard to draw a good line between the two. If you like, I could write a script that creates the PFGW input file and calls PFGW. The script could even search in parallel, to use all 4 cores and speed up the search by ~4 times (on my CPU).

P.P.S. I think LLRLLR can do the PRP tests for these numbers, and I'd expect it to be far faster than PFGW. A dedicated sieving program could also be better at factoring these numbers than PFGW's one-at-a-time. If you combined these, I'm sure you could push the bounds much higher than current solutions.

PFGW, 6067 digits, {5956}7{110}

Run PFGW with the following input file and -f100, in about 2 CPU minutes on my computer, it finds the PRP (10^(6073-6)-1)/9+6*10^110, or {5956}7{110}. I chose 6000 digits as the starting point as a nothing-up-my-sleeve number that's a little higher than all previous submissions.

ABC2 $a-$b & (10^($a-$b)-1)/9+$b*10^$c
a: primes from 6000 to 6200
b: in { 2 4 6 }
c: from 0 to 5990

Based on how quickly I was able to find this one, I could easily bump up the number of digits and still find a PRP within an hour. With how the rules are written, I might even just find the size where my CPU, running on all 4 cores, can finish one PRP test in an hour, take a long time to find a PRP, and have my "search" consist solely of the one PRP.

P.S. In some senses, this isn't a "code" solution because I didn't write anything besides the input file...but then, many Mathematica solutions to mathematical problems could be described in the same way, as could using a library that does the hard work for you. In reality, I think it's hard to draw a good line between the two. If you like, I could write a script that creates the PFGW input file and calls PFGW. The script could even search in parallel.

P.P.S. I think LLR can do the PRP tests for these numbers, and I'd expect it to be far faster than PFGW. A dedicated sieving program could also be better at factoring these numbers than PFGW's one-at-a-time. If you combined these, I'm sure you could push the bounds much higher than current solutions.

PFGW, 6067 digits, {5956}7{110}

Run PFGW with the following input file and -f100 to prefactor the numbers. In about 2-3 CPU minutes on my computer (i5 Haswell), it finds the PRP (10^(6073-6)-1)/9+6*10^110, or {5956}7{110}. I chose 6000 digits as the starting point as a nothing-up-my-sleeve number that's a little higher than all previous submissions.

ABC2 $a-$b & (10^($a-$b)-1)/9+$b*10^$c
a: primes from 6000 to 6200
b: in { 2 4 6 }
c: from 0 to 5990

Based on how quickly I was able to find this one, I could easily bump up the number of digits and still find a PRP within an hour. With how the rules are written, I might even just find the size where my CPU, running on all 4 cores, can finish one PRP test in an hour, take a long time to find a PRP, and have my "search" consist solely of the one PRP.

P.S. In some senses, this isn't a "code" solution because I didn't write anything besides the input file...but then, many Mathematica solutions to mathematical problems could be described in the same way, as could using a library that does the hard work for you. In reality, I think it's hard to draw a good line between the two. If you like, I could write a script that creates the PFGW input file and calls PFGW. The script could even search in parallel, to use all 4 cores and speed up the search by ~4 times (on my CPU).

P.P.S. I think LLR can do the PRP tests for these numbers, and I'd expect it to be far faster than PFGW. A dedicated sieving program could also be better at factoring these numbers than PFGW's one-at-a-time. If you combined these, I'm sure you could push the bounds much higher than current solutions.

Source Link
Tim S.
Tim S.
  • 615
  • 4
  • 10

PFGW, 6067 digits, {5956}7{110}

Run PFGW with the following input file and -f100, in about 2 CPU minutes on my computer, it finds the PRP (10^(6073-6)-1)/9+6*10^110, or {5956}7{110}. I chose 6000 digits as the starting point as a nothing-up-my-sleeve number that's a little higher than all previous submissions.

ABC2 $a-$b & (10^($a-$b)-1)/9+$b*10^$c
a: primes from 6000 to 6200
b: in { 2 4 6 }
c: from 0 to 5990

Based on how quickly I was able to find this one, I could easily bump up the number of digits and still find a PRP within an hour. With how the rules are written, I might even just find the size where my CPU, running on all 4 cores, can finish one PRP test in an hour, take a long time to find a PRP, and have my "search" consist solely of the one PRP.

P.S. In some senses, this isn't a "code" solution because I didn't write anything besides the input file...but then, many Mathematica solutions to mathematical problems could be described in the same way, as could using a library that does the hard work for you. In reality, I think it's hard to draw a good line between the two. If you like, I could write a script that creates the PFGW input file and calls PFGW. The script could even search in parallel.

P.P.S. I think LLR can do the PRP tests for these numbers, and I'd expect it to be far faster than PFGW. A dedicated sieving program could also be better at factoring these numbers than PFGW's one-at-a-time. If you combined these, I'm sure you could push the bounds much higher than current solutions.