Some CRUS stats

[rogue]

Originally Posted by gd_barnes

On the S53 problem, I just now tried deleting S52 and S53 from the difficulty file. Interestingly, S53 picked up the "n per 25000" from S52 instead of S48. It had previously gotten it's value from S48 before because you had me delete those 2 bases when fixing the original problem that started all of this. So we can conclude the following:

If a base has a problem that causes it to return an "n per 25000" of -2 when deleted from the difficulty file all by itself; if it is instead deleted in addition to another lower base above it, it will pick up the "n per 25000" from that base instead. So it appears that after each base is computed, a variable needs to be cleared out.

Edit: Further testing and review of the S53 *.in temp file confirmed that it is truncating the 4-digit k's before sending them to srsieve. It makes me wonder why this only happens on this base. There are many other bases of similar nature. In particular S53 has from 1 to 4-digit k's and truncates while R45 has from 2 to 5-digit k's but works fine.

It is likely a different bug in the script I just posted. I'll try to get it fixed later

[rogue]

Originally Posted by gd_barnes

Thanks Mark. That did it.

One more question: The "n per 25000" looks really high for both sides of base 2 even-n and odd-n. Could the program be sieving for all n instead? If so, programming it wouldn't be the easiest.

This may have been what mislead me into thinking that "regular" base 2 "n per 25000" was too low on both sides.

That is unlikely. I'll have to look at that as well.

gawk is a new language for me. I've clearly made a few mistakes in string manipulation. I appreciate that you are looking at the numbers in more detail.

[rogue]

Below is the output from srsieve for the various base 2 conjectures. The first three are Riesel, the remaining four are Sierpinski. The "n per 25000" on the webpage is the number of terms here multiplied by 1.25. Note that most of the remaining k have a very light weight.

srsieve started: 100001 <= n <= 110000, 3 <= p <= 1000000
Split 57 base 2 sequences into 89 base 2^24 subsequences.
Sieving 3 <= p <= 1000000 eliminated 565597 terms, 4403 remain.
Wrote 4403 terms for 57 sequences to srsieve format file `srsieve.out'.

srsieve started: 100001 <= n <= 110000, 3 <= p <= 1000000
Split 3 base 2 sequences into 31 base 2^24 subsequences.
Sieving 3 <= p <= 1000000 eliminated 27882 terms, 2118 remain.
Wrote 2118 terms for 3 sequences to srsieve format file `srsieve.out'.

srsieve started: 100001 <= n <= 110000, 3 <= p <= 1000000
Split 5 base 2 sequences into 44 base 2^24 subsequences.
Sieving 3 <= p <= 1000000 eliminated 46031 terms, 3969 remain.
Wrote 3969 terms for 5 sequences to srsieve format file `srsieve.out'.

srsieve started: 100001 <= n <= 110000, 3 <= p <= 1000000
Split 6 base 2 sequences into 9 base 2^24 subsequences.
Sieving 3 <= p <= 1000000 eliminated 59573 terms, 427 remain.
Wrote 427 terms for 6 sequences to srsieve format file `srsieve.out'.

srsieve started: 100001 <= n <= 110000, 3 <= p <= 1000000
Split 2 base 2 sequences into 13 base 2^12 subsequences.
Sieving 3 <= p <= 1000000 eliminated 18811 terms, 1189 remain.
Wrote 1189 terms for 2 sequences to srsieve format file `srsieve.out'.

srsieve started: 100001 <= n <= 110000, 3 <= p <= 1000000
Split 4 base 2 sequences into 20 base 2^12 subsequences.
Sieving 3 <= p <= 1000000 eliminated 37485 terms, 2515 remain.
Wrote 2515 terms for 4 sequences to srsieve format file `srsieve.out'.

srsieve started: 100001 <= n <= 110000, 3 <= p <= 1000000
Split 31 base 2 sequences into 118 base 2^72 subsequences.
Sieving 3 <= p <= 1000000 eliminated 307475 terms, 2525 remain.
Wrote 2525 terms for 31 sequences to srsieve format file `srsieve.out'.

[gd_barnes]

Originally Posted by rogue

Below is the output from srsieve for the various base 2 conjectures. The first three are Riesel, the remaining four are Sierpinski. The "n per 25000" on the webpage is the number of terms here multiplied by 1.25. Note that most of the remaining k have a very light weight.

You are not removing all odd n from base 2 even n conejctures and are not removing all even n from base 2 odd n conjectures. That needs to be programmed for. Even n and odd n conjectures are different from all other bases.

I just now tested this. There are many odd n remaining in the even n file. Riesel base 2, Sierp base 2, and Sierp base 2 2nd conj (i.e. the "regular" conjectures) are all correct and have an extremely low weight. But...Riesel base 2 even-n, Riesel base 2 odd-n, Sierp base 2 even-n, and Sierp base 2 odd-n are all incorrect. They "appear" to have a high weight when in fact they have a low weight, similar to regular conjectures.

[gd_barnes]

Originally Posted by gd_barnes

You are not removing all odd n from base 2 even n conejctures and are not removing all even n from base 2 odd n conjectures. That needs to be programmed for. Even n and odd n conjectures are different from all other bases.

I just now tested this. There are many odd n remaining in the even n file. Riesel base 2, Sierp base 2, and Sierp base 2 2nd conj (i.e. the "regular" conjectures) are all correct and have an extremely low weight. But...Riesel base 2 even-n, Riesel base 2 odd-n, Sierp base 2 even-n, and Sierp base 2 odd-n are all incorrect. They "appear" to have a high weight when in fact they have a low weight, similar to regular conjectures.

Some "quasi" pseudo-code for base 2 even n and odd n "n per 25000" or weight:

Riesel and Sierp base 2 even n:
Multiply the base by 2 and keep the k's the same and then run srsieve. Base 4 will give the same effect as base 2 even n.

Riesel and Sierp base 2 odd n:
Multiply the base by 2 and multiply the k's by 2 and then run srsieve. Base 4 with the k's doubled will give the same effect as base 2 odd n.

You should end up with some extremely low weight k's, just like the "regular" base 2 conjectures. The way you have it now makes the even n and odd n conjecture k's look very high weight.

[rogue]

I see now. It definitely doesn't remove the "even" or "odd" n as needed. I should be able to change the script to do that.

[rogue]

I have attached the current scripts of which three were updated. The zip file also includes the current difficulties/weights.

One script was updated to fix the problem with S53. I suspect that some other bases might have had a similar issue, although it might not have been as obvious as this one. The other two updated scripts with the even/odd conjectures. I also fixed another very small problem that caused the script to overcount remaining n after sieving. That affects all bases, which is why I've included updated difficulties/weights. I was counting number of k * 2 extra for the values.

[gd_barnes]

Only one of the 2 problems was fixed and another one was created:

1. The S53 problem is fixed but the problem with R2 even-n, R2 odd-n, S2 even-n, and S2 odd-n still exists; just not quite as bad as before.

2. New problem: The "tested n" column on the top 20 page and the unproven conjecture page has an extra right parenthese ")" after it on many bases. This only seems to happen on bases with <= 25 k's remaining.

Below are the n's that you should have remaining for R2 even-n, R2 odd-n, S2 even-n, and S2 odd-n on a sieve to P=1M for n=100001-110000. Keep in mind that you should multiply the base by 2 (i.e. always base 4) before sending to srsieve. Also...for odd-n only, all k's must be multiplied by 2.

n's that should be remaining in the difficulty.txt files:
R2 even-n (3 k's): 621
R2 odd-n (5 k's): 935
S2 even-n (2 k's): 488
S2 odd-n (4 k's): 513

This should result in "n per 25000 at 1e12" as shown on the web pages of:
R2 even-n: 776
R2 odd-n: 1169
S2 even-n: 610
S2 odd-n: 641

To make sure we're on the same page, here are examples of 2 of the files that should be sent to srsieve:
R2 even-n:
9519*4^n-1
14361*4^n-1
19401*4^n-1

S2 odd-n:
18534*4^n+1
64494*4^n+1
106266*4^n+1
168726*4^n+1


Gary

[gd_barnes]

I found another problem that I don't think previously existed before all of this started:

On the CRUS statistics and progress page, it is showing 0 bases "not started" in both of the first two tables. This is also distorting the percentages for the various categories.

I'm pretty sure this was correct prior to this entire recent exchange but I don't know which of the recent versions of the scripts caused it to become incorrect.

[rogue]

Originally Posted by gd_barnes

I found another problem that I don't think previously existed before all of this started:

On the CRUS statistics and progress page, it is showing 0 bases "not started" in both of the first two tables. This is also distorting the percentages for the various categories.

I'm pretty sure this was correct prior to this entire recent exchange but I don't know which of the recent versions of the scripts caused it to become incorrect.

It was working previously, but off the top of my head I don't know what I would have done to break that.

[rogue]

Originally Posted by gd_barnes

Only one of the 2 problems was fixed and another one was created:

1. The S53 problem is fixed but the problem with R2 even-n, R2 odd-n, S2 even-n, and S2 odd-n still exists; just not quite as bad as before.

2. New problem: The "tested n" column on the top 20 page and the unproven conjecture page has an extra right parenthese ")" after it on many bases. This only seems to happen on bases with <= 25 k's remaining.

Below are the n's that you should have remaining for R2 even-n, R2 odd-n, S2 even-n, and S2 odd-n on a sieve to P=1M for n=100001-110000. Keep in mind that you should multiply the base by 2 (i.e. always base 4) before sending to srsieve. Also...for odd-n only, all k's must be multiplied by 2.

n's that should be remaining in the difficulty.txt files:
R2 even-n (3 k's): 621
R2 odd-n (5 k's): 935
S2 even-n (2 k's): 488
S2 odd-n (4 k's): 513

This should result in "n per 25000 at 1e12" as shown on the web pages of:
R2 even-n: 776
R2 odd-n: 1169
S2 even-n: 610
S2 odd-n: 641

I understand where you are coming from, but what you are computing and what I am computing are not the same. If I change the input base to srsieve, then the range of n for the input base is no longer the same, i.e. sieving 9519*4^n-1 from 10e4 to 11e4 is not the same as sieving 9519*2^n-1 10e4 to 11e4 from 10e4 to 11e4. Sieving 9519*4^n-1 from 10e4 to 11e4 is closer to sieving 9519*2^n-1 20e4 to 22e4. My method is to sieve base 2, then only count the even/odd n in the output based upon the conjecture.