GHz-Days credit

[Miszka]

5000111 has a factor of 10000223.
I tried to find another factor by three methods: ECM; TF and PP1.
For the effect, see the attachment.
GHz-Days credit was calculated for two methods (ECM and PP1).
For the TF method, no.
The TF method has added value to the GIMPS project and should be appreciated with GHz-Days credit.
Something's not right here.

[masser]

Originally Posted by Miszka

5000111 has a factor of 10000223.
I tried to find another factor by three methods: ECM; TF and PP1.
For the effect, see the attachment.
GHz-Days credit was calculated for two methods (ECM and PP1).
For the TF method, no.
The TF method has added value to the GIMPS project and should be appreciated with GHz-Days credit.
Something's not right here.

I think you're essentially correct. Additional TF effort on partially factored Mersenne numbers should be recorded and the work credited. However, user TJAOI has systematically TF'ed (we think) all Mersenne numbers to 2^67 and continues to increase that level. Search the forum for discussions about his work. So, if you get credited for the above work, it will only be for the last bit level, 2^67 to 2^68. Also, consider that for very small exponents (<1M), the amount of ECM/P-1 already performed makes additional TF a wasted effort; the project may want to discourage TF in those ranges, so credit may not be given for wasteful TF on small exponents.

[Miszka]

Originally Posted by masser

Also, consider that for very small exponents (<1M), the amount of ECM/P-1 already performed makes additional TF a wasted effort; the project may want to discourage TF in those ranges, so credit may not be given for wasteful TF on small exponents.

Your "theory" is not confirmed in the following case.
TJAOI has probably tested all Mersenne numbers with a found factor up to the TF68 range, but it only reports results when it finds another factor.
There is nothing in the example provided M500000003 that the Mersenne count tested was tested by TJAOI or anyone using the P-1/ECM method.

[nordi]

Originally Posted by Miszka

Your "theory" is not confirmed in the following case.

But it is also not disproven by it, so the example is quite pointless. Please take masser's advise and search the forum, there has been in depth analysis of TJAOI's work.

Originally Posted by Miszka

TJAOI has probably tested all Mersenne numbers with a found factor up to the TF68 range

This is incorrect, as proven by finding a <68 bit factor for 11944129 just yesterday. TJAOI is, however, actively working on the 68 bit range as you can see here.

[Miszka]

Originally Posted by nordi

But it is also not disproven by it, so the example is quite pointless. Please take masser's advise and search the forum, there has been in depth analysis of TJAOI's work.

This is incorrect, as proven by finding a <68 bit factor for 11944129 just yesterday. TJAOI is, however, actively working on the 68 bit range as you can see here.

Sometimes there are exceptions, or perhaps rather errors.
For example, I found a factor 6864212346810681024459763168 = 2^5 × 11 × 83 × 381233 × 1081139 × 57003029 + 1 by method P-1 for 57003029 (B1=1800000).
This factor didn't find user mgpainter who tested this Mersenne number for B1=605000 and B2=14066250.

[alpertron]

Originally Posted by Miszka

Your "theory" is not confirmed in the following case.
TJAOI has probably tested all Mersenne numbers with a found factor up to the TF68 range, but it only reports results when it finds another factor.
There is nothing in the example provided M500000003 that the Mersenne count tested was tested by TJAOI or anyone using the P-1/ECM method.

TJAOI is not using Prime95 for trial factoring. He finds the Mersenne number from the prime factor by using another application. That's why he completed the 67-bit range but there are only a few reports of him of factors not found.

If you look at the table that includes more details, you will find that TJAOI found more prime factors in the range 64 to 67 bits than all other users combined.

You can see his progress at https://www.mersenne.ca/tjaoi.php . Most of the factors he finds are monotonically increasing. He is also working on other numbers trying to factor them in the range 69 to 71 bits where he probably uses Prime95.