Missed Factors (TF)

[Dobri]

This thread is intended to collect information about factors obtained with trial factoring (TF) that have been previously missed due to:
1) Incomplete TF scans for small factors;
2) Faulty TF hardware/software;
3) Fake TF reports;
etc.

An example is
M618033947
for which the 68-bit factor 276413495291788087169 was found.
It was not previously in the GIMPS database and 0.3866 GHz-Days were given for finding it.
Note that zero (0.0000) GHz-Days are given for the TF verification of a known factor.

[Dobri]

A second example is

M726064763

for which three (3) factors (34153055801363324180057, 7170112396664438582504527, and 10702385060027676180416983) have been missed for an unknown reason in previous TF tests.

[Uncwilly]

Originally Posted by Dobri

A second example is

M726064763

for which three (3) factors (34153055801363324180057, 7170112396664438582504527, and 10702385060027676180416983) have been missed for an unknown reason in previous TF tests.

Those were not missed, expect by your examination of the page you linked to.

34153055801363324180057 was reported on 2016-06-14 by Anonymous and 2022-01-21 by LordJulius

2021-12-28 alpertron F-PM1 Factor: 7170112396664438582504527 / (P-1, B1=1000000, B2=10164924)
2021-12-28 Yzzyx F-PM1 Factor: 7170112396664438582504527 / (P-1, B1=1000000, B2=10164924)

2021-12-26 alpertron F-PM1 Factor: 10702385060027676180416983 / (P-1, B1=1000000, B2=1000000)

[Uncwilly]

Originally Posted by Dobri

M618033947
for which the 68-bit factor 276413495291788087169 was found.
It was not previously in the GIMPS database and 0.3866 GHz-Days were given for finding it.
Note that zero (0.0000) GHz-Days are given for the TF verification of a known factor.

It not being in the database does not mean it was missed. It was just not found yet. It had a very small factor and those there was no need to continue looking for more factors. We are trying to eliminate exponents that that we will have to run a primality test on. We are not doing exhaustive factoring on all exponents. And even if we were, it would be a long while before we got to that one.

[Dobri]

Originally Posted by Uncwilly

Those were not missed, expect by your examination of the page you linked to.

Said factors were already known indeed.
The point is that they were missed in subsequent TF tests.

[S485122]

Originally Posted by Dobri

...
3) Fake TF reports;
An example is
M618033947
...

You mean results like :
Factor: 12965740712769703 / TF: 1-53
Factor: 37915764094923857 / TF: 1-53
Factor: 37915764094923857 / TF: 53-54
Factor: 37915764094923857 / TF: 54-55
The first factor is 53,5 bits the second 55,0 bits. How can a 55 bits factor be found trial factoring from 1 to 53 bits ?
I suppose those where tests ?
As for some no factor results, they may be the result of finishing a bit level even after a factor was found, there used to be an option to do that, but now I can only find such a setting about ECM in undoc.txt.

[axn]

Originally Posted by Dobri

The point is that they were missed in subsequent TF tests.

Which "no factor" TF lines missed which factors?

[Dobri]

Originally Posted by S485122

You mean results like :
Factor: 12965740712769703 / TF: 1-53
Factor: 37915764094923857 / TF: 1-53
Factor: 37915764094923857 / TF: 53-54
Factor: 37915764094923857 / TF: 54-55
The first factor is 53,5 bits the second 55,0 bits. How can a 55 bits factor be found trial factoring from 1 to 53 bits ?
I suppose those where tests ?

The quoted results above were generated by the mfaktc app. Sometimes, mfaktc gives such output.

What I meant was that the user Anton Repko missed 3 factors in their TF 74-75, 82-83, and 83-84 tests.

[ATH]

Yes, that is a bit strange:

2022-03-07	Dobri	F	Factor: 10702385060027676180416983 / TF: 83-84
2022-02-23	Dobri	F	Factor: 7170112396664438582504527 / TF: 82-83
.
2022-01-21	LordJulius	F	Factor: 34153055801363324180057 / TF: 74-75
.
2022-01-01	Anton Repko	NF	no factor from 2^82 to 2^83
2022-01-01	Anton Repko	NF	no factor from 2^74 to 2^75
.
2021-12-30	Anton Repko	NF	no factor from 2^83 to 2^84

[Dobri]

Currently, in addition to the correct TF results, the server generates also incorrect TF results when processing intervals for which mfaktc provides both prime and composite factors.

For example, see the exponent M54647693 where for TF:67:68 mfaktc provides

one prime factor, 226554856699193904647, interpreted correctly by the server as "Factor: 226554856699193904647 / TF: 67-68",

and one composite factor, 161694036735902117647, for which the server unnecessarily gives "no factor from 2^67 to 2^68".

[kriesel]

Originally Posted by ATH

Yes, that is a bit strange:

Hardware fails.