Algebraic factors in sieve files

[gd_barnes]

Mark,

The new srfile included with the new srsieve does not appear to be working correctly specifically when removing factors. Analysis:

1. It shows the incorrect count on input terms. It is showing one too many. It is likely counting the header line.

2. It does not show output terms at all. It just stops after showing the number of factors removed.

3. I could not make it remove the algebraic factor forms. This might be ignorance on my part. Is there a switch that needs to be set to tell it to remove algebraic factor forms? I just used the standard -k switch followed by the factors file name.

Btw #1 and #2 were caught with parallel testing. :-)


Gary

[pepi37]

If Batalov correct his script, I will update "my program with GUI" there is no problem with that!

[rogue]

Originally Posted by gd_barnes

Mark,

The new srfile included with the new srsieve does not appear to be working correctly specifically when removing factors. Analysis:

1. It shows the incorrect count on input terms. It is showing one too many. It is likely counting the header line.

2. It does not show output terms at all. It just stops after showing the number of factors removed.

3. I could not make it remove the algebraic factor forms. This might be ignorance on my part. Is there a switch that needs to be set to tell it to remove algebraic factor forms? I just used the standard -k switch followed by the factors file name.

Btw #1 and #2 were caught with parallel testing. :-)

Gary

srfile does not look for algebraic factors.

I've put additional changes on hold until I build a list of all algebraic factors that srsieve should find and remove.

[gd_barnes]

Originally Posted by rogue

srfile does not look for algebraic factors.

I've put additional changes on hold until I build a list of all algebraic factors that srsieve should find and remove.

Really?

You completely failed to read most of my post. Did you read all of my post or just #3? #1 and #2 said nothing at all about algebraic factors. The program is now not displaying counts correctly irrespective of whether the factors are algebraic.

I am running PARALLEL tests against known good versions of both srsieve and srfile. How can I run a parallel test against the old version for algebraic factors? No version has ever removed algebraic factors. I stated at the bottom that the #1 and #2 problems were caught with parallel testing.

As I stated in #1 and #2. The input count is incorrect and the output count is completely missing and that applies whether the factors are algebraic or not.

I have run some additional testing and have realized more of the specifics and that the program is not working even if the factors are not algebraic as follows:

In #1, if the file is in -a format, it will display the input count correctly. If it is in -g, -G, or -w format, the input count is in incorrect.

In #3, it will not remove even regular factors. (This is something that I just now discovered.)

Further bug: It will not write out anything at all! I simply tried to convert a -a file to -G format. It did nothing.

Let's make this really simple. I am attaching an input file for 38*40^n-1 sieved to P=10G. I am also attaching some factors for it for P=10G-20G. I suggest that you try to remove the factors and write out a new file using your new version of srsieve. It will do nothing and it will display the input count as one too high. It will say:

Read 11790 sequences from input file 'sieve-R40.txt'.

There are only 11789 terms plus a header. The old srfile does it correctly. The new version then fails to remove any factors. It then fails to say what it wrote out like the old srfile does...likely because it did not write anything.

These are basic problems. I do not know why srfile is suddenly not working. Please test your new srfile and see that it removes nothing and will not write out anything. I understand that all changes are on hold but it seems like these are ones that should not have ever happened.

[gd_barnes]

Originally Posted by pepi37

If Batalov correct his script, I will update "my program with GUI" there is no problem with that!

Question: Since srfile does not remove algebraic factors how are you removing them after running them through Batalov's script?

[rogue]

Originally Posted by gd_barnes

Really?

You completely failed to read most of my post. Did you read all of my post or just #3? #1 and #2 said nothing at all about algebraic factors. The program is now not displaying counts correctly irrespective of whether the factors are algebraic..

I misunderstood what you wrote. This is all likely related to the issue where it doesn't read from input files.

[pepi37]

Originally Posted by gd_barnes

Question: Since srfile does not remove algebraic factors how are you removing them after running them through Batalov's script?

Batalov use "feature" of srfile
if you put 3333333333333333333333 | 4*155^5000052+1 it will remove factor for npg file

[rogue]

I want to make it clear that nobody should be using the 1.1.0 release. Due to reasons I will not disclose, I haven't had time to address the problems. I fully intend to address all bugs I've introduced in 1.1.1. Gary and I are discussing offline the various algebraic forms that it should be looking for.

[MisterBitcoin]

Any updates on this topic?

[rogue]

Originally Posted by MisterBitcoin

Any updates on this topic?

I just returned from Europe. I have spent some time on it and worked out some bugs, but have more testing to do.

[rogue]

I think that I have worked out the last of the bugs, at least with the algebraic factors. I need to verify that continuing an old sieve still works.