Duplication of Effort for Smaller Aliquot Sequences

[EdH]

This has been brought up previous, but I thought I'd address it again, since I was reviewing the "Blue Page" recently and it came back to mind.

Since I probably qualify as an elf in my composite work with the db, I'm concerned about duplication of effort with such entities as yafu@home. I'm not familiar with the project enough to be sure, but if the system in use updates the db with each new line for a sequence, that new line holds a composite with <=100 digits, and the system continues with the new composite itself, there may be duplication.

Right now there are very few composites <100 dd in the db. When one shows up, it is quite often taken for factorization within a few seconds. If that same composite has already been in work by yafu@home, or another entity using a procedure that sends each new line to the db, there is duplication. (See, Is there a tool that picks off small composites constantly?.)

For the above reasons, I would like to suggest (again) that if an entity is working a sequence that reaches the db's major composite backlog, (currently the low end is 100 dd), that it either be left for the db to work**, either to termination or return above the backlog, or the sequence could be worked locally instead of updating the db at that time.

I'm not trying to pick on anyone, e.g. yafu@home, but I have seen many of my C100s turned away by the db due to already being factored by the time my machine finished. For my antique machines, a C100 can represent anywhere from twenty minutes to an hour and a half.

**If the db does terminate a sequence the credit should still be claimed by the reservation holder.

Further comments, critiques, etc. accepted...

[EdH]

Too late to add an edit...

I'm not just seeing already factored C100s. I'm seeing it for all the area from 79 dd through 100 dd, and I'm not seeing a huge number, but I've de-conflicted my machines by adding a check file, which "mostly" prevents one from working the same number as another of my machines. I also realize that there are other workers that are solely doing composites in the same area (at least C100s) and these duplications will be unavoidable.

However, I'm not sure how many may be in a situation similar to the thread I referenced earlier. I know there is at least one perl script that runs Aliquot sequences based on the last line in the db, as I have also used that script in the past.

The reason I mentioned yafu@home was because of all the reservations I saw that were passing through the 79-100 dd area. In fact, that crossing may not be as frequent as one would believe from the table on the "Blue Page," since it is a static snapshot. But, I did watch a couple of the sequences in the db and saw that my machines took many of the composites as they showed up. Without a bit of work, I don't have a method to see if my machine or yafu@home's system, (or maybe someone else), may have turned in the factors.

Of course, the overall goal is to advance the db and I'm happy doing my part. I only bring this up in case there is a manner in which we might minimize some duplication.

[yoyo]

Keep in mind, that factoring a composite doesn't increase the Aliquot Secenece in factordb. To add the next step to the Aliquot Sequence you have to fetch the sequence from factordb.

[EdH]

Originally Posted by yoyo

Keep in mind, that factoring a composite doesn't increase the Aliquot Secenece in factordb. To add the next step to the Aliquot Sequence you have to fetch the sequence from factordb.

I can see that as true. A sequence isn't advanced until queried. And, I might be quite wrong, but doesn't your script, (at least the perl one I think is from you), upload factors found and then query the db for the next composite that is blocking the sequence? If this is true, is the composite not appearing in the composite listing at that time? If so, you have initiated a factorization locally and added the composite to the db until such time as it is factored. If an "elf" picks the composite up before it is finished by your local machine, there is no way of knowing it has been factored until the elf's factors are submitted.

And, I must admit that I don't know what is being run for yafu@home scripts. I am only referencing the ali.pl script which I am a little more familiar with and I would think others are using.

In conclusion, though, based on the above, the db would not advance a sequence without intervention and therefore the only option would be offline work, which would present a different challenge.

This isn't a huge problem and I hope I didn't make it seem larger than it is. I'll go back to my own yard...

Thanks for the reply and for some great perl scripts.