Linear algebra reservations, progress and results

[Dubslow]

Originally Posted by Xyzzy

It takes us days, but we start downloading relations as soon as they start coming in. At first it is hard to keep up but things balance out nicely towards the end.

while true; do nice wget --continue --limit-rate=64k --user=rsals_data --password=***** http://escatter11.fullerton.edu/nfs_data/12_226_plus_7_226/12_226_plus_7_226.dat.gz; sleep 3600; done

I've added this to my own script:

url='http://escatter11.fullerton.edu/nfs_data'
name='4051_71_minus1'
user=****
pass=****
slptime=3600
rate=250k
exts='fb ini poly'

for ext in $exts; do
        wget -N --user=$user --password=$pass "$url/$name/$name.$ext"
done

while true; do
        nice wget --continue --limit-rate=$rate --user=$user --password=$pass "$url/$name/$name.dat.gz"
        sleep $slptime
done

I'm curious though, why the nice? Surely a simple (rate limited) TCP download consumes negligible processing power?

[Xyzzy]

Originally Posted by Dubslow

exts='fb ini poly'

Why do you need the .poly file?

I'm curious though, why the nice? Surely a simple (rate limited) TCP download consumes negligible processing power?

That is just a habit we have from way back when CPU resources were tight. It was always the polite thing to do. (And on one server we deal with even today, running a process that way allows us to skirt the process CPU time limitation.)

FWIW, when we have downloaded the .fb and .ini file, we rename them to msieve.fb and worktodo.ini so that the command line to invoke msieve is simpler.

[Dubslow]

Originally Posted by Xyzzy

Why do you need the .poly file?

That is just a habit we have from way back when CPU resources were tight. It was always the polite thing to do. (And on one server we deal with even today, running a process that way allows us to skirt the process CPU time limitation.)

FWIW, when we have downloaded the .fb and .ini file, we rename them to msieve.fb and worktodo.ini so that the command line to invoke msieve is simpler.

Hmm, there's no :shrug:

Honestly, no real reason. I had the .projection too for a while before I realized it's large and a waste.

[Jarod]

LA took 13 hours 51 minutes to complete using either 6 or 8 threads on a I 7 5960 X at 3 GHz.

prp52 factor: 7900083352228908935782292679688341172833928993595953
prp118 factor: 2028402787009378656990445950154672782198265845967931765722769217754896599149319369650857058346177958425237023520001137


Please find log attached

[jyb]

I can take 8_269_minus_7_269 for post-processing.

[unconnected]

C165_842592_8032 completed - 35 hours for 8.3M matrix (TD=140).

prp67 factor: 1050566961306451588674530244517205327913164128373334255482224505743
prp99 factor: 664910913443536911339260986610284630155514114445454204744433715236688283538021166444218857757023173

[swellman]

Reserving C178_149_35.

[Xyzzy]

We are pretty sure we know what the difference is between a "p" and a "prp".

Why does msieve sometimes report a "prp"?

[Xyzzy]

We had an idea a few days ago:

We are willing to download, package and snail-mail files in case someone is in a situation where they want to do LA work but they do not have the resources to download big files.

Mailing a USB key is slow but we think (?) there is no race to get this work done.

Anyways, if anyone wants a deal like this just let us know.

[wombatman]

Originally Posted by Xyzzy

We are pretty sure we know what the difference is between a "p" and a "prp".

Why does msieve sometimes report a "prp"?

I'm pretty sure msieve just doesn't prove primality on the factors. I believe it checks (via Fermat, maybe?) for probable primality and whether the factor divides the input number.

[VBCurtis]

Originally Posted by wombatman

I'm pretty sure msieve just doesn't prove primality on the factors. I believe it checks (via Fermat, maybe?) for probable primality and whether the factor divides the input number.

I am pretty sure this is correct, but that below a certain size prp and p are equivalent in that it has been shown that every prp via the test msieve uses below n digits is p. I don't remember what n is: 25ish?