Predict M48...

[lfm]

Originally Posted by mdettweiler

Yes, that function has been around for quite a while actually (I've experienced it myself if I get a lot of dust buildup). But, even with that, the computer still produces a bunch of heat--never mind whether it's actually overheating. Thus, much of the "summer slowdown" is probably due to people actually shutting down their machines so they produce less heat and tax their A/C less. I've definitely seen this at other projects, and would expect it to carry over to GIMPS as well.

Ok, makes sense if they're ACing them. (We don't have much AC in homes around here). I heard the latest thing for data centers is to do em without AC.

[Uncwilly]

Originally Posted by mdettweiler

Thus, much of the "summer slowdown" is probably due to people actually shutting down their machines.

And school participation likely drops when school is out.

Originally Posted by lfm

Ok, makes sense if they're ACing them.

Or wish to help reduce peak power demand. Power generated to supply the peak load is the most expensive and often dirtier.

[lycorn]

Originally Posted by petrw1

On another thread I have been reporting on the "classic" stats report over the last 5 years ro see if we are doing more or less LL/DC each year.

That sounds interesting. Which thread is it?

[Uncwilly]

:cough:

[davieddy]

Originally Posted by Uncwilly

:cough:

http://www.youtube.com/watch?v=5Y3yBhX0dhc

If you're looking for trouble you've come to the right place

[petrw1]

Originally Posted by lycorn

That sounds interesting. Which thread is it?

Terribly, terribly sorry for missing this one ... again and again.

http://www.mersenneforum.org/showthread.php?t=10853
Post #6 is my last report.

[Uncwilly]

It has been over a quarter since the last post in this thread.
Someone recently noted that we have had a long run without a prime.

[CRGreathouse]

Originally Posted by Uncwilly

Someone recently noted that we have had a long run without a prime.

Is that true? In the last 10 years (3652 days) we have discovered 9 Mersenne primes. It has been 367 days since the last Mersenne prime was discovered. A Poisson distribution with lambda = 9 * 367/3652 gives a 40.5% chance of discovering 0 and a 36.6% chance of discovering 1 in the time since the last discovery. We seem to be doing just about as expected.

Note 1: Using the Poisson distribution assumes that the increase in computer speed/number of cores/number of participants neatly offsets the increase in the size of the numbers tested. For a first approximation, this seems fair.
Note 2: There has been some question as to whether the last, say, half-dozen Mersenne primes are closer than expected. If so, then we should expect longer, making the current run all the more likely.

[Brian-E]

Originally Posted by CRGreathouse

Note 2: There has been some question as to whether the last, say, half-dozen Mersenne primes are closer than expected. If so, then we should expect longer, making the current run all the more likely.

I don't understand this statement (since you're obviously not someone who is likely to fall for the gambler's fallacy). So why would the recent suggested cluster of Mersenne primes make the last year's lack more likely?

[axn]

Originally Posted by Brian-E

I don't understand this statement (since you're obviously not someone who is likely to fall for the gambler's fallacy). So why would the recent suggested cluster of Mersenne primes make the last year's lack more likely?

Well, the average of 3652/9 days-per-find might not be the correct figure to base our calculations, if we assume that the latest bunch of close finds are anomalous. Then, we could throw them out and recalculate our averages/expectations (which would be longer than what he originally calculated)

[cheesehead]

Originally Posted by Brian-E

gambler's fallacy

I am amused to see that the automated simulation of coin tosses at the side of that article repeatedly shows a blue:red ratio that is closer to 55/45 than 50/50.