Isn't a new M. prime overdue?

[science_man_88]

Originally Posted by petrw1

I agree....

or ....I recall seeing a similar suggestion a while back; that is to give us the option on this report to exclude NF. The problem with that suggestion though is I believe the report is built at the start of each hour. I suppose an option then would be to create 2 canned reports: 1 with NF and 1 without.

On that topic I'd also ask that since it is run once per hour I'd like to see all results since the last report; knowing some could be much more than 1,000 if there was a bulk of manual assignments turned in.

AND on the topic of that topic I'd like to suggest that the Mods/Gerbils/etc create a Thread under Software with a clear topic like "Suggestion Box" where all suggestions could be collected in 1 place so that George et al would only have to look in this 1 place for suggestions.

if it's html why not govern it with a show button as a ul element or with a specific class in a table for that column ?

[retina]

Originally Posted by science_man_88

if it's html why not govern it with a show button as a ul element or with a specific class in a table for that column ?

Anything that needs JS to display hidden things is a bad option IMO. The existing menu system is already completely broken, please don't suggest to make other parts of the pages broken also.

[science_man_88]

Originally Posted by retina

Anything that needs JS to display everything is a bad option IMO. The existing menu system is already completely broken, please don't suggest to make other parts of the pages broken also.

css can help not using JS

[davar55]

For the rest of it, we still expect 1.3 more primes, because (as far as we know, anyway) future Mersenne primes are not at all determined by past ones, ...

My theory is that this assumption is false, they're not randomly
distributed, it only appears so because of the law of small numbers
( :) ).

I think we'll see a bunch, then one big gap.
Based on my YJ-Conjecture and my low number data.

[axn]

Originally Posted by davar55

I think we'll see a bunch, then one big gap.

Which is how a Poisson distribution would look.

[R.D. Silverman]

Originally Posted by axn

Which is how a Poisson distribution would look.

Sometimes

[davieddy]

Originally Posted by R.D. Silverman

Sometimes

No, always!

Say a gap is long if M(n+1)>1.47^2*M(n)
(Probability 1/e^2 ~ 1/7)

Between each long gap there is a run of "not long" gaps.
(average runlength 6)
David

[davieddy]

Originally Posted by davar55

My theory is that this assumption is false, they're not randomly
distributed, it only appears so because of the law of small numbers
( :) ).

I think we'll see a bunch, then one big gap.
Based on my YJ-Conjecture and my low number data.

"Your YJ conjecture" says the expected ratio M(n+1)/M(n) is ~1.5.

We agree, and refer to it as the Wagstaff Conjecture.

The reason the expected computing needed to find M(n+1) is
1.5^3 times that for M(n) is that each test takes 1.5^2 more,
and we expect to have to do 1.5 more of them before finding
a prime.

David

[davieddy]

Originally Posted by davieddy

And 2^3 ~ 1.5^3

David

Originally Posted by davar55

In what sense are two different real numbers with such a large
relative difference approximately equal?

In the sense that log(2^3) = 1.7*log(1.5^3)

(But I was half joking anyway)

David

[davieddy]

I think some folk need to revisit this chestnut:

If buses turn up at random, but every ten minutes on average,
your expected waiting time is ten minutes.

If no bus has turned up after an hour, the expected waiting time
is still ten minutes.

As WBLipp will confirm, in practice, bus services can get worse than this!

David

[ixfd64]

Gotta love exponential distribution.