[1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool

[Panda Mouse]

Within three months the probability of 400 samples should be at 99.56% (99.56% - 100.44%).
We have 90.6%.
This is an obvious bug.
No luck for so long demonstrates the mathematical inconsistency of assumptions.
I'm sorry, but leave the project.
I hope not for long.

Panda Mouse.

[1714221793]

1714221793

[forrestv]

How exactly did you compute that? That confidence interval is much tinier than is possible... From my simulations this outcome is clearly within the 95% confidence interval.

[DeathAndTaxes]

Within three months the probability of 400 samples should be at 99.56% (99.56% - 100.04%).
We have 90.6%.
This is an obvious bug.
No luck for so long demonstrates the mathematical inconsistency of assumptions.
I'm sorry, but leave the project.
I hope not for long.

Well you are free to leave but your math and thus the assumption based on it is flawed.

[Panda Mouse]

How exactly did you compute that? That confidence interval is much tinier than is possible... From my simulations this outcome is clearly within the 95% confidence interval.

Well:
If there is a large population sample and the distribution of the features did not challenge the validity characteristics of the distribution in the general population with normal distribution (N (m, σ)),
we calculate the sample estimate of standard deviation (s).
Then the confidence interval for standard deviation (σ) in the general population is determined by the formula:

Where:
s - standard deviation of the sample estimate,
uα value of the random variable U (sdandarised normal distribution) (N (0,1))
determined in the same way as for the arithmetic mean.

The result - easy to calculate, is in the range of 99.56-100.44 (we have 90.6).

Panda Mouse

[DeathAndTaxes]

What is your calculation for s ?
Standard Deviation for block length (in hashes) is likely much larger than you think.

[bitpop]

Please use a monte carlo simulation

[Panda Mouse]

What is your calculation for s ?
Standard Deviation for block length (in hashes) is likely much larger than you think.

The standard deviation (s) can be easily calculated from the time of calculation blocks given on p2pool.info (Recent Blocks).
For a finite population is the mean square deviation of the differences between variable values ​​and their arithmetic mean.
The standard deviation can be calculated from the formula:

where xi is another feature value in the population, u is the expected value, N is the number of observations in the population = 400.

Panda Mouse.

[Panda Mouse]

Please use a monte carlo simulation

Not a problem I made this in excel for 400 probe.
The result is exactly the same. Please try or I can send you this sheet.

BTW in last 7 days there was 2 orphaned blocks.
Why they are invisible, they should be crossed out.

Panda Mouse.

[DeathAndTaxes]

BTW in last 7 days there was 2 orphaned blocks.
Why they are invisible, they should be crossed out.

They weren't produced by the main fork of p2pool.

[DeathAndTaxes]

The standard deviation (s) can be easily calculated from the time of calculation blocks given on p2pool.info (Recent Blocks).

You can't use time.  You must use hashes and difficulty.  The avg length of a block (and expected length of a block) depends only on # of hashes and difficulty.  Technically even then there will be minor error when blocks cross over difficulty changes but it should be minimal or those blocks could be excluded.

Expected length = (2^32)(difficulty) in hashes.

To get std deviation you need the actual length of each block in hashes.

[Panda Mouse]

The standard deviation (s) can be easily calculated from the time of calculation blocks given on p2pool.info (Recent Blocks).

You can't use time.  You must use hashes and difficulty.  The avg length of a block (and expected length of a block) depends only on # of hashes and difficulty.

Expected length = (2^32)(difficulty) in hashes.

Not!!!
You are wrong.

This is only a probability and linear distribution.
The level of difficulty has nothing to do with it.
All data, together with the expected value is calculated and placed on the main page.
You just have to know how to read it!

Panda Mouse.

[DeathAndTaxes]

Well you said TIME that confused me.  Your right the length vs expected can be used but that is because it is already adjusted for what matters (hashes observed vs hashes expected).

Not sure where you are getting 400 observations.  p2pool has only found 275 blocks?

Lastly the length of block isn't a normal distribution.  It is an exponential one.

Still even using incorrect 400 observation, incorrect assumption of a normal distribution I have no idea how you determined confidence interval is ~99% to ~101% over 400 blocks.  I even tried to make mistakes and math just doesn't add up.

[echris1]

Did anybody else's graphs at /static/graphs clear recently?

I did a git pull and restarted p2pool, now graphs are just empty.  Been running for over a day now, still all the graphs are empty, and it looks like its not even registering the miners, because the section under Miners is empty.  The old ones (rrdtool) still work fine, and the info at /static looks fine as well, but all the graphs are blank.

[DeathAndTaxes]

I haven't done a gitpull but using latest released version my graphs didn't clear.

[forrestv]

That was a bug; if you pull again (as of a couple of hours ago) it will work again. No data was lost, just the graph displaying was broken.

[Krak]

Did anybody else's graphs at /static/graphs clear recently?

I did a git pull and restarted p2pool, now graphs are just empty.  Been running for over a day now, still all the graphs are empty, and it looks like its not even registering the miners, because the section under Miners is empty.  The old ones (rrdtool) still work fine, and the info at /static looks fine as well, but all the graphs are blank.

I'm getting the same thing after doing a pull last night.

[sharky112065]

Did anybody else's graphs at /static/graphs clear recently?

I did a git pull and restarted p2pool, now graphs are just empty.  Been running for over a day now, still all the graphs are empty, and it looks like its not even registering the miners, because the section under Miners is empty.  The old ones (rrdtool) still work fine, and the info at /static looks fine as well, but all the graphs are blank.

I'm getting the same thing after doing a pull last night.

Yes, there was a bug. It effected me as well. He fixed it this morning. Do another git pull and restart p2pool. The latest pull will fix the issue.

[twmz]

BTW in last 7 days there was 2 orphaned blocks.
Why they are invisible, they should be crossed out.

They weren't produced by the main fork of p2pool.

There actually were 2 orphaned blocks made by the main fork of p2pool that were missing (because my node bitcoin node never heard about them).  I added them to p2pool.info manually.  These were in addition to the other orphaned blocks from the old p2pool fork.

[Raize]

Is there a way for the users of my node to see how much BTC they have been and are about to be paid?

Also, does anyone know if Guiminer works with P2Pool or not?

[Serenata]

Holy crap @ the day we're having.

Shh...
Don't scare the luck away. 

Forget I ever said anything.

Someone did scare the luck away  
287,7% and 300,0% for today's block
123.3% and 149.9% for previous blocks