[ANN] Catcoin - 0.9.1.1 - Old thread. Locked. Please use 0.9.2 thread.

[etblvu1]

Update on Difficulty Adjustment Algorithm Simulator

Updated release dramatically reduces memory footprint, and can run the simulation for much longer. It now does "sweet spot plotting" which consists of thin blue lines showing +/- 20% of the level at which the hashrate exactly corresponds to the difficulty level. The difficulty level and hashrate cannot exactly match consistently, because there are natural random variation in how often blocks are found.

To give you an idea of what the simulation program is producing, here is the SMA36LIM12 (current incumbent algorithm), plotted against time, with consistent hashrate:



Notice the distance between the two thin lines (the +20% and -20% points), which are above and below the thick blue line (the point at which difficulty matches hashrate), and how far above and below the plot lines go in relation to those.

By contrast, here is an experimental Penta SMA algorithm, for comparison purposes:



Notice that although there is still fluctuation, they tend to be much closer to staying within the two blue lines. The Penta SMA is only experimental and only provided for illustration purposes, and may be tweaked some more over time. It is still not good enough, because it still falls outside the blue lines too much.

The simulations for Kimoto Gravity Well, Phoenixcoin's dual SMA with dampening, and Envy2010's single line modification, are not yet in the simulation, and they may provide even better performance. We will seek out and incorporate as many algorithms as are out there, that will provide us with the richest menu of possibilities, and with community participation in running the simulations and becoming familiar with the unique characteristics of each algorithm, we will have a great chance of achieving consensus or even unanimous agreement.

We want to find an algorithm that stays within the blue lines as much as possible. You can see the simulation play out for yourself by downloading the Windows simulator, available for download now at http://www.catcoin-algo.org.

Thank you for your time.

Etblvu1

[1714644312]

1714644312

[1714644312]

1714644312

[1714644312]

1714644312

[1714644312]

1714644312

[chromosoma]

Positive cats movement on Cryptsy:))
Lovely

[sturle]

I haven't read all of this thread, but I would like to add my 2 cents for the difficulty adjust algorithm.

I had a similar problem in my first algorithm for adjusting the temperature of my bathroom floor, which is heated by mining.  (Water cooled GPUs, water passes through tubes inside the floor, it is covered in another thread back in early 2011.)  My first version went like this:

Loop every 60 seconds:
 If (floor temperature > target temperature)
  Dump (foor temperature - target temperature) less heat to the floor
 elsif (floor temperature < target temperature)
  Dump (target temperature - floor temperature) more heat to the floor

The resulting temperature curve looked like the current difficulty curve of catcoin.  It was a sinusodial curve between 2 degrees over and 2 degrees under the target temperature.

The reason why this worked to badly is the fact that it takes some time for the heat to propagate from the inside of the tubes to the top of the floor where my temperature sensor is located.  When I started reducing heat output, the temperature was already at target, and it would continue to increase for a long time.  When it started to decrease again, the floor was 2 degrees above target, and the core temperature of the floor was actually below surface temperature when the surface temperature crossed my target on the way down.  I tried all kinds of dampening adjustments, but none worked.

My solution:
Instead of looking at the current temperature, I started to calculate what the temperature would be a while ahead if the current rate of change continued.  I found a half an hour to be my sweet spot.  My current algorithm will increase or reduce the amount of heat dumped to always try and reach the target temperature in exactly 30 minutes.  If the temperature increases faster, it will dump less heat.  If it increases slower, it will dump more heat.  This works very well.  The surface temperature follows the target temperature nicely without any large changes in heat output.  My target temperature isn't static.  It will wary a bit by room temperature and time of day.  (The remaining heat is dumped in my living room and in a heat exchanger before my hot water tank, btw.)

My suggestion for catcoin:
Based on the blockrate for the last 6 hours (72 blocks), always adjust the difficulty as much as needed to reach 10 minute blocks by average in 6 blocks (one hour) from now.  Never try to make up for higher or lower blockrate in the past, and by following a future target you will get the dampening you want without any static limits.

[ssjohnson1990]

Update: While my wallet claims it is only 1 hour behind, it has only processed block 21908 of 21960. Ran it all night and made progress of exactly one block although it claims to have made 20 hours worth. Extremely aggravating.

[schiavonxv]

The blockchain is in the 22014 block and in the TeamCatcoin is in the 22224.

What's happening?

[ozzymax]

FUBAR THAT'S WHAT. TOOK ME 2 DAYS TO FIGURE OUT HOW TO SYNC THE NEW WALLET AND HAD TO DO IT ON A CLEAN PC, TOOK ANOTHER 2 DAYS FOR MY COINS TO TURN UP FROM CONIUM.POOL AND NOW WAITING FOR SAME COINS TO SHOW UP AT CRYPTSY SENT 13 HOURS AGO NOT EVEN PENDING. WHAT IS THE POINT OF A COIN IF IT CAN'T BASICALLY FUNCTION. IF IT TAKES THIS LONG TO GET A COIN TO AN EXCHANGE THERE IS NO PROFIT IN IT IN MY EYES. IT'S A DEAD FISH IN THE WATER.

[etblvu1]

I haven't read all of this thread, but I would like to add my 2 cents for the difficulty adjust algorithm.

...

My suggestion for catcoin:
Based on the blockrate for the last 6 hours (72 blocks), always adjust the difficulty as much as needed to reach 10 minute blocks by average in 6 blocks (one hour) from now.  Never try to make up for higher or lower blockrate in the past, and by following a future target you will get the dampening you want without any static limits.

I would like to include this as one of the algorithms in the next version of the simulator. Just to make sure I understand, your suggestion is look at the previous 72 blocks instead of the previous 36 blocks?

Also, how would one implement in code "adjust the difficult as much as needed to reach 10 minute blocks by average in 6 blocks"? Keep in mind this is a dynamic environment and what we think we are targeting this block, will not be what we think we will be targeting next block. Are you suggesting something like (amount needed to fully adjust) / 6, to be applied at every block retarget (keeping in mind that the thing we are pointing to will be constantly moving?)

Thank you,

Etblvu1

[SlimePuppy]

FUBAR THAT'S WHAT. TOOK ME 2 DAYS TO FIGURE OUT HOW TO SYNC THE NEW WALLET AND HAD TO DO IT ON A CLEAN PC, TOOK ANOTHER 2 DAYS FOR MY COINS TO TURN UP FROM CONIUM.POOL AND NOW WAITING FOR SAME COINS TO SHOW UP AT CRYPTSY SENT 13 HOURS AGO NOT EVEN PENDING. WHAT IS THE POINT OF A COIN IF IT CAN'T BASICALLY FUNCTION. IF IT TAKES THIS LONG TO GET A COIN TO AN EXCHANGE THERE IS NO PROFIT IN IT IN MY EYES. IT'S A DEAD FISH IN THE WATER.

Thanks for the caps.  Don't confuse cryptsy with any coin - cryptsy is slow regardless.  I've sent and received feathercoin, litecoin, catcoin, bitcoin - cryptsy takes hours on a good day.

Cat, on the other hand, is working very quickly for person to person transfers and payments from coinium.  Within seconds for the transfer.

Feel free to leave your non-synchronized wallet running for a few days so the other cats in your part of the world have a reliable block source.

[envy2010]

I haven't read all of this thread, but I would like to add my 2 cents for the difficulty adjust algorithm.

...

My suggestion for catcoin:
Based on the blockrate for the last 6 hours (72 blocks), always adjust the difficulty as much as needed to reach 10 minute blocks by average in 6 blocks (one hour) from now.  Never try to make up for higher or lower blockrate in the past, and by following a future target you will get the dampening you want without any static limits.

I would like to include this as one of the algorithms in the next version of the simulator. Just to make sure I understand, your suggestion is look at the previous 72 blocks instead of the previous 36 blocks?

Also, how would one implement in code "adjust the difficult as much as needed to reach 10 minute blocks by average in 6 blocks"? Keep in mind this is a dynamic environment and what we think we are targeting this block, will not be what we think we will be targeting next block. Are you suggesting something like (amount needed to fully adjust) / 6, to be applied at every block retarget (keeping in mind that the thing we are pointing to will be constantly moving?)

Thank you,

Etblvu1

This solution is actually a variant of the weighted average (which I've coded and simulated). It turns out to be very similar to the logarithmic weighted average, though its a hyperbolic function (e.g. 1/x) instead of logarithmic (e.g. ln{x}).

You would code that as adjustment = 1+(target_time-actual_time)/target_time*1/6, which returns a factor of 1.000 when the target and actual blocktimes are equal, and a factor of 1.167 (16.7% increase) when the actual time is half as long as the target (1+[10-5]/10*1/6 = 1.16667).

To use the analogy as stated above, you want to look about as far forward as you are looking back. Plan on using 10 or 20 blocks to hit the target, or you will overshoot. When you allow more than 2x change over the averaging period, the system becomes underdamped (i.e. it may still converge, but is easily disturbed by a hashrate change)... this is easily demonstrated in the difficulty simulator.

[SlimePuppy]

The blockchain is in the 22014 block and in the TeamCatcoin is in the 22224.

What's happening?

What's happening is the same thing that's been happening for a number of days - the blockchain explorer is currently down.  We have notified the owner but no response yet.  This is a community-run site and not a resource run by the dev team so it is outside our control.

Make sure you wallet matches the major pools - teamcatcoin is one - and ignore the explorer.

This will all settle out.

[etblvu1]

Interactive Catcoin Difficulty Simulator v0.4 Released!

This is a major release, with many new feature and algorithm improvements.

Envy2010's Log Algorithm with only single line modification from SMA36LIM12 algorithm. This algorithm delivers an impressive 80% + of time and blocks in the sweet spot range, which is a record breaker for any of the algorithms in the simulator right now.

Penta SMA updated with Rapid Downward Difficulty Adjustment technology. You can click 10x three times to simulate a 1000x increase in hash, then 1/10x three times to simulate 1/1000th decrease in hash, and watch how quickly the algorithm brings hashrate into sweet range.

Sweet spot plotting updated with darker graph lines when block ends inside sweet spot.

End of simulation report that shows what percentage of time was spent in the sweet spot, and what percentage of blocks ended by adjusting inside the sweet spot.

Download your copy at http://www.catcoin-algo.org

[Shahrukh]

catcoin stucked on block number 21388 I have updated the daemon too

[sturle]

IMy suggestion for catcoin:
Based on the blockrate for the last 6 hours (72 blocks), always adjust the difficulty as much as needed to reach 10 minute blocks by average in 6 blocks (one hour) from now.  Never try to make up for higher or lower blockrate in the past, and by following a future target you will get the dampening you want without any static limits.

I would like to include this as one of the algorithms in the next version of the simulator. Just to make sure I understand, your suggestion is look at the previous 72 blocks instead of the previous 36 blocks?

It would help for some stability and robustness against random variation, while still adjusting for normal variation during the day, and reduce the efficiency of block time manipulation.

Also, how would one implement in code "adjust the difficult as much as needed to reach 10 minute blocks by average in 6 blocks"? Keep in mind this is a dynamic environment and what we think we are targeting this block, will not be what we think we will be targeting next block. Are you suggesting something like (amount needed to fully adjust) / 6, to be applied at every block retarget (keeping in mind that the thing we are pointing to will be constantly moving?)

Yes, and retarget at every block.  Looking at the previous 72 blocks, calculate the average hashrate, and what the difficulty should be to achieve 10 minute blocks at that hashrate.  Adjust the difficulty by 1/6 of that, to aim for 10 minute blocks an hour ahead.   Miners leaving when difficulty increase, and vice versa, will help to achieve the 10 minute target somewhat faster.  This is difficult to adjust to, but you could by using a shorter window than 72 blocks.  In that case you keep the timewarp attack in mind and make nodes strict about what times they will accept.

[etblvu1]

Yes, and retarget at every block.  Looking at the previous 72 blocks, calculate the average hashrate, and what the difficulty should be to achieve 10 minute blocks at that hashrate.  Adjust the difficulty by 1/6 of that, to aim for 10 minute blocks an hour ahead.   Miners leaving when difficulty increase, and vice versa, will help to achieve the 10 minute target somewhat faster.  This is difficult to adjust to, but you could by using a shorter window than 72 blocks.  In that case you keep the timewarp attack in mind and make nodes strict about what times they will accept.

Hi Sturle,

Can you take a look at the performance characteristics of Envy2010's Log function applied to the current formula, and let me know if you feel your algorithm has a good chance of performing similarly or better? I can implement your algorithm as a simulation option in any case, but I was interested in getting your input on what you thought it might do compared to this one.

Thank you,

Etblvu1

[Ramsey]

Hi

Help me please

I have not received transaction from teamcatcoin.com pool on my coinex and coinedup account

Passed more than 2 days

What could happen?

thanks

[trianglez]

Same with me, does somebody else experince this kind of problems transferring CATs from wallet to every exchange? I tried every one of them and no success. I opened many tickets and nobody can explain what's going on. Is there anyone who can tell if their daemons are old or need change or the problem is in my wallet? When i chech the transactions it says "not found", or NOt YET REDEEMED?
....... WHERE IS THE DEV to explain this mess???

[SlimePuppy]

Same with me, does somebody else experince this kind of problems transferring CATs from wallet to every exchange? I tried every one of them and no success. I opened many tickets and nobody can explain what's going on. Is there anyone who can tell if their daemons are old or need change or the problem is in my wallet? When i chech the transactions it says "not found", or NOt YET REDEEMED?
....... WHERE IS THE DEV to explain this mess???

Did you read any of this thread so you know who the devs are and to see if they're participating in this thread?

What exchanges have you sent transfers to?  Did you send the CATs directly from a pool or from your wallet?  How long did it take to get the coins from the pool to your wallet?

Are you using a wallet linked in the original post of this thread?  Did you update your catcoin.conf with reliable nodes?

Code:

addnode=cat.geekhash.org
addnode=ultimatecoinpool.org
addnode=eu.coinium.org
addnode=us.coinium.org
addnode=teamcatcoin.com

I await your reply.
Andy

[SlimePuppy]

Hi

Help me please

I have not received transaction from teamcatcoin.com pool on my coinex and coinedup account

Passed more than 2 days

What could happen?

thanks

Please do not transfer coins from your mining account directly to an exchange.  Most exchanges advise NOT to do that.

Always transfer money from pool directly into your wallet, then move to an exchange if you want.

I've used cryptsy, coinedup, and btc-e and they're all slow - minimum of 1 hour and up to 6 hours on a good day getting things into crypsy.  There's nothing anyone here in CATland can do about that.

Andy

[trianglez]

Same with me, does somebody else experince this kind of problems transferring CATs from wallet to every exchange? I tried every one of them and no success. I opened many tickets and nobody can explain what's going on. Is there anyone who can tell if their daemons are old or need change or the problem is in my wallet? When i chech the transactions it says "not found", or NOt YET REDEEMED?
....... WHERE IS THE DEV to explain this mess???

Did you read any of this thread so you know who the devs are and to see if they're participating in this thread?

I read it. I thought they maybe read as well.

What exchanges have you sent transfers to?  Did you send the CATs directly from a pool or from your wallet?  How long did it take to get the coins from the pool to your wallet?

I send it to Criptsy and Coinex, they re[lied that this is not theirs fault and cant help, because they cant see the transaction that i made, so...transaction is invalid? Why? I think its because of the blockchain. I send 3 transactions of 2 CATs just to test if they arrive, but no. It was 3 days ago...so... you know, thats a big mess with this coin...i think i will stop mining.
And... i spotted that the transactions from the pool "coinium" arrived faster when there is more hashpower on the net, maybe because of the switchpools... don't know...maybe. But i get them relatively fast...about 5-10 minutes maximum and they are in my wallet.so...

Are you using a wallet linked in the original post of this thread?  Did you update your catcoin.conf with reliable nodes?

Code:

addnode=cat.geekhash.org
addnode=ultimatecoinpool.org
addnode=eu.coinium.org
addnode=us.coinium.org
addnode=teamcatcoin.com

yes, i've updated my wallet 3 times, ive upfated my conf file, i think everything is allright here, the problem is somewhere else...it's not me. I hope you can help with some advice, because my 2000 coins are worthless right now...

I await your reply.
Andy

Thank you. Alex

[sturle]

Can you take a look at the performance characteristics of Envy2010's Log function applied to the current formula, and let me know if you feel your algorithm has a good chance of performing similarly or better? I can implement your algorithm as a simulation option in any case, but I was interested in getting your input on what you thought it might do compared to this one.

I can't run the simulator.  It is Windows only, and it would be a lot of work to reimplement it.

The only Microsoft products in my house are a book and a defunct first-generation chipmodded Xbox which used to run Linux and XBMC.  I did run MS-DOS at some time in my youth, but deleted Windows in 1989 because it was a waste of expensive hard disk space.  I found it much harder to get useful work done in the GUI compared to MS-DOS and text based editors.  Then I discovered Unix.  It was a holy-s**t-experience, and I never went back to MS-DOS since.