LWMA Failures about difficulty-algorithms HOT 3 OPEN

I found your email in my spam folder and sent the discord invite. N=30 varies too much. It will attract on-off mining. BTG has good code for BTC coins

from difficulty-algorithms.

Hello. I am looking for a simple LWMA n=30 implementation that is compatible with bitcoin 0.15.99 source. I have tried emailing you and am also available via Discord in the Pigeoncoin server. Would love to discuss this with you before implementing it from scratch.

from difficulty-algorithms.

As the POW mining situation changes, 3 more coins have seen unfriendly oscillations in LWMA caused by on-off mining. Mainly Masari who previously had the best long-term success with LWMA. Wownero and Intense have also been seeing some oscillations. BTC had by far the worst problem, and they came up with what is so far working great. There is only 6 days of data.

Here it is before and after. First on a big scale to show how big the peaks were, then on my normal scale so that it does not appear overly smooth compared to other LWMAs,

Big scale:

Normal Scaling:

They lowered they N to 45 and used the following logic::

if ( avg_10_ST < T/4 & next_D < 2*avg_10_D )
   { next_D = 2*avg_10_D; }

else if (avg_10_ST < T/2 & next_D < 1.5*avg_10_D )
    { next_D = 1.5*avg_10_D; }

if ( next_D < 0.77*prev_D )
    { next_D = 0.77*prev_D; }

I will investigate trying to optimize this because I don't think it will help Masari's nearly-awful oscillations and the less-bad oscillation situations. Time will show if on-off miners find a different oscillation pattern to get more out of the above.

Here is a link to their code which is done on targets instead of difficulty.
https://github.com/bitcoincandyofficial/bitcoincandy/blob/e92e12c1b5fb9df46e2cd2f29403c7bd44e7ec35/src/pow.cpp#L113

from difficulty-algorithms.