elective-stereophonic
elective-stereophonic
Multibranch forging approach singapore
Please login or register.

Login with username, password and session length
Advanced search  

News:

Latest Nxt Client: Nxt 1.11.15

Pages: 1 [2]  All

Author Topic: Multibranch forging approach  (Read 12084 times)

mthcl

  • Hero Member
  • *****
  • Karma: +96/-8
  • Offline Offline
  • Posts: 562
    • View Profile
Re: Multibranch forging approach
« Reply #20 on: November 30, 2014, 01:52:24 pm »

Yes, actually there could be applied different measures on branches to choose the best.

BTW, I've heard that cryptocoins with cumulative difficulty measured as SUM(1/baseTargeti2) are more stable (consensus-wise) than ones with SUM(1/baseTargeti) and SUM(1/baseTargeti3). Just a rumor, but still...
I'm almost sure this rumor is wrong, at least for the current realization of the BaseTarget changing algorithm. The reason is that the (normalized) BaseTarget can approach 0 rather easily. Then, for i when this happens, the term 1/baseTargeti2 would be so large, that it over-weights all other terms. That is, a "bad" alternative chain would still have a decent chance to win over the "good" one.

By the way, the current algo has the same deficiency, although to less extent. Imposing a reasonable lower limit on BaseTarget would solve this issue.
« Last Edit: November 30, 2014, 02:28:21 pm by mthcl »
Logged

benjyz

  • Hero Member
  • *****
  • Karma: +71/-4
  • Offline Offline
  • Posts: 508
    • View Profile
Re: Multibranch forging approach
« Reply #21 on: November 30, 2014, 04:36:19 pm »

I think there is a lot of confusion around N@S. Obviously it can't be that easy to hack Nxt, so where is the thinking going wrong? There is no expense for choosing the wrong chain, but what would be the benefit? Why should anyone forge on a malicious chain?
« Last Edit: November 30, 2014, 04:45:14 pm by benjyz »
Logged

Come-from-Beyond

  • Hero Member
  • *****
  • Karma: +794/-671
  • Offline Offline
  • Posts: 4013
    • View Profile
Re: Multibranch forging approach
« Reply #22 on: November 30, 2014, 04:42:54 pm »

I'm almost sure this rumor is wrong, at least for the current realization of the BaseTarget changing algorithm. The reason is that the (normalized) BaseTarget can approach 0 rather easily. Then, for i when this happens, the term 1/baseTargeti2 would be so large, that it over-weights all other terms. That is, a "bad" alternative chain would still have a decent chance to win over the "good" one.

By the way, the current algo has the same deficiency, although to less extent. Imposing a reasonable lower limit on BaseTarget would solve this issue.

Shouldn't we get rid of this deficiency by using 1/SQRT(baseTarget) instead of 1/baseTarget then? If the answer is positive then we could "inductively" come to 1/baseTarget0 (aka 1/1). :)
Logged

mthcl

  • Hero Member
  • *****
  • Karma: +96/-8
  • Offline Offline
  • Posts: 562
    • View Profile
Re: Multibranch forging approach
« Reply #23 on: November 30, 2014, 05:57:40 pm »

I'm almost sure this rumor is wrong, at least for the current realization of the BaseTarget changing algorithm. The reason is that the (normalized) BaseTarget can approach 0 rather easily. Then, for i when this happens, the term 1/baseTargeti2 would be so large, that it over-weights all other terms. That is, a "bad" alternative chain would still have a decent chance to win over the "good" one.

By the way, the current algo has the same deficiency, although to less extent. Imposing a reasonable lower limit on BaseTarget would solve this issue.

Shouldn't we get rid of this deficiency by using 1/SQRT(baseTarget) instead of 1/baseTarget then? If the answer is positive then we could "inductively" come to 1/baseTarget0 (aka 1/1). :)
Well, if one starts using 1/baseTargeta with a<1, then you decrease this effect, but, in compensation, the expected values of 1/baseTargeta would be closer for "good" and "bad" branches, which makes them more difficult to distinguish (as you pointed out, in the limit a -> 0 they will become indistinguishable  :) ). Yes, one should be able to optimize in a, but why doing that if one can get rid of that (former) effect for good, just by introducing a reasonable lower limit for the baseTarget?

EDIT: this "optimization in a" would not work well, since the "optimal" value of a will depend on other parameters (the stake of the "bad guy", ...) and so there is no "universal" optimal value for a. Well, one more reason to introduce a lower limit for the baseTarget...
« Last Edit: December 01, 2014, 01:42:29 pm by mthcl »
Logged

benjyz

  • Hero Member
  • *****
  • Karma: +71/-4
  • Offline Offline
  • Posts: 508
    • View Profile
Re: Multibranch forging approach
« Reply #24 on: November 30, 2014, 06:10:59 pm »

I think arbitrary values should be in general kept to the absolute minimum, not only in crypto, but especially cryptocurrency because of dynamic aspects.

Quote
In cryptography, nothing up my sleeve numbers are any numbers which, by their construction, are above suspicion of hidden properties.

http://en.wikipedia.org/wiki/Nothing_up_my_sleeve_number
Logged

Bill White

  • Newbie
  • *
  • Karma: +14/-0
  • Offline Offline
  • Posts: 10
    • View Profile
Re: Multibranch forging approach
« Reply #25 on: December 20, 2014, 12:39:19 pm »

Coq sources have been published! link removed

Thanks for doing and publishing these Coq formulations.

I'm having a little problem with POStructures.v:

coqc POStructures.v
...
File "./POStructures.v", line 5, characters 0-25:
Error: Cannot find library FixedList in loadpath

I didn't find FixedList in the Coq standard library. Do you know where I can find it?
Logged

andruiman

  • Jr. Member
  • **
  • Karma: +34/-1
  • Offline Offline
  • Posts: 37
    • View Profile
Re: Multibranch forging approach
« Reply #26 on: December 20, 2014, 10:33:15 pm »

Coq sources have been published! link removed

Thanks for doing and publishing these Coq formulations.

I'm having a little problem with POStructures.v:

coqc POStructures.v
...
File "./POStructures.v", line 5, characters 0-25:
Error: Cannot find library FixedList in loadpath

I didn't find FixedList in the Coq standard library. Do you know where I can find it?

Sorry, actually this dependence is unnecessary now. Updated POStructures.v is committed to the repository.
Also I have temporary removed the POSimul5.v as it is a little incompatible with the current version of POStructures.
I will add it in a couple of days.
Logged

Bill White

  • Newbie
  • *
  • Karma: +14/-0
  • Offline Offline
  • Posts: 10
    • View Profile
Re: Multibranch forging approach
« Reply #27 on: December 21, 2014, 01:21:45 pm »

Coq sources have been published! link removed

Thanks for doing and publishing these Coq formulations.

I'm having a little problem with POStructures.v:

coqc POStructures.v
...
File "./POStructures.v", line 5, characters 0-25:
Error: Cannot find library FixedList in loadpath

I didn't find FixedList in the Coq standard library. Do you know where I can find it?

Sorry, actually this dependence is unnecessary now. Updated POStructures.v is committed to the repository.
Also I have temporary removed the POSimul5.v as it is a little incompatible with the current version of POStructures.
I will add it in a couple of days.

It works fine now. I should have thought of trying to simply remove the line! Thanks.

I also use Coq, by the way. It's good to see other people in the cryptocurrency world using it.

I'm still studying your code and papers. Hopefully I will be able to give you guys some feedback.
Logged

coinomat

  • Hero Member
  • *****
  • Karma: +214/-18
  • Offline Offline
  • Posts: 1520
    • View Profile
Re: Multibranch forging approach
« Reply #28 on: December 24, 2014, 02:52:10 pm »

A bit of off-topic, what do you think about application of FLP result to POS blockchains? Somehow intuitively you feel that somehow it cannot apply here, but on the other hand this is a quite general result...
Logged
Time to go further

kushti

  • Board Moderator
  • Sr. Member
  • ****
  • Karma: +184/-5
  • Offline Offline
  • Posts: 384
  • Nxt Core & Apps Dev
    • View Profile
Re: Multibranch forging approach
« Reply #29 on: December 26, 2014, 05:16:21 pm »

A bit of off-topic, what do you think about application of FLP result to POS blockchains? Somehow intuitively you feel that somehow it cannot apply here, but on the other hand this is a quite general result...

I think blockchain passes by BGP/CAP/FLP by reformulating consensus into the weaker form. But we're getting all those problems back by trying to avoid blockchain usage for e.g. performance's sake.
Logged
for donations / messages: NXT-PKXM-WH25-UXXG-CJAVD (alias: kushti)
Pages: 1 [2]  All
 

elective-stereophonic
elective-stereophonic
assembly
assembly