A simple test with USEM

[rosutosefi]

If USEM faced a farm, how fast would it win? Yeah, useless facts...

I ran some tests on trainer. But after 5 minuets, I gave up. There has to be another way.
Then I tried programming to simulate games. As a newbie programmer (started learning it 2 months ago), this would be a challenge.

Spoiler for The extremely fat code:

clear
clc
matches = input('Number of games played: ');
addedttw = 0;
ttwtable = zeros(1,16);
for match = 1:matches
weapon = 0;
attack = 0;
automulligan = 0;
quanta = 0;
ehp = 100;
tplay = 1;
turn = 0;
th = 0;
%by light_sefi
while automulligan~=2 && th == 0
    tow = 10;
    swd = 2;
    fro = 6;
    coc = 6;
    hea = 6;
    sh = 0;
    fh = 0;
    ch = 0;
    hh = 0;
    for draw=1:7
        deck = tow+swd+fro+coc+hea;
        if rand(1)*deck<tow && tow ~= 0
            tow = tow-1;
            th = th+1;
        elseif rand(1)*deck<tow+swd && swd ~= 0
            swd = swd-1;
            sh = sh+1;
        elseif rand(1)*deck<tow+swd+fro && fro ~= 0
            fro = fro-1;
            fh = fh+1;
        elseif rand(1)*deck<tow+swd+fro+coc && coc ~= 0
            coc = coc-1;
            ch = ch+1;
        elseif rand(1)*deck<=tow+swd+fro+coc+hea && hea ~=0
            hea = hea-1;
            hh = hh+1;
        end
    end
automulligan = automulligan+1;
%by light_sefi
end
    if rand(1)<0.5
        deck = tow+swd+fro+coc+hea;
        if rand(1)*deck<tow && tow ~= 0
            tow = tow-1;
            th = th+1;
        elseif rand(1)*deck<tow+swd && swd ~= 0
            swd = swd-1;
            sh = sh+1;
        elseif rand(1)*deck<tow+swd+fro && fro ~= 0
            fro = fro-1;
            fh = fh+1;
        elseif rand(1)*deck<tow+swd+fro+coc && coc ~= 0
            coc = coc-1;
            ch = ch+1;
        elseif rand(1)*deck<=tow+swd+fro+coc+hea && hea ~=0
            hea = hea-1;
            hh = hh+1;
        end
    end
    while ehp > 0 %the game
        while th ~=0
            th = th-1;
            tplay = tplay + 1;
            quanta = quanta + 1;
        end
        while quanta > 0 && sh > 0 && weapon < 1
            sh = sh-1;
            weapon = 1;
            attack = attack + 6;
            quanta = quanta - 1;
        end
        while quanta > 1 && fh > 0
            fh = fh-1;
            attack = attack + 5;
            quanta = quanta - 2;
        end
        while quanta > 2 && ch > 0
            ch = ch-1;
            attack = attack + 5;
            quanta = quanta - 3;
        end
        %endturn
        turn = turn+1;
        if sh+fh+ch+hh == 8
            if sh == 1 && weapon == 1
                sh = sh-1;
            elseif hh > 0
                hh = hh-1;
            elseif ch > 0
                ch = ch-1;
            elseif fh > 0
                fh = fh-1;
            end
        end
        ehp = ehp-attack;
        quanta = quanta + tplay;
        deck = tow+swd+fro+coc+hea;
        if rand(1)*deck<tow && tow ~= 0
            tow = tow-1;
            th = th+1;
        elseif rand(1)*deck<tow+swd && swd ~= 0
            swd = swd-1;
            sh = sh+1;
        elseif rand(1)*deck<tow+swd+fro && fro ~= 0
            fro = fro-1;
            fh = fh+1;
        elseif rand(1)*deck<tow+swd+fro+coc && coc ~= 0
            coc = coc-1;
            ch = ch+1;
        elseif rand(1)*deck<=tow+swd+fro+coc+hea && hea ~=0
            hea = hea-1;
            hh = hh+1;
        end
    end
ttwtable(turn) = ttwtable(turn) + 1;
end
%by light_sefi
for add = 1:16
    addedttw =  addedttw + add*ttwtable(add);
    if ttwtable(add)~=0
    fprintf('%d turns: %d (%g%%)\n', add, ttwtable(add), ttwtable(add)*100/matches)
    end
end
averagettw = addedttw/matches;
fprintf('Average turns to win: %g\n', averagettw)
And the results!

USEMulator Results

Spoiler for Hidden:

Test #1
Number of games played: 130000000
4 turns: 31776 (0.0244431%)
5 turns: 79881767 (61.4475%)
6 turns: 45115511 (34.7042%)
7 turns: 4203277 (3.23329%)
8 turns: 629129 (0.483945%)
9 turns: 117379 (0.0902915%)
10 turns: 18767 (0.0144362%)
11 turns: 2104 (0.00161846%)
12 turns: 270 (0.000207692%)
13 turns: 20 (0.0000153846%)
Average turns to win: 5.43043

Test #2
Number of games played: 100000000
4 turns: 24292 (0.024292%)
5 turns: 61450018 (61.450018%)
6 turns: 34702273 (34.702273%)
7 turns: 3233596 (3.233596%)
8 turns: 483087 (0.483087%)
9 turns: 90585 (0.090585%)
10 turns: 14327 (0.014327%)
11 turns: 1608 (0.001608%)
12 turns: 200 (0.0002%)
13 turns: 14 (0.000014%)
Average turns to win: 5.430396

Test #3
Number of games played: 10000000
4 turns: 2427 (0.02427%)
5 turns: 6147018 (61.47018%)
6 turns: 3469132 (34.69132%)
7 turns: 321968 (3.21968%)
8 turns: 48664 (0.48664%)
9 turns: 9133 (0.09133%)
10 turns: 1470 (0.01470%)
11 turns: 160 (0.0016%)
12 turns: 24 (0.00024%)
13 turns: 4 (0.00004%)
Average turns to win: 5.430167

Test #4
Number of games played: 10000000
4 turns: 2365 (0.02365%)
5 turns: 6146981 (61.46981%)
6 turns: 3468637 (34.68637%)
7 turns: 322742   (3.22742%)
8 turns: 48451 (0.48451%)
9 turns: 9147 (0.09147%)
10 turns: 1490 (0.01490%)
11 turns: 156 (0.00156%)
12 turns: 29 (0.00029%)
13 turns: 2 (0.00002%)
Average turns to win: 5.43023

Test #5
Number of games played: 10000000
4 turns: 2427 (0.02427%)
5 turns: 6142749 (61.4275%)
6 turns: 3472992 (34.7299%)
7 turns: 322989 (3.22989%)
8 turns: 48158 (0.48158%)
9 turns: 9073 (0.09073%)
10 turns: 1427 (0.01427%)
11 turns: 162 (0.00162%)
12 turns: 20 (0.0002%)
13 turns: 3 (0.00003%)
Average turns to win: 5.43056

Test %6
Number of games played: 10000000
4 turns: 2325 (0.02325%)
5 turns: 6146605 (61.4661%)
6 turns: 3468873 (34.6887%)
7 turns: 322883 (3.22883%)
8 turns: 48559 (0.48559%)
9 turns: 9136 (0.09136%)
10 turns: 1417 (0.01417%)
11 turns: 172 (0.00172%)
12 turns: 27 (0.00027%)
13 turns: 2 (0.00002%)
14 turns: 1 (0.00001%)
Average turns to win: 5.43029

Test #7
Number of games played: 10000000
4 turns: 2432 (0.02432%)
5 turns: 6145233 (61.4523%)
6 turns: 3470169 (34.7017%)
7 turns: 323089 (3.23089%)
8 turns: 48323 (0.48323%)
9 turns: 9132 (0.09132%)
10 turns: 1438 (0.01438%)
11 turns: 151 (0.00151%)
12 turns: 30 (0.0003%)
13 turns: 3 (0.00003%)
Average turns to win: 5.43037

Test #8
Number of games played: 10000000
4 turns: 2355 (0.02355%)
5 turns: 6144181 (61.4418%)
6 turns: 3471146 (34.7115%)
7 turns: 323536 (3.23536%)
8 turns: 48328 (0.48328%)
9 turns: 8963 (0.08963%)
10 turns: 1336 (0.01336%)
11 turns: 135 (0.00135%)
12 turns: 19 (0.00019%)
13 turns: 1 (0.00001%)
Average turns to win: 5.43043

Test #9
Number of games played: 10000000
4 turns: 2459 (0.02459%)
5 turns: 6146103 (61.461%)
6 turns: 3469458 (34.6946%)
7 turns: 323147 (3.23147%)
8 turns: 48156 (0.48156%)
9 turns: 9102 (0.09102%)
10 turns: 1402 (0.01402%)
11 turns: 156 (0.00156%)
12 turns: 17 (0.00017%)
Average turns to win: 5.43022

Test #10
Number of games played: 200000000
4 turns: 48594 (0.024297%)
5 turns: 122896525 (61.4483%)
6 turns: 69410180 (34.7051%)
7 turns: 6464142 (3.23207%)
8 turns: 967406 (0.483703%)
9 turns: 180806 (0.090403%)
10 turns: 28679 (0.0143395%)
11 turns: 3199 (0.0015995%)
12 turns: 436 (0.000218%)
13 turns: 32 (0.000016)
14 turns: 1 (0.0000005%)
Average turns to win: 5.43041

TTW of 5.430 and 0.024% chance to get a 4-turn win. It also got two 14-turn wins out of 500000000 games.
Interesting, but useless.

And yes, it was played at maximum efficiency. AI for USEM is very easy to program. 

That's what happens when I get bored with no internet  thanks to the typhoon.

[majofa]

That's pretty cool.

I have an idea for the next time your internet goes out... What is the fastest mono deck for each element.

Requirements:
All cards must be of the specific element. (No other cards or off-element marks)
Imagine this deck facing an opponent who only has pillars in his deck. (So you can totally maximize on speed, plus you don't have extra programming)

I'd start with upped versions of the deck.

[Xenocidius]

That's pretty cool.

I have an idea for the next time your internet goes out... What is the fastest mono deck for each element.

Requirements:
All cards must be of the specific element. (No other cards or off-element marks)
Imagine this deck facing an opponent who only has pillars in his deck. (So you can totally maximize on speed, plus you don't have extra programming)

I'd start with upped versions of the deck.

Could be quite hard, some may require more complex AI.

[rosutosefi]

I'll try programming simple decks when I get bored again. 

[The_Mormegil]

I imagine the following:
  will be simple, as they probably aren't that different from USEM. Life might be a bit more trouble if you want it to use Mitosis, but Adrenarush is probably faster. Darkness has Eclipse, and Death has Arsenic, but they aren't too difficult to code.
  will be slightly worse, because for air, you probably need to consider Fireflies + UGs and the very least Sky Blitz; for gravity there's Acceleration and Catapult; for time there's extra draws to count (precog, probably hourglasses).
  will be a lot more difficult due to Cremations, Cremation targets, Lava Golem's growth from excess Cremation quanta, if any, and phoenixes' rebirth.
And I think the best way to rush with Entropy (although I might be off) involves Improved Mutation... Or maybe just abomidragon spam?

[Djhopper :)]

And I think the best way to rush with Entropy (although I might be off) involves Improved Mutation... Or maybe just abomidragon spam?

For pure rush, you want abominations, discords and dragons. (as long as your cards don't HAVE to be upped (abominations) )

[rosutosefi]

Light won't be too hard. I'll try a blessed pegasi deck when I have time to do it. Shriekers will be next. I'll start looking for decks and save them so that I can just pull one and test when I feel bored. 

[bored_ninja777]

you have matlab? i used that in my AI class in college.. that is strange IDE to program in.. really no rules..

[jmdt]

Matlab...I approve

[bored_ninja777]

can you get it free somewhere? my professor made it seem like you have to pay a lot for it.. and its a very powerful tool :)used it in my class to help diagnose breast cancer patients based on raw data and had to improve it.. all the crazy math involved O_o

[jmdt]

can you get it free somewhere? my professor made it seem like you have to pay a lot for it.. and its a very powerful tool :)used it in my class to help diagnose breast cancer patients based on raw data and had to improve it.. all the crazy math involved O_o

Free...no chance.

[bored_ninja777]

can you get it free somewhere? my professor made it seem like you have to pay a lot for it.. and its a very powerful tool :)used it in my class to help diagnose breast cancer patients based on raw data and had to improve it.. all the crazy math involved O_o

Free...no chance.

didnt think so.