Loading...
  OR  Zero-K Name:    Password:   

Locust vs aa analyse

6 posts, 419 views
Post comment
Filter:    Player:  
sort
when Locusts brawl vs AA (pure damage vs hp)

define brawl value (dps*hp) (BV) of a cluster of units, whose amont determines win lose directly

define brawl factor square (BF_sq) is that a cluster of units cost C has BV = BF_sq*Cost*Cost

BF_sq = dps/Cost * hp/Cost

define (standard) brawl factor is BF = sqrt(BF_sq)

so BV = (BF*Cost)^2

A unit with Cost/BF_A vs B unit with cost Cost/BF_B draw when brawl, we say brawl ratio of unit A,B is BF_A:BF_B

brawl factor: Locust 1.0082, Gremlin 1.5996, Toad 1.4257, Tarantula 1.1638, Ettin 1.1224, Reaver 2.2406, Razor 2.3877, Stardust 3.7344.

glaive 2.3907, trident 0.8606, raptor 1.1055.

assume brawl ratio Locust vs AA is 1:1.4

Assume locust has overwelming speed, so that:

AA must spread evenly across the front line, which is normally map width.

Locusts attacks and pass thorugh AA range immediately so we dont calculate damage that AA can done before locust attack.

AA range can affect how much AA can attack Locusts.

When rich ( players expanded enough and have frontline):

1 unit is 512 elmo(iirc) ( width x height of map's description).

assume typecal AA range 900

for width 14 map, when locusts attack, only (900*2) / (14*512) ~= 0.25 of AA unit can actually attack locusts.

So overall locust advantage ratio is Locust vs AA 2.85:1. You must cost 2.85x locust value on AA to defent them properly.

This calculation varies depends on map width, the wider the map the more advantage air have.

This calculation didnt take riots in count, which only affects when locust want to attack a regular army.

And the calculation may not be correct.

When expanding:

assume players expands as a circle, they need AA forces to cover then edge of circle.

define expand cost: cost they need to spend on army to protect area / area

more idea is required
+3 / -0
20 days ago
What in the Lobstermatics.
+2 / -0
The overwhelming speed assumption is a poor approximation, particularly the corollary that Locusts don't take significant damage closing with AA units. AA can shoot down Ravens that don't need to stick around to apply their damage, and in modest quantities can do so while the Ravens are inbound. But as a more fundamental problem, if Locusts really were that fast, the perimeter defense scheme you describe would be pointless, because the Locusts would only take trivial amounts of damage for flying past the perimeter into the undefended inner economy.

Your AA suffers from an assumption in the other direction that AA only moves while Locust raids are not underway, yet has efficiency closer to mobile AA than static AA. Use the numbers for Razors in your perimeter defense instead of static Toads and the Locusts' calculated advantage almost disappears. Or maneuver the AA units defensively while you're under attack by Locusts, moving them into range of your anti-ground defenses and thus violating the assumption that riots don't matter.

Finally, you seem to be ignoring AoE, which where present is a practical limit on Locusts' ability to scale up, and the advantage of frontloading burst damage.

Between these various assumptions, you've crippled every single AA option. An Artemis must wait for the Locusts to start hitting it, then pick them off one at a time. Anglers are stuck plinking away at a rate of one missile every three seconds, with a pair of them taking nine seconds' worth of incoming Locust damage before scoring a single kill. Crashers cannot group up under fire, nor can they run for the apathetic support of any nearby Rippers, because riots do not care unless they themselves are under attack. And Razors need to slow down their DPS to match the mobile options.
+3 / -0
I think calculating geometry is important to analyzing this.
You are making quite a few geometrical assumptions that don't apply very often.

quote:
assume typecal AA range 900

for width 14 map, when locusts attack, only (900*2) / (14*512) ~= 0.25 of AA unit can actually attack locusts.

An Ettin that has 900 range running away with 96 speed takes 4.44 seconds before the Locusts with 260 range and 250 speed enters range.
If a supporting Ettin goes in the opposite direction of the fleeing Ettin, it takes about 6 seconds before the supporting Ettin reaches the fleeing Ettin.
This means that the lone Ettin gets 4.4 seconds of free damage and only 1.6 seconds for the Locusts to deal damage before the supporting Ettin reaches combat.

With radar coverage, this should give you even more time to group up your AA before the clash.

quote:
..This calculation varies depends on map width, the wider the map the more advantage air have...

...assume players expands as a circle, they need AA forces to cover then edge of circle.

It is true that larger maps favor faster units over slower units.
However players rarely expand in a circle. On most maps you start next to the edge of the map or even in the corner. With a corner start and radar, it should in most cases be sufficient to cover your initial expansion with 1 group of anti-air. I'd say gunship starts are only viable when there is terrain that prevents land units from moving freely like Mecharavaged or Lonely Oasis, where it also blocks Line of Fire and hides the initial Locusts from most radar.

Another important point is that what you are actually targeting with the Locusts.
On the front line, there might a lot of good flex AA like raiders, riots, defenses and even commanders. Since those typically have high brawl factor square, I see most AA units as a way to extend the range of those strongpoints.
If you have a combination of AA and raiders running around in groups, the brawling strength is the combined brawling strength of the groups while the AA provides the range to prevent Locusts for making cost effective trades against your expansion where the Locusts must have enough time and strength to kill defenses while taking damage from mobile AA and do it before the raiders reach the expansion, and the Locusts have to do it all costeffectively.
+2 / -0

Kiting



unit a vs b

assume a.range > b.range, a.speed < b.speed

range advantage: a.range-b.range

speed advantage: a.speed-b.speed

kiting time advantage (KT): (a.range-b.range) / (b.speed-a.speed)

kiting damage advantage (KD): KT>0: KT*a.dpspc

assume a_cost, b_cost:

when a kite b, first a deal burst damage, then a deal dps for KT time, then b deam burst on a, then a b brawl

b_cost = b_cost_0 - a_cost_0* (a.burstpc + tdadv * a.dpspc) / b.hppc

then a_cost = a_cost_0 - b_cost*b.burstpc / a.hppc

compare a_cost * a.BF = b_cost *b.BF, we get who win.

by going backward, we can calculate braw percentage of a,b.

Asked ai to simplify :) b_cost_0/a_cost_0 = ( KT * a.dpspc + a.burstpc )/b.hppc + 1/ ( b.BF / a.BF + b.burstpc / a.hppc )

Had written some code to calculate.

This may soon become generic unit comparer.

[Spoiler]

Reaver's BF: 2.2406267344797315
Raptor's BF: 1.1055415967851332
Trident's BF: 0.8608781046763305
Bandit's BF: 2.2639984295243867
Glaive's BF: 2.390810810449239
Hacksaw's BF: 1.108011094956061
Scallop's BF: 2.299216490349261
Toad's BF: 1.4245841498486498
Locust's BF: 1.0020639856840274
Gremlin's BF: 1.5957517385580724
Tarantula's BF: 1.1644425790483877
Angler's BF: 1.3033274920143167
Stardust's BF: 3.7344719829897692

Locust vs Reaver overall advantage: 0.4193698094839383
Locust vs Trident overall advantage: 0.5942703442974699
Locust vs Angler overall advantage: 0.41327089064940464
Locust vs Hacksaw overall advantage: 0.32941084485932304
Locust vs Toad overall advantage: 0.5335701178767338
Locust vs Gremlin overall advantage: 0.4840650632010012
Locust vs Tarantula overall advantage: 0.5394871091728761
Locust vs Scallop overall advantage: 0.37622805053289354
Locust vs Stardust overall advantage: 0.2470960327476647

Locust vs Bandit brawl advantage: 0.4426080745535401
Locust vs Glaive brawl advantage: 0.4191314433180672
Locust vs Raptor brawl advantage: 0.906400979029632

Locust vs Angler is not very correct because angler shot 4 per 12 second while calculation assume they deal continuous dps, which overrate Angler.

Stardust heat is not counted.

Aoe are not counted.
+1 / -0


6 days ago
It seems decent.

If you are counting in burst damage from the first free shot you do when the engagement start, I think you should also calculate and factor in overkill DPS.
To get overkill DPS you can just take:
DPS * (math.floor(DPS / HP) / (DPS / HP))

Basically if a unit does 200 burst damage, it will waste 25% of its DPS against a unit with 300 HP.

+0 / -0