• Log InLog In
  • Register
Liquid`
Team Liquid Liquipedia
EDT 19:48
CEST 01:48
KST 08:48
  • Home
  • Forum
  • Calendar
  • Streams
  • Liquipedia
  • Features
  • Store
  • EPT
  • TL+
  • StarCraft 2
  • Brood War
  • Smash
  • Heroes
  • Counter-Strike
  • Overwatch
  • Liquibet
  • Fantasy StarCraft
  • TLPD
  • StarCraft 2
  • Brood War
  • Blogs
Forum Sidebar
Events/Features
News
Featured News
[ASL22] Ro8 Preview: In A Tizzy9[ASL22] Ro16 Preview: Holy Diver5[ASL22] Ro16 Preview: Rough Waters10[ASL22] Ro24 Preview: Siren's Call8[ASL22] Ro24 Preview: Summer's End9
Community News
Weekly Cups (Sep 13-20): herO scores triple1BSL Season 235Weekly Cups (Sep 7-12): SHIN, ByuN, MaxPax double down1StarCraft open world shooter announced at BlizzCon101Weekly Cups (Aug 30-Sep 7): herO thrives amid growing schism10
StarCraft 2
General
Blizzard Classic Cup @ BlizzCon 2026 - $100k prize pool Balance hotfix patch 5.0.16b (July 16) How do you feel about the StarCraft shooter announcement at BlizzCon 2026? The Death of Cheese: From a Professional Cheeser Weekly Cups (Sep 13-20): herO scores triple
Tourneys
Stellar Fest TWO the Moon (Dec 16-20) Sparkling Tuna Cup - Weekly Open Tournament 2026 GSTL Announcement RSL Revival: Season 6 - Qualifiers and Main Event RSL goes to London! 2026 Offline Finals Nov 21-22
Strategy
[H] ZvP Mid-Late Game: Stalkers Collossi HT
Custom Maps
Nexus Wars 2021 GUIDE [M] (2) Industrial Park
External Content
Mutation # 544 Double Trouble The PondCast: SC2 News & Results Mutation # 543 Enhanced Defenses Mutation # 542 The Ascended
Brood War
General
BW General Discussion screpdb: new Starcraft reporting tool Bot on ladder [ASL22] Ro8 Preview: In A Tizzy new ranked ladder hack?
Tourneys
[ASL22] Ro8 Day 2 [ASL22] Ro16 Group D [ASL22] Ro8 Day 1 [IPSL] IPSL is Back With Winter 26-27!
Strategy
Cliff Jump Revisited (1 in a 1000 strategy) Replay Review Process - What do you do? Simple Questions, Simple Answers Odyssey Mineral Stack Saturation
Other Games
General Games
Diablo IV Nintendo Switch Thread Warcraft III: The Frozen Throne Stormgate/Frost Giant Megathread EVE Corporation
Dota 2
Dota 2 Champions League Season 3 Begins April 25! Official 'what is Dota anymore' discussion
League of Legends
[TL LoL EUW IHs] Teemo shall perish
Heroes of the Storm
Heroes of the Storm 2.0
Hearthstone
Deck construction bug
TL Mafia
TL Mafia Community Thread
Community
General
Russo-Ukrainian War Thread US Politics Mega-thread Artificial Intelligence Thread Things Aren’t Peaceful in Palestine All you football fans (soccer)!
Fan Clubs
MarineLorD Fan Club The Creator Fan Club The ShoWTimE Fan Club
Media & Entertainment
Movie Discussion! [Manga] One Piece Diablo Animated Series on Netflix
Sports
Football (Soccer) Thread TeamLiquid Health and Fitness Initiative For 2023 MLB/Baseball 2023
World Cup 2022
Tech Support
Computer Build, Upgrade & Buying Resource Thread
TL Community
Recent Gifted Posts
Blogs
Gaming Intensity, Problemati…
TrAiDoS
38 yo Retired SWE loo…
PurE)Rabbit-SF
Can Bots Beat Pros?? Starcr…
namkraft
[meme] I finally understa…
LUCKY_NOOB
Regacy Esports:Our Goa…
regacyesports
Dreaming of BW patches (mod…
c3rberUs
Customize Sidebar...

Website Feedback

Closed Threads



Active: 7578 users

Towards a good SC bot - P1 - Economy (1/2)

Blogs > imp42
Post a Reply
imp42
Profile Blog Joined November 2010
398 Posts
Last Edited: 2016-09-04 09:45:39
August 23 2016 18:40 GMT
#1
Towards a good StarCraft bot - Part 1 - "Economy (1/2)"

Summary:
+ Show Spoiler +
This post examines StarCraft economy from a static perspective. How much units can be produced given some income in minerals and gas? This is important, because it limits the solution space. The post concludes with a formula to calculate all possible combinations for Terran. It does not take into account time.


Introduction:
Before any attempt is made at creating a good StarCraft bot, we must thoroughly understand the mechanics of the game. Here, I discuss the mechanisms defining what a player can produce given his available resources. To draw an analogy between chess and StarCraft: In chess there are a total of 32 pieces. You start out with 16 pieces each and the number of pieces on the board can only decrease. It is quite easy to teach this knowledge to an AI. But in StarCraft, it is a problem on its own. Units can be produced. Furthermore, what can be produced depends on what has been produced in the past.

In Economics, there is a concept called the “Production Possibilities Frontier”. It defines what a given economy is maximally able to produce in terms of a combination of different goods. Typically, it is illustrated using a two-goods economy and it is assumed that there is a marginal cost per unit because of the law of diminishing returns. That is, each additional unit is worth less than the previous one (if you’re hungry, you might pay 5$ for your first slice of pizza. But probably won’t even want to pay 1$ for the 10th slice, because you are too full). This leads to production possibilities frontiers looking like this:

[image loading]

As can be seen, the depicted economy can either produce 10 computers or 100 books or any combination within the area defined by the border line. If the combination is on the border line it is a maximal production. Examples: 40 books and 6 computers is not maximal, but 60 books and 8 computers is. I both economics and StarCraft we want to be on the border line = on the production possibilities frontier. This translates to having few resources in the bank and producing as many units as possible.

But StarCraft is a bit more complicated than this.
1. There is not one single currency, but two (minerals and gas – there is no way to trade one for the other).
2. There are not two goods, but between 12 and 14, depending on race (counting producible units in Brood War).
3. there is no direct marginal cost. At least not as long as we ignore time.
4. the income we are allowed to spend is not constant, but it 1) increases over time and 2) depends on what we produce (workers increase income, all other units don’t).

As pointed out in part 0 of this series, my approach is to simplify, simplify, simplify. Point 4 is really complicated and probably requires differential equations to solve, so I leave it out for now. furthermore, it is often useful and simpler to focus on the upper bound of economies. An upper bound does not define what you can produce, but it guarantees that it is impossible to produce more than the upper bound.

I start really easy by examining one currency and one unit, no buildings, or supply cost: Just SCVs.

Just SCVs

1. “Given x minerals, how many SCVs can I produce?”

OK, this is really quite easy. An SCV costs 50 minerals. Up to 49 minerals the answer is zero, from 50 to 99 it is one, and so on. Mathematically: round(x / 50). Note that just x / 50 (without rounding) also represents an upper bound, albeit a less precise one.

Anyways, let’s raise the bar just a little bit:

SCVs and vultures

2. “Given x minerals, how many SCVs and / or vultures can I produce?”

The chart for an income of 1000 minerals looks like this:

[image loading]

We can produce any combination between 20 SCVs and 13 vultures, as long as we stay on the blue line (i.e. we don’t spend more than 1000 minerals). The reason it looks a bit like a stair is because a vulture costs 1.5x as much as a SCV. It therefore doesn’t matter whether we produce 11 or just 10 SCVs, we still can only produce 6 vultures with the remaining minerals.

Mathematically, we solve x * 50 + y * 75 <= 1000. x = number of SCVs, y = number of vultures
Economically speaking, we can trade 1 vulture for 7/5 SCVs, or the other way around, we can trade 1 SCV for 5/7 vultures.

Still too easy for you? Let’s look at all 13 units then.

All 13 units

Now, it’s not really feasible to plot a graph anymore, since the production space is 13-dimensional. However, fortunately the mathematical equation is still simple:
x1*c1 + x2*c2 + x3*c3 + x4*c4 + x5*c5 + x6*c6 + x7*c7 + x8*c8 + x9*c9 + x10*c10 + x11*c11 + x12*c12 + x13*c13 <= Income
where xi denotes the amount of unit i and ci the cost of the unit i. Income = available minerals.

The result is a valid upper bound that actually applies to the game. However, it is imprecise. The reasons:
- We did not account for production buildings and supply depots
- We did not consider gas cost

So let’s do gas cost first. This simply turns cost and income into a 2-dimensional variable. Instead of cost = 50 we now write cost = (50,0) indicating 50 minerals and 0 gas. Instead of Income = 1000 minerals we now write Income = (m, g).
The constraint above: x1*c1 + … x14 * c14 <= Income must now be written twice. Once for minerals and once for gas.

1) x1*cm1 + … x14 * cm14 <= minerals income
2) x1*cg1 + … x14 * cg14 <= gas income

Now, let’s account for production buildings and supply depots.

a) Production buildings

Production buildings are easy. In economics terms, they just represent a fixed cost, as opposed to the variable cost of the unit. So instead of stating “total cost = number of units * cost of unit” the equation now states “total cost = number of units * cost of unit + fixed cost”
I briefly list the equation for all Terran units considering their technology tree, to not lose touch with StarCraft
Note that I list the cost of a required building on the tech tree only once (lazy...).

List of variable and fixed cost for each Brood War Terran unit, accounting for tech tree requirements:
+ Show Spoiler +
Cost of SCVs = x * (50,0)

Cost of marines = x * (50,0) + (150,0)
requirements: barracks (150,0)

Cost of firebats = x * (50,25) + (300,0)
requirements: barracks, academy (150,0)

Cost of medics = x * (50,25) + (300,0)
requirements: barracks, academy

Cost of ghosts = x * (25,75) + (800,400)
requirements: barracks, factory, starport, academy, science facility, covert ops (50,50)

Cost of Vulture = x * (75,0) + (350,100)
requirements: barracks, factory (200,100)

Cost of Siege tank = x * (150,100) + (400,150)
requirements: barracks, factory, machine shop (50,50)

Cost of Goliath = x * (100,50) + (450,150)
requirements: barracks, factory, armory (100,50)

Cost of Wraith = x * (150,100) + (500,200)
requirements: barracks, factory, starport (150,100)

Cost of Dropship = x * (100,100) + (550,250)
requirements: barracks, factory, starport, control tower (50,50)

Cost of Science Vessel = x * (100,225) + (650,400)
requirements: barracks, factory, starport, control tower, science facility (100,150)

Cost of Battlecruiser = x * (400,300) + (650,400)
requirements: barracks, factory, starport, science facility, physics lab (50,50)

Cost of Valkyrie = x * (250,125) + (650,300)
requirements: barracks, factory, starport, control tower, armory

This list does not account for synergies. The total cost of 1 marine is 200 minerals, the total cost of 1 vulture is 425 minerals and 100 gas. However, the total cost of 1 marine + 1 vulture is 475 minerals and 100 gas, because only 1 barracks is needed.



Let's account for supply cost:

b) Supply depots

Supply makes matters worse. Not only do we have to factor in an additional 100 minerals every 8 supply, but now we also have to introduce further constraints and account for supply cost of all units. The mineral cost of supply is supply * 12.5 (supply / 8 depots, costing 100 minerals each).

The additional constraints therefore are:

1) x1*s1 + … x14 * s14 <= max supply, where si is the supply cost of unit i
2) x1*c1 + … x14 * c14 <= income – (x1*s1 + … x14 * s14)*12.5

Note that 2) can be written as x1*(c1+S1*12.5) + … + x14*(c14+S14*12.5) <= income

Let’s wrap it up by stating the final equation for an upper bound of the PPF, considering mineral cost, gas cost, supply cost, and technology tree for Terran units:

Income(minerals,gas) >=
#SCVs * (50+12.5,0) + #marines * (50+12.5,0) + #firebats * (50+12.5,25) + #medics * (50+12.5,25) + #ghosts * (25+12.5,75) + #vultures * (75+25,0) + #siegetanks * (150+25,100) + #goliaths * (100+25,50) + #wraiths * (150+25,100) + #dropships * (100+25,100) + #science vessels * (100+25,225) + #battle cruisers * (400+75,300) + #valkyries * (250+37.5,125) + (1050,600)

Supply constraint:
200 = #SCVs * 1 + #marines * 1 + #firebats * 1 + #medics * 1 + #ghosts * 1 + #vultures * 2 + #siegetanks * 2 + #goliaths * 2 + #wraiths * 2 + #battlecruisers * 6 + #valkyries * 3

Conclusion:

The real challenge in StarCraft economy is the fact, that income increases over time and depends on what has been produced in the past (If I made more SCVs in the past, I will have a larger income now). This post examined the static view only. However, it is a necessary preliminary work for the next post to come, which will examine income over time.

Both concepts together will then allow us to provide a reasonable upper bound of what each player can possible have produced at any given time in the game.

Disclaimer: I double checked numbers, but did calculations manually – there might be some typos
Data was retrieved from http://wiki.teamliquid.net/starcraft/Technology_tree and http://wiki.teamliquid.net/starcraft/Units



****
50 pts Copper League
Endymion
Profile Blog Joined November 2009
United States3701 Posts
August 23 2016 19:38 GMT
#2
since you're writing terran AI, how are you going to factor travel time into your mineral/gas projections when you're accounting for the cost of producing a building? each time you build a building you'll have a fairly random start point for an SCV, as well as a difficult to determine from an AI perspective construction site for the building. are you just going to go with an average time and work on the complicated algorithms related to building placement later, or are you going to do it before??? because if you don't account for it your build orders will always be off by like (x*8)*2 where x is the number of buildings you have constructed, leading to some pretty material impacts on your builds, especially if you're building your AI to think off of flawed logic
Have you considered the MMO-Champion forum? You are just as irrational and delusional with the right portion of nostalgic populism. By the way: The old Brood War was absolutely unplayable
Kau *
Profile Joined March 2007
Canada3502 Posts
August 23 2016 19:44 GMT
#3
I think there's a miscalculation for the cost of ghost's prerequisite buildings?

How does knowing what a player can have at any time in the game help?
Moderator
Endymion
Profile Blog Joined November 2009
United States3701 Posts
August 23 2016 19:52 GMT
#4
On August 24 2016 04:44 Kau wrote:
I think there's a miscalculation for the cost of ghost's prerequisite buildings?

How does knowing what a player can have at any time in the game help?


it's the same idea as when you're playing as a human.. if you're scouting a terran in BW at like 22 supply as a zerg, you can count his marines and say "hmm, okay, if he has 8 marines then he has 2 or 3 rax that he built at standard timings" which gives you insight into his built, allowing you to more accurately adapt your build to kill whatever his build is moving towards.. if you're an AI, you can count exactly how much minerals and gas worth of buildings and structures you have seen, and get an even more accurate picture of the information that you haven't seen yet.. as you get later in the game information tells you less and less because there are more possible leadups from the past, but it's pretty common early game (and one of the strongest strengths of AI in rts).
Have you considered the MMO-Champion forum? You are just as irrational and delusional with the right portion of nostalgic populism. By the way: The old Brood War was absolutely unplayable
Randomaccount#77123
Profile Blog Joined May 2010
United States5003 Posts
Last Edited: 2016-08-23 20:26:55
August 23 2016 20:25 GMT
#5
--- Nuked ---
imp42
Profile Blog Joined November 2010
398 Posts
August 24 2016 09:52 GMT
#6
On August 24 2016 04:44 Kau wrote:
I think there's a miscalculation for the cost of ghost's prerequisite buildings?

How does knowing what a player can have at any time in the game help?


Thanks, I fixed Ghost fixed cost. Great find!

As others pointed out, a good approximation of the PPF helps judging strength of enemy force as well as BO based on limited scouting information. I'd say it is mostly useful the first 5 minutes.

Remember also that to an AI it is not obvious that there can't be a Battlecruiser out the first 30 seconds :p
You need to somehow create that knowledge. The PPF is one way to accomplish this.
50 pts Copper League
imp42
Profile Blog Joined November 2010
398 Posts
August 24 2016 10:00 GMT
#7
On August 24 2016 04:38 Endymion wrote:
since you're writing terran AI, how are you going to factor travel time into your mineral/gas projections when you're accounting for the cost of producing a building? each time you build a building you'll have a fairly random start point for an SCV, as well as a difficult to determine from an AI perspective construction site for the building. are you just going to go with an average time and work on the complicated algorithms related to building placement later, or are you going to do it before??? because if you don't account for it your build orders will always be off by like (x*8)*2 where x is the number of buildings you have constructed, leading to some pretty material impacts on your builds, especially if you're building your AI to think off of flawed logic


Building time and travel time are very valid concerns. I have planned to account for them as follows:

Travel time: I just take a factor, such as say 3 seconds, and increase it by a small amount for every additional building.

Building time is much more relevant IMO, because an SCV building is an SCV not mining. Let's say you loose 5 minerals travelling to build a barracks. Actually building it looses you another 80 minerals (build time is 80 seconds).

The calculations might be off for some cases, but I intend to compare calculations with actual game data and then adjust coefficients in the calculations until it fits well.
50 pts Copper League
Endymion
Profile Blog Joined November 2009
United States3701 Posts
August 24 2016 10:13 GMT
#8
On August 24 2016 19:00 imp42 wrote:
Show nested quote +
On August 24 2016 04:38 Endymion wrote:
since you're writing terran AI, how are you going to factor travel time into your mineral/gas projections when you're accounting for the cost of producing a building? each time you build a building you'll have a fairly random start point for an SCV, as well as a difficult to determine from an AI perspective construction site for the building. are you just going to go with an average time and work on the complicated algorithms related to building placement later, or are you going to do it before??? because if you don't account for it your build orders will always be off by like (x*8)*2 where x is the number of buildings you have constructed, leading to some pretty material impacts on your builds, especially if you're building your AI to think off of flawed logic


Building time and travel time are very valid concerns. I have planned to account for them as follows:

Travel time: I just take a factor, such as say 3 seconds, and increase it by a small amount for every additional building.

Building time is much more relevant IMO, because an SCV building is an SCV not mining. Let's say you loose 5 minerals travelling to build a barracks. Actually building it looses you another 80 minerals (build time is 80 seconds).

The calculations might be off for some cases, but I intend to compare calculations with actual game data and then adjust coefficients in the calculations until it fits well.

building time is important because it's a larger amount of time, but it's wholly predictable.. however, saying you have a 3 second travel time for an SCV on 투혼 going from your main to ramp or natural to create a rax is going to be drastically different than building a depot right next to your minerals.. imo the easiest way is to just tell the AI where the best place to put buildings for each starting point is, but that kinda screws you if you need to play python instead of FS or something.. either way i think it's going to make it hard to be efficient without writing actual builds into it
Have you considered the MMO-Champion forum? You are just as irrational and delusional with the right portion of nostalgic populism. By the way: The old Brood War was absolutely unplayable
imp42
Profile Blog Joined November 2010
398 Posts
August 24 2016 10:28 GMT
#9
On August 24 2016 19:13 Endymion wrote:
Show nested quote +
On August 24 2016 19:00 imp42 wrote:
On August 24 2016 04:38 Endymion wrote:
since you're writing terran AI, how are you going to factor travel time into your mineral/gas projections when you're accounting for the cost of producing a building? each time you build a building you'll have a fairly random start point for an SCV, as well as a difficult to determine from an AI perspective construction site for the building. are you just going to go with an average time and work on the complicated algorithms related to building placement later, or are you going to do it before??? because if you don't account for it your build orders will always be off by like (x*8)*2 where x is the number of buildings you have constructed, leading to some pretty material impacts on your builds, especially if you're building your AI to think off of flawed logic


Building time and travel time are very valid concerns. I have planned to account for them as follows:

Travel time: I just take a factor, such as say 3 seconds, and increase it by a small amount for every additional building.

Building time is much more relevant IMO, because an SCV building is an SCV not mining. Let's say you loose 5 minerals travelling to build a barracks. Actually building it looses you another 80 minerals (build time is 80 seconds).

The calculations might be off for some cases, but I intend to compare calculations with actual game data and then adjust coefficients in the calculations until it fits well.

building time is important because it's a larger amount of time, but it's wholly predictable.. however, saying you have a 3 second travel time for an SCV on 투혼 going from your main to ramp or natural to create a rax is going to be drastically different than building a depot right next to your minerals.. imo the easiest way is to just tell the AI where the best place to put buildings for each starting point is, but that kinda screws you if you need to play python instead of FS or something.. either way i think it's going to make it hard to be efficient without writing actual builds into it


I agree, but remember that the calculations mostly serve the purpose of estimating what the opponent could possibly have. And since I examine upper bounds I have to consider shortest possible travel time.
This becomes relevant e.g. when you are zerg and want to know the maximum number of marines the opponent could possibly have by the time you walk into his base.

Also, I will start with only 1 matchup on 1 map, so I can adjust coefficients for travel time to accomodate for the map, if necessary.
A topic I haven't touched yet: a machine learning mechanism will be able to adjust all results based on actually observed numbers.
50 pts Copper League
Endymion
Profile Blog Joined November 2009
United States3701 Posts
August 24 2016 10:56 GMT
#10
ah okay sorry i missed that part, i thought you were using this as a core system for writing build order logic as well, as an estimation you're right that losing 8 minerals here and there probably won't make a difference in ai vs ai
Have you considered the MMO-Champion forum? You are just as irrational and delusional with the right portion of nostalgic populism. By the way: The old Brood War was absolutely unplayable
Please log in or register to reply.
Live Events Refresh
Next event in 12m
[ Submit Event ]
Live Streams
Refresh
StarCraft 2
ViBE156
Has 41
PiGStarcraft22
StarCraft: Brood War
Britney 11460
GuemChi 727
Artosis 590
Rain 244
Hyun 20
910 13
Dota 2
monkeys_forever544
League of Legends
Doublelift6303
Counter-Strike
minikerr18
Super Smash Bros
hungrybox346
Heroes of the Storm
Grubby4151
Other Games
summit1g9736
Coldzera 814
shahzam559
Day[9].tv454
JimRising 393
C9.Mang0288
CosmosSc2 29
Organizations
Other Games
gamesdonequick617
[ Show 13 non-featured ]
StarCraft 2
• Hupsaiya 35
• mYiSmile128
• musti20045 23
• davetesta7
• AfreecaTV YouTube
• intothetv
• Kozan
• IndyKCrew
• Migwel
StarCraft: Brood War
• BSLYoutube
• STPLYoutube
• ZZZeroYoutube
Other Games
• Day9tv454
Upcoming Events
PiGosaur Cup
12m
Kung Fu Cup
11h 12m
The PondCast
1d 10h
Replay Cast
2 days
Korean StarCraft League
3 days
CranKy Ducklings
3 days
GSL
4 days
Yamato Cup
4 days
Replay Cast
5 days
Afreeca Starleague
5 days
[ Show More ]
WardiTV Weekly
5 days
Sparkling Tuna Cup
6 days
Afreeca Starleague
6 days
Liquipedia Results

Completed

Super Anchor Qualifying S3
Blizzard Classic Cup 2026
Big Dog Cup 2026 Div 1

Ongoing

ASL Season 22
CSL 2026 AUTUMN (S22)
Acropolis #5
Acropolis #5 - GSA
Calamity Invitational
Logitech G Play Connect 2026
SL StarSeries Fall 2026
FISSURE Playground #3
BLAST Open Fall 2026
Esports World Cup 2026
BLAST Bounty Summer 2026
BLAST Bounty Summer Qual
Stake Ranked Episode 3
XSE Pro League 2026

Upcoming

Acropolis #5 - GSB
Acropolis #5 - GSC
BSL Season 23
SC4ALL II: Brood War
BSL 23: Non-Korean Championship
HSC XXX
Stellar Fest 2: Lunar Cup
SC4ALL II: StarCraft II
Kung Fu Cup 2026 Grand Finals
RSL Offline Finals
Custodian Cup
Copium Cup
PGL Major Singapore 2026
Stake Ranked Episode 6
BLAST Rivals Fall 2026
IEM Beijing 2026
Stake Ranked Episode 5
PGL Masters Bucharest 2026
1win Private Club #2
Thunderpick World Champ. '26
ESL Pro League Season 24
Stake Ranked Episode 4
1win Private Club #1
TLPD

1. ByuN
2. TY
3. Dark
4. Solar
5. Stats
6. Nerchio
7. sOs
8. soO
9. INnoVation
10. Elazer
1. Rain
2. Flash
3. EffOrt
4. Last
5. Bisu
6. Soulkey
7. Mini
8. Sharp
Sidebar Settings...

Advertising | Privacy Policy | Terms Of Use | Contact Us

Original banner artwork: Jim Warren
The contents of this webpage are copyright © 2026 TLnet. All Rights Reserved.