geniusyield/cardano-hackathon-argentina-2022

This is stars and forks stats for /geniusyield/cardano-hackathon-argentina-2022 repository. As of 28 Apr, 2024 this repository has 3 stars and 8 forks.

Testing tokens Minting Policy: 789824bb4bcdc53d96c9114a9aec50149eb0ec7198d292c172556ec1 Token 1: TokenA Token 2: TokenB Competition asset classes fa2261efef35419afb055426b651b404356d803901a89c14cee9a926.TokenA fa2261efef35419afb055426b651b404356d803901a89c14cee9a926.TokenB DEX architecture diagram geniusyield-orderbot This is an order matching bot implementation that is meant to be modular and polymorphic. It uses backpack to support this goal. Solid resource for learning backpack: https://github.com/danidiaz/really-small-backpack-example Development environment In theory, you only really need cabal (and similar Haskell tooling) - and some generally expected system libraries and tooling to work with this project. You should also have cardano-cli available to generate a signing key. You may also need to install a few system libraries: libsecp256k1 and libsodium. Depending on your linux distribution, you may have to install the -dev versions which include the C headers. In particular, a sufficiently updated version of libsecp256k1 must be available. Older versions do not have secp_extrakeys.h which is required. If your distribution does not have up to date packages, you may have to build from source. libsecp256k1 should also be built with the flags --enable-module-schnorrsig --enable-experimental. If you want to use a haskell.nix based environment instead which includes all these (but takes a long time to initialize the first time), you can use the provided flake.nix. Of course, you must have nix set up with the flakes feature enabled. You must also set up the IOG binary caches following the official guide. If you do not do this, you'll end up building several copies of GHC from source - which is an extremely long process that may potentially fail. Once you have the above all set up, you can run nix develop within the project root which will drop you inside a shell with everything you need, including cabal. After this point, you can do cabal build all and similar commands. You can also utilize the provided Makefile commands, such as make orderbot-blockfrost - which will start the orderbot with the config-blockfrost.json config. Do note that the first time you set up this nix shell, it will take a long time to build everything. However, next runs should be fast unless you severely modify the dependencies. Adding some deps from hackage into your cabal file is fine - but modifying the index-state may lead to build failures as newer packages might not be supported by the libraries! If you must use hackage packages newer than what is provided by the index-state, ask the organizers for help. They may be able to provide you with cabal constraints and allow-newer stanzas to fix build failures. Setting up config files The two provided configs are missing the respective API key/token, which you will have to provide. Components The orderbot has three modular components, two of them are modularized through backpack, and one of them is modularized just by virtue of being a higher order function. The backpack-modularized components are: The data source where the orderbook is populated from (lib-datasource) The orderbook data structure (lib-orderbook) The other component is the matching strategy. Which is a function that has access to an orderbook for each order asset pair, and performs order matchings to yield information on which orders/pools to match. The returned information is then used to create and submit matching transactions. Adding implementation for components The implementation for the data source component is already provided for participants to use. It is defined in impl/datasource-providers. Participants will have to implement their own orderbook. An example implementation is provided in impl/orderbook-list. But of course, it is intentionally suboptimal! However, it should serve as a good example to get started with and even run the bot with! You'll note that the Orderbook signature defines many orderbook querying functions. These are only provided as hints for the participants. You are not required to be limited by only using these functions in your strategy - you may have more functions here that you import in your Main file! Though you must provide at least these functions to make sure the signature is compatible with the implementation. There is extensive documentation on understanding the types that are to expected to be used by all orderbook implementations, in GeniusYield.OrderBot.Types. It explains concepts with regards to asset pairs, orders and how they are separated into buy/sell etc. You should especially read mkOrderInfo if you want to understand how DEX orders (which are always "sell" orders) are distinguished into buy and sell orders. This distinction is made to make it easier for the matching strategy writer to match orders. Provided implementations and running the bot Once you have these two implementations, you can link them up in the Main file that will contain the bot execution itself. The provided Main file uses the provided datasource component (which Participants will also use), as well as the example orderbook-list implementation. You may check out the respective stanza in the geniusyield-orderbot.cabal file for this executable (geniusyield-orderbot-exe), which showcases how cabal mixins may be used to link up backpack signatures with their implementations: executable geniusyield-orderbot-exe import: common-lang import: common-deps import: common-ghc-opts main-is: Main.hs build-depends: , geniusyield-orderbot-framework , geniusyield-orderbot-framework:common , geniusyield-orderbot:datasource-providers , geniusyield-orderbot:orderbook-list mixins: , geniusyield-orderbot:orderbook-list requires (GeniusYield.OrderBot.DataSource as GeniusYield.OrderBot.DataSource.Providers) , geniusyield-orderbot-framework requires ( GeniusYield.OrderBot.DataSource as GeniusYield.OrderBot.DataSource.Providers , GeniusYield.OrderBot.OrderBook as GeniusYield.OrderBot.OrderBook.List ) ghc-options: -O2 -threaded -rtsopts -with-rtsopts=-N Most likely, participants will replace the geniusyield-orderbot:orderbook-list dependency with their own implementation library. Once they have done that, the mixin must be accurately updated. Consider, a custom orderbook implementation defined in a cabal sublibrary geniusyield-orderbot:orderbook-tree. Suppose, the implementation module in said library is: GeniusYield.OrderBot.OrderBook.Tree. Then, the new stanza would look like: executable geniusyield-orderbot-exe import: common-lang import: common-deps import: common-ghc-opts main-is: Main.hs build-depends: , geniusyield-orderbot-framework , geniusyield-orderbot-framework:common , geniusyield-orderbot:datasource-providers , geniusyield-orderbot:orderbook-tree mixins: , geniusyield-orderbot:orderbook-tree requires (GeniusYield.OrderBot.DataSource as GeniusYield.OrderBot.DataSource.Providers) , geniusyield-orderbot-framework requires ( GeniusYield.OrderBot.DataSource as GeniusYield.OrderBot.DataSource.Providers , GeniusYield.OrderBot.OrderBook as GeniusYield.OrderBot.OrderBook.Tree ) ghc-options: -O2 -threaded -rtsopts -with-rtsopts=-N Customizing the bot Now that you know how to set up the project itself and link components. Of course the next step is to write a matching strategy that beats out everyone else's! A very simple sample strategy is provided in the example Main file. Of course, it uses the list based orderbook and therefore it uses list functions in the strategy. You can use your custom functions in your strategy once you import the respective Orderbook implementation. What else is there to customize? Quite a lot actually! Here's what the OrderBot type looks like: -- See: GeniusYield.OrderBot -- | The order bot is product type between bot info and "execution strategies". data OrderBot = OrderBot { -- | Signing key of the bot. botSkey :: !GYPaymentSigningKey , -- | UTxO ref of the collateral UTxO in the bot's wallet. botCollateral :: !GYTxOutRef , -- | The execution strategy, which includes and governs the matching strategy. botExecutionStrat :: !ExecutionStrategy , -- | Function that can be used to filter out uninteresting orders/pools. -- -- Before retrieving all information on an order/pool, this function will be applied -- on the 'GYAssetClass'(s) it contains. If it returns true for _any of the assets_ -- contained within, the order/pool will be fetched. botAssetFilter :: GYAssetClass -> Bool , -- | Handler for exceptions that may be raised during bot run. It'll also have access to the matched orders. -- -- NOTE: This handler is when used during tx build and submission step. It is assumed all other -- steps are pure and should not raise exceptions under usual circumstances. This also applies to the -- OrderBook population step (when user supplied 'populateOrderBook') is called - which is similarly -- assumed to not fail. -- -- If you find exceptions outside of the tx build/submission step common, you may wrap 'runOrderBot' with -- a "Control.Exception.try" or similar and handle/restart your bot from there. botOnException :: [MatchExecutionInfo] -> SomeException -> IO () , -- | How many microseconds to wait after a tx submission before rescanning the chain for orders. botRescanDelay :: Int } Once you have this data type constructed, you simply pass it to the function GeniusYield.OrderBot.runOrderBot. Let's see what the example strategy uses: OrderBot { botSkey = skey , botCollateral = collateral , botExecutionStrat = MultiAssetTraverse $ \_ bk -> sampleStrategy bk , botRescanDelay = 50_000_000 -- 50 Seconds , botAssetFilter = const True -- Fetch everything! , botOnException = \matchSet err -> do print err BSL.putStr $ AesonE.encodePretty matchSet putStrLn "\n" } We will discuss the skey and collateral below, for now: notice that you are able to set the following: The execution strategy itself, which contains the respective matching strategy. Currently, there is only one matching strategy: MultiAssetTraverse. What does this mean? In the real world, the DEX orders will consist of many many unique asset pairings; which means there will be an OrderBook for each asset pair. The MultiAssetTraverse is meant to independently traverse over each of these OrderBooks and yield matches. For the hackathon, there will only be one asset pair. The identifier of which, will be provided to you by the organizers. So in that sense, MultiAssetTraverse is the only execution strategy that makes sense and of course it will only have one orderbook within it. The execution strategy then takes a function as an argument. This is your matching strategy. You will have access to the orderbook and its respective unique OrderAssetPair within this strategy, and you are expected to return information on matches in form of MatchExecutionInfo. There is extensive documentation on OrderAssetPair and relevant types in GeniusYield.OrderBot.Types - you should read this to understand how to work with them in your matching strategy. Documentation on MatchExecutionInfo can be found in GeniusYield.OrderBot.MatchingStrategy. The rescan delay. This is the amount of micro seconds to wait after a transaction submission before the blockchain is scanned again for changes in the DEX. Remember that changes in the blockchain take time! You cannot submit transactions back to back. It is the participants' responsibility to figure out the optimal waiting time! Make it too high, and your bot will slow down compared to others. Make it too low and you'll be wasting precious computation resources on transactions that ultimately fail! An asset filtering function. Only orders with assets for which this function returns true will be fully fetched. It is highly recommended that Participants utilize this to increase throughput. The final competition will use a constant PolicyId, which will be given to you by the organizers - you may filter for only the asset classes with that policy id. Of course, the example simply fetches everything. The exception handler. This handles exceptions raised during transaction build and submission. There will be various exceptions during such a stateful operation. It might range from user errors like improper matching, to contention errors on the chain due to other bots trying to match the same orders! It's the participants' responsibility to potentially scrutinize and handle these exceptions. The handler will also have access to the list of MatchExecutionInfo returned by your strategy that caused the transaction to fail. What about exceptions raised during other operations in runOrderBot, other than transaction build/submission? These are not wrapped by your provided exception handler. The assumption is that the other components should not fail, ideally. However, if you do see them failing, wrap the runOrderBot call with a try, or catch, or similar where you can have another exception handler. However, do note that you'll of course have to call runOrderBot again to restart the bot in this handler, since it crashed! Logistics and Orchestration The Main executable takes a path to a config file containing some information to run the bots. This is meant to be passed as a command line argument. There are two sample configs provided: config-maestro.json and config-blockfrost.json. There are also two make commands to run the bot with them, respectively: make orderbot-maestro and make orderbot-blockfrost The configuration file describes the blockchain information provider to use. The supported providers can be seen in: GeniusYield.OrderBot.Config.GYCoreProviderInfo. We expose two easy to use providers in this repository: The Maestro blockchain API, provided by Maestro. Blockfrost Both of these need an API token. The Maestro API token is simply a number; config-maestro.json uses 42 as an example - but it is not the correct API token! Please replace it with the proper API token, which will be provided to you by the Hackathon organizers. You can obtain a Blockfrost API key by creating a Blockfrost account and then a project with the Preprod testnet. Paste said key inside config-blockfrost.json. Finally, you need to create a signing key that the bot will use, as well as a collateral UTxO within the bot's wallet. Save the signing key to a file. The example main file sets up the scaffolding for using the skey and the collateral: skey <- readPaymentSigningKey "bot.skey" let collateral = "ca54517fd33bb9abafc1c48eb7e3ee6a7681b5ef38a4eaa8e56ab717b7705622#1" Replace bot.skey with the path to your bot's signing key. Replace collateral with a UTxO reference of the collateral in the bot's wallet which it will use to interact with smart contracts. Notice that the collateral can be specified using a string that contains the Transaction Id hash and the UTxO index, joined by a #. For instructions on creating a signing key and collateral, refer to: Setup wallet instructions You must provide the wallet address of the bot to be eligible for the final competition! Matching strategy tips Remember that you should only match orders where the overall arbitrage profit is at least a whole number. Payment to a DEX order is always ceiled, so if the calculated payment is 2.13 tokens, the order actually gets paid 3 tokens. So if you expected arbitrage profits in change, it's not going to happen! Don't return too many 'MatchExecutionInfo's! There is a limit to how many transactions can be matched in a single transaction. The exact limit, will have to be found out by the Participants. When you try to match too many orders in one transaction, you'll see ExUnitsTooBigUTxO within the raised exception. Common Transaction Exceptions ExUnitsTooBigUTxO in the exception means you are trying to match too many orders at once, making the transaction size cross the limit. BalancingErrorInsufficientFunds in the exception indicates there are not enough of one or many tokens to construct the tranasction. If you see ada in the value that is printed afterwards, it means your bot is out of ada. More often however, this error will be raised if your matching strategy does not return proper order matches and there aren't enough tokens in the transaction bucket to pay an order. GYTxMonadException "partiallyFillPartialOrder: amount x must be smaller than offered amount x - You are trying to partially fill an order, but the partial fill amount is the max volume of the order. Use CompleteFill instead. See GeniusYield.OrderBot.MatchingStrategy.FillType for more information. BadInputsUTxO in the exception that is raised during tx submission, not creation/balancing, usually indicates contention. An order you are trying to match is being matched by another transaction. Troubleshooting HLS will not work in signature modules, nor will it work in modules importing a signature module. Importing implementation modules is fine though! The order of type variables in a forall must match exactly in the signature and implementation. Remember that the order of type variables in implicit foralls is well defined - that is, it is simply in the order they appear from left to right, in the type signature. If you see an error like this in within the nix shell: ghc: panic! (the 'impossible' happened) (GHC version 8.10.7: getPackageDetails: couldn't find package blockfrost-api-0.6.0.0-<hash> Exit and re-enter the nix shell: exit, followed by nix develop. DEX Table of Content Introduction Orders Partially Fillable Orders Liquidity Positions NFT Introduction The purpose of the DEX smart contracts is to implement a DEX, a Decentralized Exchange, i.e. a means for users to swap (Cardano native) tokens, which include ada and arbitrary custom tokens. This is Decentralized in the sense that no central authority (like a traditional, centralized exchange) is needed: Users can swap without needing to trust some authority or each other. We anticipate two types of users, those who are primarily interested in swapping one kind of token for another and liquidity providers, who facilitate swaps by providing liquidity (i.e. large amounts of tokens). Normal users have one sort of token and want to exchange it for another. Liquidity providers want to earn fees by making it easier for normal users to perform their swaps. The process works as follows: Users who want to swap place orders on the blockchain (in the form of an UTxO "sitting" at a script address). They include their "offer" (the tokens they want to swap) in their order and state their price (in the datum). Anybody who is willing to pay the price can do so by filling the order, i.e. paying the price in exchange for the offered tokens. Liquidity providers open liquidity positions by placing tokens of two sorts in an UTxO on the blockchain at another script address. These liquidity positions specify a price, and anybody can put tokens of the one sort into a position in exchange for getting tokens of the other sort out - at the specified price. Liquidity positions make it easier to fill orders: If an order offers some amount of a token A in exchange for some amount of a token B and if there is a liquidity position containing A's and B's in sufficient quantity and at an appropriate price, then the order can be filled by utilizing the liquidity position. Orders and liquidity positions are not "aware" of each other and can be freely combined in transactions. It is in principle possible to fill several orders in one transaction using several liquidity positions. Anybody can fill orders and use liquidity positions for swaps. We anticipate that there will be a large number of "bots", programs that scan the blockchain for open orders and use diverse strategies to fill them - with or without the help of liquidity positions. Those bots can "pocket" any arbitrage they encounter, which will motivate bot authors and lead to a healthy competition among them, resulting in orders being filled swiftly and the market becoming as efficient as possible. Our DEX architecture is very open and extensible, but at the moment, we support two types of orders and one type of liquidity positions. In total, we have four different smart contracts powering our DEX: Orders. A simple order that has to be filled in one transaction. Partially Fillable Orders. A more complex order that can be filled in several transaction, each only taking part of the offered tokens and paying part of the price. Liquidity Positions. Deposits of two types of tokens that can be swapped in both directions at a specified price. NFT. A simple NFT minting policy which is used by both partially fillable orders and liquidity positions to track their "identity" over their lifetime. There is no mutable state in the EUTxO-model. To simulate state, UTxO's have to be consumed, and new ones (with modified value and/or datum) have to be created. NFT's are passed from one UTxO to the other to provide continuity and to identify the old and new UTxO as representing the "same" abstract order respectively liquidity position. Orders Orders contain an arbitrary value (which can in principle be a bag of different tokens) and specify a price (which is again an arbitrary value, not necessarily just consisting of one token). Their datum specifies the pubkey hash of the owner (the user placing the order), the owner's address and the price. Orders support two types of "actions" (ways to consume them), they can be cancelled or filled. A transaction cancelling an order needs to be signed by the owner. To fill an order, the specified price needs to be paid to the owner's address. Orders are supposed to be composable with other orders and with liquidity positions, so we must be careful to not run afoul of the double satisfaction problem, where a malicious player could fill several orders but only pay once. It is surprisingly difficult to write a validator in a way that prevents this, because validators run independently of each other and can't communicate, so a validator can't simply "claim" a specific payment for itself. We solve this by making use of the fact that every UTxO on the Cardano blockchain can have a datum attached - not just UTxO's at script addresses. We will attach a reference to the filled order as datum to each payment, thus making payments for different orders distinguishable. Partially Fillable Orders The partial order mechanism allows one to lock their offered tokens inside a validator, which can then be either completely or partially filled/bought by another user - according to the price. Each order contains a specific NFT to identify it. Its TokenName must be present in the order datum upon order creation. The CurrencySymbol of this NFT is defined in NFT.hs (see NFT!). This order validator is parameterized over three arguments: The NFT minting policy ID A constant lovelace amount to pay for the order-filler's (buyer's) transaction fees, for a single transactions. A total fee amount is calculated based on this initial fee parameter, and the maximum number of partial filling transactions that can be done for the order - this total fee amount should be present at the UTxO during initial construction. For example, if the offer amount is 110. min filling is 20, and fee is 2 Ada, then the maximum number of partial transactions that can be done is $\lceil 110/20\rceil = 6$. So, $6\cdot 2 = 12$ Ada should be provided to the UTxO during construction to pay for each of those transactions' fees. A constant lovelace amount to pay for the deposit after a partial fill. This is similar to the fee amount in purpose. This corresponds to a minimum Ada amount that has to be present in an UTxO alongside other tokens. In the above example, there can be at most 6 fillings, so at most 5 partial fillings, which means the deposit should be included five times upon order creation. Once an order is properly constructed, it can be interacted with in one of three ways: Cancel - The order is cancelled, allowing the UTxO with the offer amount to be reclaimed. Requirements: Must be signed by the owner of the order. The owner's pub key hash is put into the order datum upon creation. The NFT identifying the order must be burnt, lest it escapes into another UTxO. Complete fill - The full offered amount is bought, for the full price. Requirements: Payment must be made to a UTxO at podOwnerAddr, containing a datum that contains the TxOutRef of the order UTxO. (i.e ownRef within the validator). The owner address is put into the datum during construction. The payment must be greater than, or equal to, the price (in the asked-for asset) for the full offer amount, it must also include the full deposit amount. However, the transaction fee for one transaction (corresponding to the fee parameter) may be deducted. The NFT identifying the order must be burnt, lest it escapes into another UTxO. Partial fill - a portion of the offered amount is bought, for the corresponding price. Requirements: The portion/amount must be positive, but less than the remaing total amount, and it must be greater than or equal to podMinFilling, which is put inside the datum during order creation. This is effectively the minimum amount one must buy even during a partial filling. Payment must be made to a UTxO at podOwnerAddr, containing a datum that contains the TxOutRef of the order UTxO. (i.e ownRef within the validator) The owner address is put into the datum during construction. The payment must be greater than, or equal to, the price (in the asked-for asset) for the corresponding portion of the offer amount, it must also include the deposit amount (single UTxO deposit), as given by the parameter. However, the transaction fee for one transaction (corresponding to the fee parameter) may be deducted. A single continuing output must be present, with the remaining amount inside, alongside an updated datum. In particular, the podOfferedAmount will be subtracted by the amount that has just been sold, and one deposit can be deducted. Liquidity Positions Liquidity providers can open a liquidity position by placing two sorts of tokens (with arbitrary amounts) at a UTxO "sitting" at the script address. Valid liquidity positions need to contain an NFT to track their identity, and they need to specify various parameters in the datum. The most important operation supported by such a liquidity position is that of swapping one sort of tokens for the other: A transaction consumes the UTxO and recreates a new one at the same address with adjusted amounts of tokens (according to the price). Each such swapping transaction needs to pay a fee to the liquidity provider. This fee is paid in the token being swapped and remains in the position, so that the amount of tokens in the position grows with each swap. Part1 of that fee is meant for GeniusYield, but also remains in the position until it is withdrawn or the position is closed. The datum contains the following information: data LiquidityDatum = LiquidityDatum { ldOwner :: !PubKeyHash -- ^ Pubkey hash of the owner (i.e. the liquidity provider). , ldNFT :: !TokenName -- ^ Token name of the NFT tracking the identity of the liquidity position. , ldFrom :: !AssetClass -- ^ One of the tokens in the position. , ldTo :: !AssetClass -- ^ The other token in the position. , ldPrice :: !(PlutusTx.Integer, PlutusTx.Integer) -- ^ The price for one unit of the "from" token in terms of the "to" token, given as numerator and denominator of the actual rational price. , ldFee :: !(PlutusTx.Integer, PlutusTx.Integer) -- ^ The fee due for each swap, given as numerator and denominator of the actual rational proportion. , ldGYFeeFrom :: !PlutusTx.Integer -- ^ GeniusYield fees in terms of the "from" token contained in the position. , ldGYFeeTo :: !PlutusTx.Integer -- ^ GeniusYield fees in terms of the "to" token contained in the position. , ldMinFrom :: !PlutusTx.Integer -- ^ Minimal amount of "from" tokens that need to be swapped. } For a liquidity position to be considered valid, a number of conditions have to be met. Our offchain code will not consider invalid positions, and nobody else should either - all bets are off for invalid position. The validator must guarantee that no operation causes a position to become invalid. These condiftions are: The specified NFT must be present in the position. ldFrom and ldTo must be different. ldPrice must be positive. ldFee must not be negative. ldGYFeeFrom must not be negative, and the amount of "from" tokens in the position must be greater or equal than ldGYFeeFrom. ldGYFeeTo must not be negative, and the amount of "to" tokens in the position must be greater or equal than ldGYFeeTo. The following operations are supported: data LiquidityAction = Swap -- ^ Perform a swap. | CollectFees -- ^ Collect the GeniusYield fees. | Update -- ^ Update the parameters of the position. | Close -- ^ Close the position. These operations work as follows: Swap: A transaction consumes the position and recreates it with updated values and datum. Let us consider an example where the position contains 127 "from" tokens and 315 "to" tokens, where the price is (3,2) (i.e. 1.5), the fees are (1,20) (i.e. 5%) and the GeniusYield fees are 10%. Let us assume that ldGYFeeFrom is 10 and ldGYFeeTo is 5, i.e. 10 of the 127 "from" tokens belong to Geniusyield, as do 5 of the 315 "to" tokens. A user wants to swap 104 "from" tokens for "to" tokens. The fees are 5% of 104, i.e. 5.2, which are rounded up to 6 "from" tokens. Of those, 10% go to GeniusYield, which is - we round up again - one "from" token. 104 - 6 = 98 tokens remain for the swap. Using the price of 1.5, that gives us 147 "to" tokens2. So the user places 104 "from" tokens in the position and gets 147 "to" tokens out of it. The new position contains 127+104=231 "from" tokens (of which 10+1=11 belong to GeniusYield) and 315-147=168 "to" tokens (of which 5 belong to GeniusYield). The datum of the new UTxO is identical to the old datum with exception of the ldGYFeeFrom field, which is now 11 instead of 10. Let us now assume that another user wants to swap 77 "to" tokens. The fees are 5% of 77, which is 3.85, rounded up to 4. 10% of that, again rounded up, is one token for GeniusYield. Of the 77 "to" tokens, 77-4=73 remain for the swap. Seeing as we swap in the opposite direction, the price is now 2/3, which gives us 48.666... "from" tokens. These are rounded down to 48. The user therefore puts 77 "to" tokens in, leaving 168+77=245 "to" tokens, and takes 48 out, leaving 183. Of those, 11 "from" tokens belong to GeniusYield, as do 5+1=6 "to" tokens. The datum of the newest UTxO is again identical except for the ldGYFeeTo field, which is now 6. Users can swap arbitrary amounts, but they obviously can only get out what is there. So in our example, a user swapping "from" tokens can at most receive 245-6=239 "to" tokens, and a user swapping "to" tokens can at most receive 183-11=172 "from" tokens. She would probably refrain from swaps giving her less than the price dictates, but she is free to do so. All the validator checks is that the liquidity provider and GeniusYield get what is their due (and that the NFT remains in the position and that the datum is updated correctly). CollectFees: GeniusYield can submit a transaction to collect their fees. Such a transaction must be signed by GeniusYield, it must (at most) take out the amount specified by the ldGYFeeFrom and ldGYFeeTo fields, and it must update the datum to ldGYFeeFrom = ldGYFeeTo = 0. Update: The liquidity provider can update certain datum fields and take liquidity out or put new liquidity in, provided The NFT remains in the position. The GeniusYield fee remains in the position (and is not changed in the datum). The two assets "from" and "to" are not changed. The position remains valid (positive price, non-negative fees). Close: The liquidity provider can close the position, provided the NFT is burnt and GeniusYield is paid their fees[^3]. NFT The purpose of this minting policy is to guarantee that every token with this policy - independent of its token name - is an NFT, i.e. a non-fungible token, a token that can only exist at most once. The minting policy is not parameterized and thus statically known, which makes it easy to use in other contracts that are in need of an NFT. Once one has convinced oneself that each token with this policy is indeed an NFT, one can simply use this policy whenever an NFT is required. This is achieved by making use of a redeemer for the minting transaction. The redeemer specifies a UTxO (via its TxOutRef) that has to be consumed by the minting transaction. The minting policy checks that the specified UTxO is indeed consumed, that exactly one token is minted (whereas arbitrary burning is allowed) and that the token name of that one token is the hash of the TxOutRef of that UTxO. Since UTxO's are unique and can only exist once, this guarantees that only one token with this minting policy and token name can ever be minted (except, of course, if there is a hash collision, but we deem this risk to be negligible). Footnotes The ratio of fees going to GeniusYield is a global system parameter. ↩ If the amount of "to" tokens was not integral, we would round down to the next integer. ↩