Be advised that the analysis takes several weeks of computing time, upwards of 50 GB of available RAM, and about 400 GB of storage space to execute. The scripts call [this](https://gist.githubusercontent.com/jeffwong/5925000/raw/bf02ed0dd2963169a91664be02fb18e45c4d1e20/sqlitewritetable.R) external script file and downloads [this](https://github.com/Rucknium/CashFusionStats/raw/main/data/fusions_df.rds) R data file.
The main analysis is done with the R statistical programming language. R itself can be downloaded [here](https://cloud.r-project.org/). [RStudio](https://www.rstudio.com/products/rstudio/download/#download) is a good IDE for R. Install the necessary R packages with:
Linux users may have difficulties installing the required packages due to external dependencies (as they must be compiled from source). Ubuntu users will require ```build-essential```, ```libcurl4-openssl-dev``` and ```libgmp3-dev```. Please raise an issue here if you encounter any further issues.
You must have a Bitcoin Cash (BCH) full node synced with the transaction index enabled with the `-txindex` flag. As of now, the analysis has been tested with the [Bitcoin Unlimited](https://www.bitcoinunlimited.info/) node implementation.
The [R/extract-tx-graphs.R](R/extract-tx-graphs.R) script file issues JSON-RPC queries to `bitcoind`, the Bitcoin Cash node daemon. Make sure `bitcoind` is running before initiating this script.
In the R script file you must set `bitcoin.conf.file` to the filepath of your bitcoin.conf file and `data.dir` to the directory where you want files to be stored. Note that the `data.dir` in the R file should be a different directory than the `bitcoind` data dir where the BCH blockchain is stored. Your bitcoin.conf file should contain these lines at a minimum:
Input \<USER\> and \<PASSWORD\> of your choice. Perform a quick sanity check to ensure you can access your node via JSON-RPC. The following should return data, and not refuse your connection: (user being the value of rpcuser / pass the value of rpcpass)
The script spawns multiple R process threads to accelerate queries to `bitcoind` and will take several hours to execute. In the specified data directory, a set of files named `tx_graph_height_BEGIN _to_END.rds` will be created.
The R object format of these files is:
```
List of 2
$ incoming: Classes 'data.table and 'data.frame': N obs. of 3 variables:
..$ txid : chr [txid of the transaction]
..$ origin.txid : chr [txid of the output that is being spent by the transaction]
..$ origin.position : num [the position (of the output being spent) within the output's transaction, indexed from one]
$ outgoing: Classes 'data.table and 'data.frame': N obs. of 4 variables:
..$ txid : chr [txid of the transaction]
..$ address : chr [address that the transaction is being spent to]
..$ position : int [position of the output being produced by the transaction, indexed from one]
..$ value : num [quantity of BCH being spent to each address]
```
Note that, unlike the underlying blockchain data, the position of outputs is indexed from one, not from zero.
The [R/construct-edgelist.R](R/construct-edgelist.R) script file produces a SQL database that contains the [edge list](https://en.wikipedia.org/wiki/Edge_list) of the BCH transaction [graph](https://en.wikipedia.org/wiki/Graph_(discrete_mathematics)). Set `data.dir` to the same directory same as you used in `extract-tx-graphs.R`.
The script assigns an integer index to every output. Converting the transaction ID + output position to integer indixes is necessary so that the transaction graph can be stored in RAM. This script is single-threaded so as to avoid conflicts when writing to the database. It should take a few hours to complete.
The [R/get-coinbases.R](R/get-coinbases.R) script file produces a `.rds` file that contains the coinbase outputs. Coinbases are special vertices in the transaction graph since they have no inputs. Set `data.dir` to the same directory same as you used in `extract-tx-graphs.R`.
The [R/determine-descendants.R](R/determine-descendants.R) script file uses the `igraph` package to determine which outputs in the unspent transaction output (UTXO) set can be traced back to an earlier CashFusion transaction. The dataset produced by the [CashFusionStats](https://github.com/Rucknium/CashFusionStats) repository is used to identify the CashFusion transactions.
Set `data.dir` as before. These operations are time-consuming and may take weeks. There is an option to interrupt the process and restart it.
The [R/descendant-statistics.R](R/descendant-statistics.R) script file merges the CashFusion descendants, which has been identified by integer indices, with their transaction ID identifiers. It then calculates a simple total of proportion of the UTXO set that is a CashFusion descendant as well as the value in BCH terms. A `CashFusion-Descendants.csv` csv file is output with the following columns: