GAMS Transfer is a package to maintain GAMS data outside a GAMS script in a programming language like Python, Matlab, or R. It allows the user to add GAMS symbols (Sets, Aliases, Parameters, Variables and Equations), to manipulate GAMS symbols, as well as read/write symbols to different data endpoints. GAMS Transfer's main focus is the highly efficient transfer of data between GAMS and the target programming language, while keeping those operations as simple as possible for the user. In order to achieve this, symbol records - the actual and potentially large-scale data sets - are stored in native data structures of the corresponding programming languages. The benefits of this approach are threefold: (1) The user is usually very familiar with these data structures, (2) these data structures come with a large tool box for various data operations, and (3) optimized methods for reading from and writing to GAMS can transfer the data as a bulk - resulting in the high performance of this package.

Getting Started

Install

The user must download and install the latest version of GAMS in order to install GAMS Transfer R. GAMS Transfer R can then be installed from either the source package or from the binary package.

The installation from the source will install gamstransfer and all of its dependencies.

The users can also install gamstransfer without compiling the package source code using gamstransfer binary package. The binary packages are platform dependent and the instructions for each supported platform are shown below.

Windows

Linux

Mac OS X

GAMS Transfer R depends on packages R6, assertthat, R.utils, and Rcpp.

Examples

GDX Read

Reading in all symbols can be accomplished with one line of code (we reference data from the trnsport.gms example).

All symbol data is organized in the data field (a list) – m$data[<symbol_name>]$records – records are stored as data frames.

GDX Write

There are five symbol classes within GAMS Transfer R: Sets, Parameters, Variables, Equations, and Aliases. For purposes of this quick start, we show how to recreate the distance data structure from the trnsport.gms model (the parameter d). This brief example shows how users can achieve "GAMS-like" functionality, but within an R environment – GAMS Transfer R leverages the object oriented programming to simplify the syntax.

This example shows a few fundamental features of GAMS Transfer R:

1. A Container is analogous to a GDX file
2. Symbols will always be linked to a Container (notice that we always pass the Container reference m to the symbol constructor)
3. Records can be added to a symbol with the setRecords() method, through the records constructor argument (internally calls setRecords()), or through directly setting the records field. GAMS Transfer R will convert many common R data structures into a standard format.
4. Domain linking is possible by passing domain set objects to other symbols – this will also enable domain checking (violations will show up as <NA> as shown in this example)
5. Writing a GDX file can be accomplished in one line with the write() method.

Full Example

It is possible to use GAMS Transfer R to recreate the trnsport.gms results in GDX form. As part of this example, we also introduce the Container methods write() method (and generate new.gdx). We will discuss it in more detail in the following section: Data Exchange with GDX.

It can be observed from the above example that a typical work flow for writing using GAMS Transfer R is creating a container, filling it with symbols (Sets, Parameters, Variables, Equations, and Aliases), and write it to a GDX file. To read a GDX file, a Container can simply be initialized with the GDX file name as an argument.

These examples introduced the reader to the GAMS Transfer R syntax. In the remaining sections, we will present details about the core functionality and dig further into the syntax.

Manual

Container

Storing, manipulating, and transforming sparse data requires that it lives within an environment – this data can then be linked together to enable various operations. In GAMS Transfer R, we refer to this "environment" as the Container, it is the main repository for storing and linking our data. Linking data enables data operations like implicit set growth, domain checking, data format transformations (to dense/sparse matrix formats), etc –. A Container also enables different read/write operations.

Creating a Container is a simple matter of initializing an object. For example:

This new Container object, here called m, contains a number of convenient fields and methods that allow the user to interact with the symbols that are in the Container. Some of these methods are simply used to filter out different types of symbols, other methods are used to numerically characterize the data within each symbol.

Container fields

Symbols are organized in the Container under the data field. The $ notation (m$data) is used to access the underlying named list; symbols in this named list can then be retrieved with the standard bracket (or double bracket) notation (m$data[<symbol_name>]).

Container methods

Symbols

In GAMS Transfer R, a symbol is either a GAMS Set, Parameter, Variable, Equation, Alias. In GAMS Transfer R, a symbol cannot live on its own, but is always part of a container.

Create a symbol

There are two different ways to create a GAMS set and add it to a Container.

1. Use symbol constructor for Set, Parameter, Variable, Equation, Alias

1. Use the Container methods addSet, addParameter, addVariable, addEquation, addAlias (which internally calls the symbol constructor)
   m = Container$new()

   p = Parameter$new(m, "p")

Add Symbol Records

Three possibilities exist to assign symbol records:

1. Setting the argument records in the symbol constructor/Container method (internally calls setRecords)
2. Using the symbol method setRecords
3. Setting the field records directly

If the data is in a convenient format, a user may want to pass the records directly within the set constructor. This is an optional keyword argument and internally the symbol constructor will simply call the setRecords method. The symbol method setRecords is a convenience method that transforms the given data into an approved data frame format (see Standard Data Formats). Many native R data types can be easily transformed into data frames, so the setRecords method will accept a number of different types for input. The setRecords method is called internally on any data structure that is passed through the records argument.

The examples of setting records using the above-mentioned methods can be found in the respective documentation for each symbol ( Set, Parameter, Variable, Equation, and Alias ).

Set

Set Constructor

Set Fields

Set Methods

Adding Set Records

Three possibilities exist to assign symbol records to a set: We show a few examples of ways to create differently structured sets.

Example #1 - Create a 1D set from a vector

m = Container$new()

i = Set$new(m, "i", records = c("seattle", "san-diego"))

# NOTE: the above syntax is equivalent to -

# i = Set$new(m, "i")

# i$setRecords(c("seattle", "san-diego"))

# NOTE: the above syntax is also equivalent to -

# m$addSet("i", records= c("seattle", "san-diego"))

# NOTE: the above syntax is also equivalent to -

# i = m$addSet("i")

# i$setRecords(c("seattle", "san-diego"))

# NOTE: the above syntax is also equivalent to -

# m$addSet("i")

# m$data[["i"]]$setRecords(c("seattle", "san-diego"))

> i$records

uni_1 element_text

1 seattle

2 san-diego

Example #2 - Create a 2D set from a list

m = Container$new()

k = Set$new(m, "k", c("*", "*"), records=list("seattle", "san-diego"))

# NOTE: the above syntax is equivalent to -

# k = Set$new(m, "k", c("*", "*"))

# k$setRecords(list("seattle", "san-diego"))

# NOTE: the above syntax is also equivalent to -

# m$addSet("k", c("*","*"), records=list("seattle", "san-diego"))

# NOTE: the above syntax is also equivalent to -

# k = m$addSet("k", c("*","*"))

# k$setRecords(list("seattle", "san-diego"))

# NOTE: the above syntax is also equivalent to -

# m$addSet("k", c("*", "*"))

# m$data[["k"]]$setRecords(list("seattle", "san-diego"))

> k$records

uni_1 uni_2 element_text

1 seattle san-diego

Example #3 - Create a 1D set from a data frame slice

m = Container$new()

dist = data.frame(

from = c("seattle", "seattle", "seattle",

"san-diego", "san-diego", "san-diego"),

to = c("new-york", "chicago", "topeka",

"new-york", "chicago", "topeka"),

thousand_miles = c(2.5, 1.7, 1.8, 2.5, 1.8, 1.4)

)

l = Set$new(m, "l", records = unique(dist[["from"]]))

# NOTE: the above syntax is equivalent to -

# l = Set$new(m, "l")

# l$setRecords(unique(dist[["from"]]))

# NOTE: the above syntax is also equivalent to -

# m$addSet("l", records=unique(dist[["from"]]))

# NOTE: the above syntax is also equivalent to -

# l = m$addSet("l")

# l$setRecords(unique(dist[["from"]]))

# NOTE: the above syntax is also equivalent to -

# m$addSet("l")

# m$data[["l"]]$setRecords(unique(dist[["from"]]))

> l$records

uni_1 element_text

1 seattle

2 san-diego

Set element text is very handy when labeling specific set elements within a set. A user can add a set element text directly with a set element. Note that it is not required to label all set elements, as can be seen in the following example.

Example #4 - Add set element text

The primary advantage of the setRecords method is that GAMS Transfer R will convert many different (and convenient) data types into the standard data format (a data frame). Users that require higher performance will want to directly pass the Container a reference to a valid data frame, thereby skipping some of these computational steps. This places more burden on the user to pass the data in a valid standard form, but it speeds the records setting process.
In this section, we walk the user through an example of how to set records directly.

Example #5 - Directly set records (1D set)

Stepping through this example we take the following steps:

1. Create an empty Container
2. Create a GAMS set i in the Container, but do not set the records
3. Create a data frame (manually, in this example) taking care to follow the standard format
4. The data frame has the right shape and column labels so we can proceed to set the records.
5. We need to cast the uni_1 column as a factor, so we create a custom ordered category type using factor
6. Finally, we set the records directly by passing a reference to df_i into the symbol records attribute. The setter function of records checks that a data frame is being set, but does not check validity. Thus, as a final step, we call the $isValid() method to verify that the symbol is valid.

Note

Users can debug their data frames by running <symbol_name>$isValid(verbose=TRUE) to get feedback about their data.

Example #6 - Directly set records (1D subset)

This example is more subtle in that we want to create a set j that is a subset of i. We create the set i using the setRecords method but then set the records directly for j. There are two important details to note: 1) the column labels in df_j now reflect the standard format for a symbol with a domain set (as opposed to the universe) and 2) we create the factors by referencing the parent set (i) for the levels (instead of referencing itself).

Parameter

Parameter Constructor

Argument	Type	Description	Required	Default
container	Container	A reference to the Container object that the symbol is being added to	Yes	-
name	string	Name of symbol	Yes	-
domain	list	List of domains given either as a string ("*" for universe set) or as a reference to a Set object, an empty domain list will create a scalar parameter	No	NULL
records	many	Symbol records	No	NULL
domainForwarding	logical	Flag that forces set elements to be recursively included in all parent sets (i.e., implicit set growth)	No	FALSE
description	string	Description of symbol	No	""

Parameter Fields

Parameter Methods

Adding Parameter Records

Three possibilities exist to assign symbol records to a parameter: We show a few examples of ways to create differently structured parameters:

Example #1 - Create a GAMS scalar

m = Container$new()

pi = Parameter$new(m, "pi", records = 3.14159)

# NOTE: the above syntax is equivalent to -

# pi = Parameter$new(m, "pi")

# pi$setRecords(3.14159)

# NOTE: the above syntax is also equivalent to -

# m$addParameter("pi", records=3.14159)

# NOTE: the above syntax is also equivalent to -

# pi = m$addParameter("pi")

# pi$setRecords(3.14159)

# NOTE: the above syntax is also equivalent to -

# m$addParameter("pi")

# m$data[["pi"]]$setRecords(3.14159)

> pi$records

value

1 3.14159

Note

GAMS Transfer R will still convert scalar values to a standard format (i.e., a data frame with a single row and column).

Example #2 - Create a 2D parameter (defined over a set) from a

data frame slice

Example #3 - Create a 2D parameter from an array using setRecords

As with Sets, the primary advantage of the setRecords method is that GAMS Transfer will convert many different (and convenient) data types into the standard data format (data frame). Users that require higher performance will want to directly pass the Container a reference to a valid data frame, thereby skipping some of these computational steps. This places more burden on the user to pass the data in a valid standard form, but it speeds the records setting process. In this section, we walk the user through an example of how to set records directly.

Example #4 - Correctly set records (directly)

In this example, we create a large parameter (31,536,000 records and 8880 unique domain elements – we mimic data that is labeled for every second in one year) and assign it to a parameter with a$records. GAMS Transfer R requires that all domain columns must be ordered factors. The records setter function does very little work other than checking if the object being set is a data frame. This places more responsibility on the user to create a data frame that complies with the standard format. In Example #1 we take care to properly reference the factor from the domain sets – and in the end a$isValid() = TRUE.

Users will need to use the $isValid(verbose=TRUE) method to debug any structural issues. As an example, we incorrectly generate categorical data types by passing the data frame constructor the generic factor argument. This creates factor columns, but they are not ordered and they do not reference the underlying domain set. These errors result in a being invalid.

Example #5 - Incorrectly set records (directly)

Variable

Variable Constructor

Variable Fields

Variable Methods

Adding Variable Records

Three possibilities exist to assign symbol records to a variable: We show a few examples of ways to create differently structured variables:

Example #1 - Create a GAMS scalar variable

m = Container$new()

pi = Variable$new(m, "pi", records = data.frame(level = 3.14159))

# NOTE: the above syntax is equivalent to -

# pi = Variable$new(m, "pi", "free")

# pi$setRecords(data.frame(level = 3.14159))

# NOTE: the above syntax is also equivalent to -

# m$addVariable("pi", "free", records=data.frame(level = 3.14159))

> pi$records

level marginal lower upper scale

1 3.14159 0 -Inf Inf 1

Example #2 - Create a 2D positive variable, specifying no numerical data

Example #3 - Create a 2D variable (defined over a set) from a matrix

m = Container$new()

i = Set$new(m, "i", "*", records = paste0("i", 1:5))

j = Set$new(m, "j", "*", records = paste0("j", 1:5))

# creating records for parameter a

ij = list(i_1 = paste0("i", 1:5), j_2 = paste0("j", 1:5))

df = rev(expand.grid(rev(ij)))

df$value = 1:25

a = Parameter$new(m, "a", c(i, j), records = df)

# create a free variable and set the level and marginal attributes from matricies

v = Variable$new(m, "v", domain = c(i, j), records = list(level = a$toDense(), marginal = a$toDense()))

# if not specified, the toDense() method will convert the level values to a matrix

> v$toDense()

[,1] [,2] [,3] [,4] [,5]

[1,] 1 2 3 4 5

[2,] 6 7 8 9 10

[3,] 11 12 13 14 15

[4,] 16 17 18 19 20

[5,] 21 22 23 24 25

As with Sets, the primary advantage of the setRecords method is that GAMS Transfer will convert many different (and convenient) data types into the standard data format (data frame). Users that require higher performance will want to directly pass the Container a reference to a valid data frame, thereby skipping some of these computational steps. This places more burden on the user to pass the data in a valid standard form, but it speeds the records setting process. In this section, we walk the user through an example of how to set records directly.

Example #4 - Correctly set records (directly)

In this example, we create a large variable (31,536,000 records and 8880 unique domain elements – we mimic data that is labeled for every second in one year) and assign it to a variable with a$records. GAMS Transfer R requires that all domain columns must be ordered factors. The records setter function does very little work other than checking if the object being set is a data frame. This places more responsibility on the user to create a data frame that complies with the standard format. In Example #1 we take care to properly reference the factor from the domain sets – and in the end a$isValid() = TRUE.

As with Set and Parameters, users can use the $isValid(verbose=TRUE) method to debug any structural issues.

Equation

Equation Constructor

Equation Fields

Equation Methods

Adding Equation Records

Three possibilities exist to assign symbol records to an equation: We show a few examples of ways to create differently structured equations:

Example #1 - Create a GAMS scalar equation

m = Container$new()

pi = Equation$new(m, "pi", records = data.frame(level = 3.14159))

# NOTE: the above syntax is equivalent to -

# pi = Equation$new(m, "pi", "free")

# pi$setRecords(data.frame(level = 3.14159))

# NOTE: the above syntax is also equivalent to -

# m$addEquation("pi", "free", records=data.frame(level = 3.14159))

> pi$records

level marginal lower upper scale

1 3.14159 0 -Inf Inf 1

Example #2 - Create a 2D positive equation, specifying no numerical data

Example #3 - Create a 2D equation (defined over a set) from a matrix

m = Container$new()

i = Set$new(m, "i", "*", records = paste0("i", 1:5))

j = Set$new(m, "j", "*", records = paste0("j", 1:5))

# creating records for parameter a

ij = list(i_1 = paste0("i", 1:5), j_2 = paste0("j", 1:5))

df = rev(expand.grid(rev(ij)))

df$value = 1:25

a = Parameter$new(m, "a", c(i, j), records = df)

# create a free variable and set the level and marginal attributes from matrices

e = Equation$new(m, "e", "nonbinding", domain = c(i, j), records = list(level = a$toDense(), marginal = a$toDense()))

> e$records

i_1 j_2 level marginal lower upper scale

1 i1 j1 1 1 -Inf Inf 1

2 i2 j1 6 6 -Inf Inf 1

3 i3 j1 11 11 -Inf Inf 1

4 i4 j1 16 16 -Inf Inf 1

5 i5 j1 21 21 -Inf Inf 1

6 i1 j2 2 2 -Inf Inf 1

7 i2 j2 7 7 -Inf Inf 1

8 i3 j2 12 12 -Inf Inf 1

9 i4 j2 17 17 -Inf Inf 1

10 i5 j2 22 22 -Inf Inf 1

11 i1 j3 3 3 -Inf Inf 1

12 i2 j3 8 8 -Inf Inf 1

13 i3 j3 13 13 -Inf Inf 1

14 i4 j3 18 18 -Inf Inf 1

15 i5 j3 23 23 -Inf Inf 1

16 i1 j4 4 4 -Inf Inf 1

17 i2 j4 9 9 -Inf Inf 1

18 i3 j4 14 14 -Inf Inf 1

19 i4 j4 19 19 -Inf Inf 1

20 i5 j4 24 24 -Inf Inf 1

21 i1 j5 5 5 -Inf Inf 1

22 i2 j5 10 10 -Inf Inf 1

23 i3 j5 15 15 -Inf Inf 1

24 i4 j5 20 20 -Inf Inf 1

25 i5 j5 25 25 -Inf Inf 1

# if not specified, the toDense() method will convert the level values to a matrix

> e$toDense()

[,1] [,2] [,3] [,4] [,5]

[1,] 1 2 3 4 5

[2,] 6 7 8 9 10

[3,] 11 12 13 14 15

[4,] 16 17 18 19 20

[5,] 21 22 23 24 25

As with sets, parameters, and variables the primary advantage of the setRecords method is that GAMS Transfer will convert many different (and convenient) data types into the standard data format (data frame). Users that require higher performance will want to directly pass the Container a reference to a valid data frame, thereby skipping some of these computational steps. This places more burden on the user to pass the data in a valid standard form, but it speeds the records setting process. In this section, we walk the user through an example of how to set records directly.

Example #4 - Correctly set records (directly)

In this example, we create a large equation (31,536,000 records and 8880 unique domain elements – we mimic data that is labeled for every second in one year) and assign it to an equation with a$records. GAMS Transfer R requires that all domain columns must be ordered factors. The records setter function does very little work other than checking if the object being set is a data frame. This places more responsibility on the user to create a data frame that complies with the standard format. In Example #1 we take care to properly reference the factor from the domain sets – and in the end a$isValid() = TRUE.

As with sets, parameters, and variables, users can use the $isValid(verbose=TRUE) method to debug any structural issues.

Alias

Alias Constructor

Example - Creating an alias from a set

GAMS Transfer R only stores the reference to the parent set as part of the alias structure – most properties that are called from an alias object simply point to the properties of the parent set (with the exception of refContainer, name, and aliasWith). It is possible to create an alias from another alias object. In this case, a recursive search will be performed to find the root parent set – this is the set that will ultimately be stored as the aliasWith field. We can see this behavior in the following example:

Alias Fields

Alias Methods

Adding Alias Records

The linked structure of Aliases offers some unique opportunities to access some of the setter functionality of the parent set. Specifically, GAMS Transfer allows the user to change the domain, description, dimension, and records of the underlying parent set as a shorthand notation. We can see this behavior if we look at a modified Example #1 from R_GAMSTRANSFER_ADD_SET_RECORDS.

Example - Creating set records through an alias link

Note

An alias $isValid()=TRUE when the underlying parent set is also valid – if the parent set is removed from the Container the alias will no longer be valid.

Additional Features

Validating Data

GAMS Transfer R requires that the records for all symbols exist in a standard format (Standard Data Formats) in order for them to be understood and written successfully. It is certainly possible that the data could end up in a state that is inconsistent with the standard format (especially if setting symbol attributes directly). GAMS Transfer R includes the $isValid() method in order to determine if a symbol is valid and ready for writing; this method returns a logical. For example, we create two valid sets and then check them with $isValid() to be sure.

Note

It is possible to run $isValid() on both the Container as well as the symbol object – $isValid() will also return a logical if there are any invalid symbols in the Container object.

Example (valid data)

m = Container$new()

i = Set$new(m, "i", records = c("seattle", "san-diego", "washington_dc"))

j = Set$new(m, "j", i, records = c("san-diego", "washington_dc"))

> i$isValid()

[1] TRUE

> j$isValid()

[1] TRUE

> m$isValid()

[1] TRUE

Now we create some data that is invalid due to domain violations in the set j.

In this example, we know that the validity of the data is compromised by the domain violations, but there could be other subtle discrepancies that must be remedied before writing data. The user can get more detailed error reporting if the verbose argument is set to TRUE. For example:

The $isValid() method checks:

1. If the symbol belongs to a Container
2. If all domain set symbols exist in the Container
3. If all domain set symbols objects are valid
4. If records are a data frame (or None)
5. The shape of the records is congruent with the dimensionality of the symbol
6. If records column headings are in standard format
7. If all domain columns are factors and also ordered
8. If all domain factors are referenced properly ($records for referenced domain sets cannot be NULL in order to create categoricals properly)
9. If there are any domain violations
10. If there are any duplicate domain members
11. That all data columns are type numerical
12. To make sure that all domain categories are type string

Note

Calling $isValid() too often may have a significant impact on the performance.

Domain Forwarding

GAMS includes the ability to define sets directly from data using the implicit set notation (see: Implicit Set Definition (or: Domain Defining Symbol Declarations)). This notation has an analogue in GAMS Transfer R called domainForwarding.

Note

It is possible to recursively update a subset tree in GAMS Transfer R.

Domain forwarding is available as an argument to all symbol object constructors; the user would simply need to pass domainForwarding=TRUE.

In this example, we have raw data that is in the dist data frame, and we want to send the domain information into the i and j sets – we take care to pass the set objects as the domain for parameter c.

Note

The element order in the sets i and j mirrors that in the raw data.

In this example, we show that domain forwarding will also work recursively to update the entire set lineage – the domain forwarding occurs at the creation of every symbol object. The correct order of elements in set i is (z, a, b, c) because the records from j are forwarded first, and then the records from k are propagated through (back to i).

Describing Data

The methods describeSets, describeParameters, describeVariables, and describeEquations allow the user to get a summary view of key data statistics. The returned data frame aggregates the output for a number of other methods (depending on symbol type). A description of each Container method is provided in the following subsections:

describeSets

Returns: data frame

The following table includes a short description of the column headings in the return.

Example #1

m = Container$new("trnsport.gdx")

> m$describeSets()

name isAlias isSingleton domain domainType dim numberRecs cardinality

1 i FALSE FALSE * none 1 2 NA

2 j FALSE FALSE * none 1 3 NA

sparsity

1 NA

2 NA

describeParameters

Returns: data frame

The following table includes a short description of the column headings in the return.

describeVariables

Returns: data frame

The following table includes a short description of the column headings in the return.

describeEquations

Returns: data frame

The following table includes a short description of the column headings in the return.

describeAliases

Returns: data frame

The following table includes a short description of the column headings in the return.

Example #1

m = Container$new()

i = Set$new(m, "i", records = paste0("i",1:5))

j = Set$new(m, "j", records = paste0("j",1:10))

ip = Alias$new(m, "ip", i)

ipp = Alias$new(m, "ipp", ip)

jp = Alias$new(m, "jp", j)

> m$describeAliases()

name aliasWith isSingleton domain domainType dim numberRecs cardinality

1 ip i FALSE * none 1 5 NA

2 ipp i FALSE * none 1 5 NA

3 jp j FALSE * none 1 10 NA

sparsity

1 NA

2 NA

3 NA

Matrix Generation

GAMS Transfer R stores data in a "flat" format, that is, one record entry per data frame row. However, it is often necessary to convert this data format into a matrix/array format – GAMS Transfer R enables users to do this with relative ease using the toDense symbol methods This method will return a dense N-dimensional array (matrix for 2-Dimensions) with each dimension corresponding to the GAMS symbol dimension; it is possible to output an array up to 20 dimensions (a GAMS limit).

Example (1D data w/o domain linking (i.e., a relaxed domain))

Note that the parameter a is not linked to another symbol, so when converting to a matrix, the indexing is referenced to the data structure in a$records. Defining a sparse parameter a over a set i allows us to extract information from the i domain and construct a very different dense matrix, as the following example shows:

Example (2D data w/ domain linking)

The Universe Set

A Unique Element List (UEL) (aka the "universe" or "universe set") is an (i,s) pair where i is an identification number for a string s. GAMS uses UELs to efficiently store domain entries of a record by storing the UEL ID i of a domain entry instead of the actual string s. This avoids storing the same string multiple times. The concept of UELs also exists in R and is called a "factor". GAMS Transfer R leverages these types in order to efficiently store strings and enable domain checking within the R environment.

Each domain column in a data frame can be a factor, the effect is that each symbol maintains its own list of UELs per dimension. R lets the user choose if levels in a factor should be regarded as ordered. GAMS Transfer R relies exclusively on ordered factors (in order for a symbol to be valid it must have only ordered factors). By using ordered factors, GAMS Transfer R will order the UEL such that elements appear in the order in which they appeared in the data (which is how GAMS defines the UEL). GAMSTransfer allows the user to reorder the UELs with the uelPriority argument in the $write() method.

GAMS Transfer R does not keep track of the UEL separately from other symbols in the Container, it will be created internal to the $write() method and is based on the order in which data is added to the container. The user can access the current state of the UEL with the $getUniverseSet() container method. For example, we set a two dimensional set:

R also allows the user to modify (rename, reorder, etc.) factor levels. These methods may be useful for advanced users, but most users will probably find that modifying the original data structures and resetting the symbol records provides a simpler solution. The design of GAMS Transfer R should enable the user to quickly move data back and forth, without worrying about the deeper mechanics of factors.

Removing Symbols

Removing symbols from a Container is easy when using the removeSymbols container method; this method accepts either a string or a list of string.

Attention

Once a symbol has been removed, it is possible to have hanging references as domain links in other symbols. The user will need to repair these other symbols with the proper domain links in order to avoid validity errors.

Reordering Symbols

In order to write the contents of the Container, it is required that the symbols are sorted such that, for example, a Set used as a domain of another symbol appears before that symbol. The Container will try to establish a valid ordering when writing the data. This type of situation could be encountered if the user is adding and removing many symbols (and perhaps rewriting symbols with the same name) – users should attempt to only add symbols to a Container once, and care must be taken when naming. The method reorderSymbols attempts to fix symbol ordering problems. The following example shows how this can occur:

Example Symbol reordering

Since the link to i is broken, Set j is now invalid. The user has to manually set the domain again.

Now even though j is valid, the symbols are out of order. The order can be fixed as follows.

GAMS Special Values

The GAMS system contains five special values: UNDEF (undefined), NA (not available), EPS (epsilon), +INF (positive infinity), -INF (negative infinity). These special values must be mapped to their R equivalents. GAMS Transfer R follows the following convention to generate the 1:1 mapping:

• +INF is mapped to Inf
• -INF is mapped to -Inf
• EPS is mapped to -0.0 (mathematically identical to zero)
• NA is mapped to a NA
• UNDEF is mapped to NaN

GAMS Transfer R syntax is designed to quickly get data into a form that is usable in further analyses or visualization. The user does not need to remember these constants as they are provided within the class SpecialValues as SpecialValues$POSINF, SpecialValues$NEGINF, SpecialValues$EPS, SpecialValues[["NA"]], and SpecialValues$UNDEF. Some examples are shown below.

The following data frame for x would look like:

Note

The syntax SpecialValues$NA is not allowed in R. Therefore, to access NA, one has to use SpecialValues[["NA"]]. As shown in the example above, double bracket syntax works for other special values too.

Standard Data Formats

This section is meant to introduce the standard format that GAMS Transfer R expects for symbol records. It has already been mentioned that we store data as a data frame, but there is an assumed structure to the column headings and column types that will be important to understand. GAMS Transfer R includes convenience functions in order to ease the burden of converting data from a user-centric format to one that is understood by GAMS Transfer R. However, advanced users will want to convert their data first and add it directly to the Container.

Set Records Standard Format

All set records (including singleton sets) are stored as a data frame with n number of columns, where n is the dimensionality of the symbol + 1. The first n-1 columns include the domain elements while the last column includes the set element explanatory text. Records are organized such that there is one record per row.

The names of the domain columns follow a pattern of <domain_set_name>_<index_position>; a symbol dimension that is referenced to the universe is labeled uni_<index position>. The explanatory text column is called element_text and must take the last position in the data frame.

All domain columns must be factors and the element_text column must be a string type.

Some examples:

Parameter Records Standard Format

All parameter records (including scalars) are stored as a data frame with n number of columns, where n is the dimensionality of the symbol + 1. The first n-1 columns include the domain elements while the last column includes the numerical value of the records. Records are organized such that there is one record per row. Scalar parameters have zero dimension, therefore they only have one column and one row.

The names of the domain columns follow a pattern of <domain_set_name>_<index_position>; a symbol dimension that is referenced to the universe is labeled uni_<index_position>. The value column is called value and must take the last position in the data frame.

All domain columns must be factors and the value column must be a numeric type. GAMS Transfer R requires that all the factor levels are of type string.

Some examples:

Variable/Equation Records Standard Format

Variables and equations share the same standard data format. All records (including scalar variables/equations) are stored as a data frame with n number of columns, where n is the dimensionality of the symbol + 5. The first n-5 columns include the domain elements while the last five columns include the numerical values for different attributes of the records. Records are organized such that there is one record per row. Scalar variables/equations have zero dimension, therefore they have five columns and one row.

The names of the domain columns follow a pattern of <domain_set_name>_<index position>; a symbol dimension that is referenced to the universe is labeled uni_<index_position>. The attribute columns are called level, marginal, lower, upper, and scale. These attribute columns must appear in this order. Attributes that are not supplied by the user will be assigned the default GAMS values for that variable/equation type. It is possible to not pass any attributes, GAMS Transfer R would then simply assign default values to all attributes.

All domain columns must be factors and the attribute columns must be a numeric type. GAMS Transfer R requires that all the factor levels are of type string.

Some examples:

Data Exchange with GDX

Up until now, we have been focused on using GAMS Transfer R to create symbols in an empty Container using the symbol constructors (or their corresponding container methods). These tools will enable users to ingest data from many different formats and add them to a Container – however, it is also possible to read in symbol data directly from GDX files using the read container method. In the following sections, we will discuss this method in detail as well as the write method, which allows users to write out to new GDX files.

Reading from GDX

There are two main ways to read in GDX based data.

• Pass the file path directly to the Container constructor (will read all symbols and records)
• Pass the file path directly to the read method (default read all symbols, but can read partial files)

The first option here is provided for convenience and will, internally, call the read method. This method will read in all symbols as well as their records. This is the easiest and fastest way to get data out of a GDX file and into your R environment. For the following examples, we leverage the GDX output generated from the trnsport.gms model file.

A user could also read in data with the read method as shown in the following example.

Example (reading full data w/ read method)

It is also possible to read in a partial GDX file with the read method, as shown in the following example:

This syntax assumes that the user will always want to read in both the metadata as well as the actual data records, but it is possible to skip the reading of the records by passing the argument records=FALSE.

Attention

The read method attempts to link the domain objects together (in order to have a "regular" domainType) but if domain sets are not part of the read operation there is no choice but to default to a "relaxed" domain_type. This can be seen in the last example where we only read in the variable x and not the domain sets (i and j) that the variable is defined over. All the data will be available to the user, but domain checking is no longer possible. The symbol x will remain with "relaxed" domain type even if the user were to read in sets i and j in a second read call.

Writing to GDX

A user can write data to a GDX file by simply passing a file path (as a string). The write method will then create the GDX and write all data in the Container.

Note

It is not possible to write the Container when any of its symbols are invalid. If any symbols are invalid an error will be raised and the user will need to inspect the problematic symbols (perhaps using a combination of the listSymbols(isValid=FALSE) and isValid(verbose=TRUE) methods).

Example

m$write("path/to/file.gdx")

Example (write a compressed GDX file)

m$write("path/to/file.gdx", compress = TRUE)

Advanced users might want to specify an order to their UEL list (i.e., the universe set); recall that the UEL ordering follows that dictated by the data. As a convenience, it is possible to prepend the UEL list with a user specified order using the uelPriority argument.

Example (change the order of the UEL)

The original UEL order for this GDX file would have been c("a", "b", "c"), but this example reorders the UEL with uelPriority – the positions of b and c have been swapped. This can be verified with the gdxdump utility (using the uelTable argument):

gdxdump foo.gdx ueltable=foo

Set foo /
  'a' ,
  'c' ,
  'b' /;
$onEmpty

Set i(*) /
'a',
'c',
'b' /;

$offEmpty