GDXInterface.jl

High-level Julia interface for reading and writing GDX files (GAMS Data Exchange).

Uses the gdx_jll package to provide the GDX C library independently of GAMS.

No GAMS installation required.

For more information on the GDX file format, see the blog post 'GDX source code published on GitHub'.

Installation

using Pkg
Pkg.add(url="https://github.com/jd-foster/GDXInterface.jl.git")

Or in the Pkg REPL:

pkg> add https://github.com/jd-foster/GDXInterface.jl.git

Run tests with:

pkg> test GDXInterface

These instructions will be updated if/when the package is registered.

Quick Start

Reading GDX files

using GDXInterface

gdx = read_gdx("transport.gdx")

# List symbols by type
list_sets(gdx)
list_parameters(gdx)
list_variables(gdx)
list_equations(gdx)

# Access data as DataFrames
demand = gdx[:demand]       # bracket syntax
demand = gdx.demand         # property syntax (with tab completion)

# Access the full symbol object (includes name, description, domain)
sym = get_symbol(gdx, :demand)
sym.name         # "demand"
sym.description  # explanatory text from GAMS
sym.domain       # ["j"]
sym.records      # the DataFrame

Writing GDX files

using GDXInterface, DataFrames

# Write DataFrames as parameters
supply = DataFrame(i = ["seattle", "san-diego"], value = [350.0, 600.0])
demand = DataFrame(j = ["new-york", "chicago", "topeka"], value = [325.0, 300.0, 275.0])
write_gdx("output.gdx", "supply" => supply, "demand" => demand)

# Round-trip: read a GDX file and write it back (preserves all symbol types)
gdx = read_gdx("model.gdx")
write_gdx("copy.gdx", gdx)

Selective reading

For large GDX files, load only the symbols you need:

gdx = read_gdx("big_model.gdx", only=[:x, :demand])

API Reference

Types

Type	Description
GDXFile	Container for all symbols in a GDX file
GDXSet	GAMS set (elements + explanatory text)
GDXParameter	GAMS parameter (domain elements + values)
GDXVariable	GAMS variable (level, marginal, lower, upper, scale)
GDXEquation	GAMS equation (level, marginal, lower, upper, scale)

Reading

read_gdx(filepath; parse_integers=true, only=nothing) -> GDXFile

• parse_integers: convert set elements like "2020" to Int
• only: vector of symbol names to load (e.g. [:x, :demand])

Writing

# Write DataFrames as parameters (convenience)
write_gdx(filepath, "name" => DataFrame, ...)

# Write a full GDXFile (sets, parameters, variables, equations)
write_gdx(filepath, gdxfile::GDXFile)

Querying a GDXFile

gdx[:name]               # records DataFrame (bracket access)
gdx.name                 # records DataFrame (property access)
get_symbol(gdx, :name)   # full GDXSymbol object

list_sets(gdx)            # list set names
list_parameters(gdx)      # list parameter names
list_variables(gdx)       # list variable names
list_equations(gdx)       # list equation names
list_symbols(gdx)         # list all symbol names

haskey(gdx, :name)        # check if symbol exists
length(gdx)              # number of symbols
for (k, v) in gdx ...    # iterate over symbols

Special Values

GAMS special values are mapped to Julia equivalents when reading:

GAMS	Julia	Notes
UNDEF	NaN	Undefined value
NA	NaN	Not available
+INF	Inf	Positive infinity
-INF	-Inf	Negative infinity
EPS	-0.0	"Explicitly zero" in sparse data

When writing, NaN maps to GAMS NA, Inf/-Inf map to +INF/-INF, and -0.0 maps back to GAMS EPS. This preserves EPS semantics through round-trips. Regular 0.0 stays as a normal zero.

Acknowledgments

Derived from GDX file access functionality originally developed for GAMS.jl.