Julia Interoperability Guide — Calling Python, C, and Foreign Languages

In this tutorial, you will learn about Julia Interoperability Guide. We cover key concepts, practical examples, and best practices to help you master this topic.

Julia interoperability with PyCall enables calling Python libraries (NumPy, pandas, Scikit-Learn), ccall provides direct C function calling without wrappers, and RCall allows R integration -- making existing library ecosystems available from Julia with minimal overhead.

PyCall

using PyCall

# Import Python modules
np = pyimport("numpy")
pandas = pyimport("pandas")
sklearn = pyimport("sklearn.ensemble")

# Use NumPy
arr = np.array([1, 2, 3, 4, 5])
result = np.mean(arr)

# Use pandas
df = pandas.DataFrame(Dict(
    :name => ["Alice", "Bob"],
    :age  => [30, 25]
))

C Calls

# Call C standard library
# strlen
strlen = ccall((:strlen, "libc"), Csize_t, (Cstring,), "hello")
println(strlen)  # 5

# sin from math.h
result = ccall((:sin, "libm"), Cdouble, (Cdouble,), 1.0)

# Custom C library
# ccall((:my_function, "mylib"), Cint, (Cint, Cint), 1, 2)

RCall

using RCall

# Run R code
R"library(ggplot2)"
R"x <- c(1, 2, 3, 4, 5)"
R"y <- x^2"

# Transfer data
@rput df  # send Julia DataFrame to R
@rget result  # get R variable back

# Plot with ggplot
R"""
library(ggplot2)
p <- ggplot(df, aes(x=age)) + geom_histogram()
ggsave("histogram.png", p)
"""

CC++ Support

using Cxx

# Include and use C++ headers
cxx"""
#include <vector>
#include <string>
"""

# Create C++ objects
vec = icxx"std::vector<int>();"
push!(vec, 10)
push!(vec, 20)
println(vec[1])  # 10

Common Mistakes

1. Python package not found

PyCall uses the system Python. Use ENV["PYTHON"] to set path, or use Conda.jl: ] add Conda; Conda.add("numpy").

2. C type mismatches

ccall types must match exactly. Use Cint, Clong, Cdouble from Base for portable types.

3. Performance overhead

Cross-language calls have overhead. Batch operations (e.g., passing arrays) rather than calling per-element.

Practice Questions

1. How do you call a Python function from Julia? pyimport("module").function(args) or py"module.function(args)".

2. How do you call a C function? ccall((:function_name, "library"), ReturnType, (ArgTypes,), args).

3. How do you transfer data to R? @rput julia_variable sends to R. @rget r_variable brings back to Julia.

FAQ

{{< faq question="Is PyCall slow?" >}} Individual calls have overhead, but array data is passed without copying (via pointer). For batch operations, performance is good. {{< /faq >}}

{{< faq question="Can I use Python virtual environments?" >}} Yes. Set ENV["PYTHON"] = "/path/to/venv/bin/python" before using PyCall. Or use ] build PyCall to reconfigure. {{< /faq >}}

{{< faq question="Can I call Julia from Python?" >}} Yes. Use PyJulia (Python package) or the juliacall Python package to call Julia from Python. {{< /faq >}}

What's Next

Now learn about Julia best practices.