7.2 Rust array ecosystem

Explanation

The course default is simple and explicit:

  • Use slices for one-dimensional function boundaries.
  • Use Vec<f64> for owned one-dimensional numerical data.
  • Use tenferro typed tensors, currently tenferro_tensor::TypedTensor, for two-dimensional and higher arrays or tensors.

The Rust ecosystem also includes important alternatives. ndarray is a widely used N-dimensional array library. ndarray-linalg connects ndarray-style arrays to BLAS and LAPACK routines for established dense linear algebra workflows.

Choosing a library is a scientific-coding decision, not only a syntax decision. Check the current API, maintenance state, dependencies, memory layout, and the validation needs of the calculation. A library that is convenient for one operation may be a poor fit if it hides shapes, makes reproducibility harder, or brings in dependencies that are difficult to build on the target machine.

Nested vectors are not the normal representation for scientific arrays in this course. They make shape checks and memory layout less explicit than tensor or array types designed for numerical work.

Things to look up

  • ndarray
  • ndarray-linalg
  • BLAS
  • LAPACK
  • Contiguous memory
  • Crate maintenance

Exercise

Compare the tenferro-tensor and ndarray documentation for one basic array operation, such as creating a small two-dimensional array or checking its shape. Then inspect ndarray-linalg documentation for one BLAS/LAPACK-style operation, such as solving a linear system or calling a matrix factorization.

For each library, record:

  • Where you found the documentation.
  • What shape information the example makes visible.
  • Whether the operation needs optional features, backend crates, or external libraries.
  • One validation check you would add before using the result scientifically.

Notes for the exercise

  • Do not switch libraries only because an AI agent suggests one.
  • Prefer the course default unless the exercise or project has a clear reason to use another library.
  • For BLAS or LAPACK workflows, check both the Rust crate documentation and the system library requirements.