Edit on GitHub

Robust32s.jl

Increased accuracy using Float32 data

How To Use This Package

There are two exported types: Robust32 and ComplexR32. Use them as you would Float32 and ComplexF32. It Just Works.

The Only Requirements

  • You are comfortable working with your data as Float32s.
  • You are comfortable reporting your results as Float32s.

Bringing in data

Your raw data rawdata is stored as Ints or as Float32s or as Float64s or other Real type.

data = Robust32.(rawdata);

Computing on the data

Just do it.

processed_data = your_computation(data)

Collecting results

Your processed data processed_data is stored as Robust32s. Report the results as Float32s.

results = Float32.(processed_data);