Derivative-free optimisation in javascript

Very simple optimisation, using the Simplex (Nelder-Mead) method or Brent's method (for single-variable functions).

We provide two interfaces. In the first, you can dfoptim a function in a single go:

const point = dfoptim.fitSimplex(target, start);

which will look for the minimum of the vector-valued function target, starting from location start.

Running the optimisation may take a while, and no information can be retrieved while it runs, so we also provide a more stateful interface. The function above can be implemented as:

const opt = new dfoptim.Simplex(target, start)
while (!opt.step()) {
  // do something
}
const point = opt.result();

Where

• opt is our optimiser. At this point, it has done basic set up (creating the first simplex) but not taken any steps
• The step() method advances the algorithm one step, which will take one or two evaluations of the target function and may or may not find a better point than our current best. It returns true if we have converged.
• The result() method returns information about the best point.

The same pair of interfaces is provided for the Brent's method via dfoptim.fitBrent and dfoptim.Brent.

Example

Run

npm run build
npm run webpack

Then open example/index.html for a simple example.

Licence

MIT © Imperial College of Science, Technology and Medicine

Please note that this project is released with a Contributor Code of Conduct. By participating in this project you agree to abide by its terms.