Unable to use a function from an npm library ml-levenberg-marquardt

Community Help
1

I want to reproduce the example in the npm library ml-levenberg-marquardt in an Observable notebook. I run the following in a cell and a block:

fit_data = {
  let data = {
    x: [0, 1, 2],
    y: [1, 1, 1]
  }
  return data
}

{
    const LM = require('ml-levenberg-marquardt@2.1.1/lib/index.js').catch(() => window["_interopDefault"]);
    function sinFunction([a, b]) {
    return (t) => a * Math.sin(b * t);
  }

  const options = {
    damping: 1.5,
    gradientDifference: 10e-2,
    maxIterations: 100,
    errorTolerance: 10e-3
  };

  let fittedParams = LM(fit_data, sinFunction, options);
  return fittedParams
}

and I get the error message TypeError: LM is not a function. I could be importing this function completely incorrectly.

I am importing the library by following this guide.

2

Try adding ‘await’ to the require. Change

const LM = require('ml-levenberg-marquardt@2.1.1/lib/index.js').catch(() => window["_interopDefault"]);`

to

const LM = await require('ml-levenberg-marquardt@2.1.1/lib/index.js').catch(() => window["_interopDefault"]);`

or you could put the require in its own cell (outside the code block), since a cell has an implicit await. I hope that helps!

1 Like
3

@Cobus Thanks for the reply. Importing in a cell worked, but I am still unable to use the functions I want. On the guide it said that:

There may be multiple variables attached to the window that way, so for completeness, here are all of them: _interopDefault , isArray , mlMatrix , errorCalculation , gradientFunction , matrixFunction , step , levenbergMarquardt .

What does this mean, and how do I start using these functions?

4

Looking at the file your trying to “require” it has not been packaged correctly for browser usage: https://cdn.jsdelivr.net/npm/ml-levenberg-marquardt@2.1.1/lib/index.js

Typically that library would be run through a tool like RollupJS or WebPack which wraps it in a layer that is compatible with browser and NodeJS usage (and will also pull in dependencies as needed)

5

One way around this is to just extract the functions into individual cells in your notebook (instead of requiring it from npm).

Here is an example: https://observablehq.com/d/98f3b940a1389dc4

I would include the license from the NPM package in the notebook to make it clear that you copied it from there…

1 Like