Show HN: Stochastic – JavaScript library for simulating stochastic processes

At line 116:

   if (sum != 1){
     return false;

where you're checking that the transition matrix sums to one across the rows, I think you can have problems due to finite precision. The sum can be epsilon different than unity. A reasonable check would be that abs(1-sum) < n*eps where eps is the floating-point precision (say, 4e-16).

In fact, a strict check like this can sometimes be annoying because you might save or transmit the values in a JSON format with less than full precision.

Shameless plug: if you're looking for something similar with a little more practical use cases: https://github.com/jprichardson/stochasm which is a fork of the great https://github.com/heydenberk/stochator

Consider also continuous time, second order stationary of known power spectra, say, for dimension integer n > 0.

For n > 1, consider Brillinger's book on time series.

Then, sure, bring along some FFT code.

Awesome, going to use it in my trading bot. Thanks!

I want to see some demos using this!

Nice work Nathan!