Fix indexing error in particle resampling

[danscr]

xp has size n_observations, length(x). The resampling returns a vector of length n_observations, with each element contain length(x) indices. I think the index of choices in Line 24 should be i, not j, to avoid a BoundsError if length(x) > n_observations. But please double check that this is correct. It's just my best guess how to fix an error I kept getting for very long state vectors.

Not-so-MWE:

using LowLevelParticleFilters, MonteCarloMeasurements, Distributions

dynamics_noise = MvNormal(30,.1)
measurement_noise = MvNormal(30,.1)
rng = Random.Xoshiro()

function dynamics(x,u,p,t,noise=false)
  if noise
    return x .+ rand(rng,dynamics_noise)
  else
    return x
  end
end

function measurement(x,u,p,t,noise=false)
  if noise
    return x .+ rand(rng,measurement_noise)
  else
    return x
  end
end

function measurement_likelihood(x,u,y,p,t)
  logpdf(measurement_noise, x .- y)
end
state_init = MvNormal(30,.1)

apf = AdvancedParticleFilter(50,dynamics,measurement, measurement_likelihood,nothing,state_init)

sol = forward_trajectory(apf, zeros(5), [ones(30) for i=1:5], nothing)

Particles(sol.x,sol.we) #BoundsError

[baggepinnen]

Thanks for this fix, I believe you are correct :)