Particle Filter Estimator
Estimate a one-dimensional hidden state with a reproducible bootstrap particle filter.
Description
Estimate a changing scalar state from noisy observations with seeded sequential Monte Carlo resampling.
Particle Filter Estimator: Estimate a changing scalar state from noisy observations with seeded sequential Monte Carlo resampling.
When to use Particle Filter
Use this estimator to approximate a changing hidden state from sequential noisy observations when a nonlinear or non-Gaussian model makes simpler closed-form filtering unsuitable.
- Observations
- Required list input.
- Particles
- Required integer input.
- Process noise
- Required number input.
- Observation noise
- Required number input.
- Initial mean
- Required number input.
- Initial spread
- Required number input.
- Seed
- Required string input.
How Particle Filter works
Estimate a changing scalar state from noisy observations with seeded sequential Monte Carlo resampling. The tool evaluates the supplied inputs together and returns the named outputs below; it does not infer omitted operating conditions or change the units shown.1
- Estimate
- The resulting estimate returned as an object.
Limitations and assumptions
- Particle-filter quality depends on the state model, observation likelihood, proposal distribution, particle count, resampling rule, initialization, and numerical stability. Weight degeneracy and sample impoverishment can produce confident-looking but poor estimates.
- Use finite inputs in the displayed units and preserve more precision than the final presentation requires. Independently verify safety-critical, financial, compliance, or production decisions.
Alternative or Complementary approaches
Inspect effective sample size and repeated-run variability, test against simulated ground truth, and compare with Kalman-family or grid methods where their assumptions apply.
References
-
Particle filter — Wikipedia contributors
Similar or alternative tools
- Sequential Monte Carlo Gaussian Filter
Run a bootstrap particle filter for a one-dimensional Gaussian random walk observed with Gaussian noise; report weighted posterior means and resampled final particles.
- Finite-State Markov Chain Monte Carlo Sampler
Sample a finite-state Markov chain from a caller-supplied row-stochastic transition matrix and initial state. This samples the chain, not an arbitrary target distribution.
- Monte Carlo European Option Pricing Calculator
Estimate a European call or put price under geometric Brownian motion with antithetic normal pairs and report a Monte Carlo standard error. This is a model estimate, not a market quote.