quant: Quantitative tests of pypomp

These quantitative tests, or simply quant tests, are designed to assess the accuracy and performance of pypomp for problems existing on a scale too large to be run on a laptop within the unit tests in pypomp:pypomp/tests.

The quant tests also provide additional examples of pypomp, focused on technical issues that extend the simpler examples in pypomp:tutorials.

---

🚀 Quick Links

• 💻 Core Repository — Source code and issue tracker for the pypomp package.
• 📖 Read the Documentation — Core library API reference and user guide.
• 🎓 Tutorials — Examples and tutorials for learning pypomp.

---

Quantitative Tests Index

Below is a list of quantitative test reports available in this repository:

1. SPX (S&P 500) Model

• Parameter Estimation & Convergence (tests/spx/estimation): Compares parameter estimation traces and log-likelihood estimates on the SPX index dataset using pypomp (CPU/GPU) and R’s pomp.
• Runtime & Performance Benchmark (tests/spx/timing): Benchmarks execution speed, speedup factors, and CPU/GPU throughput relative to R pomp.
• Fixed-Parameter Likelihood Validation (tests/spx/loglik): Evaluates particle filter log-likelihood estimates at fixed parameter values (rep3600) via two-sample KS test.
• Computational Scaling Benchmark (tests/spx/scaling): Evaluates execution speed and VRAM memory scaling across particle counts (\(J\)) and parallel search chains on GPU.

2. Dhaka Cholera Model

• IF2 vs IFAD (tests/dacca/algorithms): Compares how far iterated filtering alone and iterated filtering with gradient training get on the same likelihood for a comparable wall-clock budget.
• Runtime & Performance Benchmark (tests/dacca/timing): Benchmarks IF2 and particle filter execution speed in pypomp (CPU/GPU) against R pomp, on identical work.
• Fixed-Parameter Likelihood Validation (tests/dacca/loglik): Checks that the particle filter yields the same distribution of log-likelihoods in pypomp and pomp at the published MLE.

3. Random Number Generators

• Random Number Generators Benchmark & Comparison (tests/samplers): Benchmarks the execution speed and validates the statistical accuracy of pypomp’s fast approximate inverse CDF samplers (fast_poisson, fast_binomial, fast_gamma, fast_nbinomial) against jax.random and scipy.stats.

4. Measles Model

• Fixed-Parameter Likelihood Validation (tests/measles/loglik): Compares the distribution of particle filter log-likelihoods at the He et al. (2010) estimates in pypomp (32- and 64-bit) against R’s pomp.
• Parameter Estimation (tests/measles/estimation): Compares the distribution of IF2 parameter estimates from a global search in pypomp versus pomp, from identical starting points.
• Runtime & Throughput Benchmark (tests/measles/timing): Benchmarks IF2 and particle filter execution speed on the discrete measles model. Contrasts pypomp on GPU and CPU with R pomp, and the fast samplers in pypomp.random with stock JAX samplers.

5. Panel Measles Model

• Fixed-Parameter Likelihood Validation (tests/panel_measles/loglik): Compares the distribution of panel particle filter log-likelihoods at the He et al. (2010) estimates in pypomp against R’s panelPomp.
• Parameter Estimation (tests/panel_measles/estimation): Compares the distribution of block-IF2 parameter estimates from a global search in pypomp versus panelPomp, from identical starting points, and reports the pypomp-only mix-and-match step.
• Runtime & Throughput Benchmark (tests/panel_measles/timing): Benchmarks block-IF2 and particle filter execution speed on the 4-unit mixed panel model, contrasting pypomp on GPU and CPU with R panelPomp.