import streamlit as st
from pathlib import Path

st.set_page_config(page_title="Calibration Benchmark", page_icon="🏠", layout="wide")

# Sidebar navigation
st.sidebar.title("Navigation")
st.sidebar.page_link("streamlit_app.py", label="Home", icon="🏠")
st.sidebar.page_link("pages/OptimizationLeaderboard.py", label="Optimization Leaderboard", icon="📊")
st.sidebar.page_link("pages/UQLeaderboard.py", label="UQ Leaderboard", icon="🎯")
st.sidebar.page_link("pages/MethodDetails.py", label="Methods", icon="📘")
st.sidebar.page_link("pages/RawData.py", label="Get Data", icon="🧾")

st.title("Calibration Benchmark")
st.markdown(
    "A benchmark comparing parameter-calibration methods on chaotic dynamical systems. "
    "Methods are ranked by **forward-model run efficiency** — how many forward-model "
    "evaluations are needed, on average across random seeds, to reach a target accuracy."
)

st.divider()

col1, col2 = st.columns(2, gap="large")

with col1:
    st.subheader("📊 Optimization Leaderboard")
    st.markdown(
        "Ranks methods by mean forward-model runs to reach an **RMSE target** on "
        "Lorenz-63 and Lorenz-96 benchmarks. "
        "Lower is better; failed runs are tracked separately as a failure rate."
    )
    st.page_link("pages/OptimizationLeaderboard.py", label="Go to Optimization Leaderboard →")

with col2:
    st.subheader("🎯 UQ Leaderboard")
    st.markdown(
        "Ranks methods by mean forward-model runs to reach an **uncertainty quantification "
        "target**. Same metric and benchmarks as the Optimization Leaderboard, evaluated "
        "at a UQ-specific convergence criterion."
    )
    st.page_link("pages/UQLeaderboard.py", label="Go to UQ Leaderboard →")

st.divider()

st.subheader("Benchmarks")
st.markdown(
    "Results are reported on four benchmark configurations of the [Lorenz system]"
    "(https://en.wikipedia.org/wiki/Lorenz_system), a standard testbed for "
    "data-assimilation and calibration algorithms:"
)

_media = Path(__file__).parent / "media"

st.markdown("- **L63** — Lorenz-63 — 3-variable chaotic attractor; learn 2 parameters, strongly nonlinear.")
st.image(
    str(_media / "posterior_ribbons_20_13_k5.png"),
    caption="Example prior-posterior & truth. L: Difference to true parameter. R: data-sample/output predictived distribution (state-mean [1:3],  state-covariance (diag [4:6] and off-diag [7:9]))",
    width=800,
)


st.markdown("- **L96** — Lorenz-96 (40-variable); learn 1-parameter constant forcing.")
st.image(
    str(_media / "posterior_ribbons_const-force_12_1_k3.png"),
    caption="Example prior-posterior & truth. L: Difference to true parameter. C: parameter-induced forcing of the L96 system. R: data-sample/output predictived distribution ([1:40] state-mean [41:80] state-std)",
    width=1200,
)

st.markdown("- **L96_SPATIAL_FORCING** — Lorenz-96 (40-variable) with spatially-varying forcing; learn 40 parameters; moderately correlated prior.")
st.image(
    str(_media / "posterior_ribbons_vec-force_65_1_k3.png"),
    caption="Example prior-posterior & truth. L: Difference to true parameters. C: parameter-induced forcing of the L96 system. R: data-sample/output predictived distribution ([1:40] state-mean [41:80] state-std)",
    width=1200,
)


st.markdown("- **L96_NN_FORCING** — Lorenz-96 (100-variable) with a neural-network forcing; Learn 61 parameters (weights and biases of the network). Reasonable prior given.")
st.image(
    str(_media / "posterior_ribbons_flux-force_80_1_k3.png"),
    caption="Example prior-posterior & truth. L: Difference to true parameters (weights). C: parameter-induced forcing of the L96 system. R: data-sample/output predictived distribution ([1:100] state-mean [101:200] state-std)",
    width=1200,
)

st.subheader("Method taxonomy")
st.markdown(
    "Every method carries four independent tags — how it searches, what update "
    "mechanism drives each step, what it's built to report, and whether/when it uses "
    "a surrogate model. See the **📘 Methods** page for citations and per-method "
    "performance charts."
)
st.markdown(
    """
- **Parallelism** — how the search explores parameter space
    - **Serial** — `ADAM`, `LM` — a single point estimate advanced step by step.
    - **Parallel-independent** — `ABC`, `HM` — a population of candidates updated
      with no coupling between members (accepted samples / per-wave resampling).
    - **Parallel-interacting** — `TEKI`, `ETKI`, `IEKF`, `UKI`, `CES-EKI-DMC`, `CES-EKI-CONST`,
      `CES-IEKF-CONST` — an ensemble whose members are coupled through a shared update
      each iteration.
- **Update type** — the mechanism driving each update step
    - **Gradient** — `ADAM`, `LM` — follow the loss gradient (or a Gauss-Newton
      approximation of it) directly.
    - **Kalman** — `TEKI`, `ETKI`, `IEKF`, `UKI`, `CES-EKI-DMC`, `CES-EKI-CONST`,
      `CES-IEKF-CONST` — a (possibly linearized or unscented) Kalman-style ensemble update.
    - **General** — `ABC`, `HM` — neither gradient- nor Kalman-based (rejection
      sampling, implausibility cuts).
- **Method goal** — what the method is built to report
    - **Optimization** — `TEKI`, `ETKI`, `UKI`, `ADAM`, `LM` — a single best-fit
      parameter estimate.
    - **UQ** — `IEKF`, `ABC`, `HM`, `CES-EKI-DMC`, `CES-EKI-CONST`, `CES-IEKF-CONST` — the
      full posterior / parameter uncertainty. Can still be scored on the Optimization
      leaderboard, but tends to be less competitive there since it's not optimizing for
      speed-to-target.
- **Emulator use** — when/whether a surrogate model of the forward model is used
    - **None** — `TEKI`, `ETKI`, `IEKF`, `UKI`, `ADAM`, `LM`, `ABC` — samples/evaluates
      the true forward model throughout.
    - **Within-optimize** — `HM` — refits a surrogate at each iteration (wave) of the
      search itself.
    - **After-optimize** — `CES-EKI-DMC`, `CES-EKI-CONST`, `CES-IEKF-CONST` — fits a
      surrogate (e.g. a GP) once, after calibration finishes, and samples the posterior
      through it.
"""
)
st.caption(
    "Note: Kalman methods are Bayesian in spirit too (they're approximate Gaussian "
    "posterior updates) — update type is about mechanism (gradient vs. Kalman vs. "
    "general), not whether a method is 'Bayesian'."
)

st.subheader("Key metric")
st.markdown(
    "The reported metric is the **mean number of forward-model evaluations** required "
    "to reach the target, averaged over random seeds. "
    "A value of **−1** indicates a failed run (did not reach the target); "
    "the **failure rate** shows the fraction of seeds that failed."
)