docs: replace ASCII diagrams with Mermaid visuals + brand CSS

- Add sphinxcontrib-mermaid>=0.9.2 to requirements.txt
- Add custom.css with orange brand theme (admonitions, tables, mermaid containers)
- Replace Complete Algorithm ASCII pseudocode with Mermaid flowchart (differential_evolution.md)
- Add Mermaid stateDiagram-v2 for Bull/Normal/Bear regime transitions (hmm.md)
- Replace Rust module file tree with Mermaid graph TD (point_processes.md)
- Configure mermaid_version=10.9.0 and dark+orange themeVariables in conf.py

docs/requirements.txt

@@ -4,3 +4,4 @@ furo==2023.9.10
 myst-parser==2.0.0
 sphinx-autodoc-typehints==1.24.0
 charset-normalizer>=3.4.0
+sphinxcontrib-mermaid>=0.9.2

docs/source/_static/custom.css

@@ -0,0 +1,138 @@
+/* ─── Optimiz-rs Documentation Custom Styles ──────────────────────────────── */
+/* Brand: orange (#f97316) on furo dark theme */
+
+:root {
+    --brand-orange: #f97316;
+    --brand-orange-light: #fb923c;
+    --brand-orange-dim: rgba(249, 115, 22, 0.15);
+    --brand-orange-border: rgba(249, 115, 22, 0.35);
+}
+
+/* ── Mermaid diagram container ─────────────────────────────────────────────── */
+.mermaid {
+    background: transparent !important;
+    padding: 1.5rem 0;
+    text-align: center;
+    overflow-x: auto;
+}
+
+.mermaid svg {
+    max-width: 100%;
+    border-radius: 12px;
+    filter: drop-shadow(0 4px 16px rgba(249, 115, 22, 0.12));
+}
+
+/* ── Better admonitions ─────────────────────────────────────────────────────── */
+.admonition {
+    border-radius: 10px !important;
+    border-left-width: 4px !important;
+    margin: 1.5rem 0 !important;
+}
+
+.admonition.note {
+    border-left-color: var(--brand-orange) !important;
+    background: var(--brand-orange-dim) !important;
+}
+
+.admonition.tip {
+    border-left-color: #22c55e !important;
+    background: rgba(34, 197, 94, 0.1) !important;
+}
+
+.admonition.warning {
+    border-left-color: #f59e0b !important;
+    background: rgba(245, 158, 11, 0.1) !important;
+}
+
+.admonition.important {
+    border-left-color: #ef4444 !important;
+    background: rgba(239, 68, 68, 0.1) !important;
+}
+
+/* ── Section headers ────────────────────────────────────────────────────────── */
+h1, h2, h3 {
+    scroll-margin-top: 4rem;
+}
+
+h2 {
+    border-bottom: 2px solid var(--brand-orange-border);
+    padding-bottom: 0.4rem;
+}
+
+/* ── Code blocks ────────────────────────────────────────────────────────────── */
+.highlight {
+    border-radius: 8px !important;
+    border: 1px solid rgba(255, 255, 255, 0.08) !important;
+}
+
+/* Inline code */
+code.literal {
+    background: rgba(249, 115, 22, 0.08) !important;
+    border: 1px solid var(--brand-orange-border) !important;
+    padding: 0.1em 0.4em !important;
+    border-radius: 4px !important;
+    color: var(--brand-orange-light) !important;
+    font-size: 0.88em !important;
+}
+
+/* ── Tables ─────────────────────────────────────────────────────────────────── */
+table.docutils {
+    border-collapse: collapse;
+    width: 100%;
+    margin: 1.2rem 0;
+    border-radius: 8px;
+    overflow: hidden;
+    font-size: 0.9rem;
+}
+
+table.docutils th {
+    background: var(--brand-orange-dim) !important;
+    color: var(--brand-orange-light) !important;
+    font-weight: 700;
+    padding: 0.6rem 0.9rem;
+    border-bottom: 2px solid var(--brand-orange-border);
+}
+
+table.docutils td {
+    padding: 0.5rem 0.9rem;
+    border-bottom: 1px solid rgba(255, 255, 255, 0.06);
+}
+
+table.docutils tr:hover td {
+    background: rgba(249, 115, 22, 0.04);
+}
+
+/* ── Performance metric boxes (for benchmarks) ──────────────────────────────── */
+.perf-box {
+    background: linear-gradient(135deg, rgba(249,115,22,0.12) 0%, rgba(249,115,22,0.04) 100%);
+    border: 1px solid var(--brand-orange-border);
+    border-radius: 10px;
+    padding: 1rem 1.5rem;
+    margin: 1rem 0;
+    display: inline-block;
+    min-width: 140px;
+    text-align: center;
+}
+
+.perf-box .val {
+    font-size: 1.8rem;
+    font-weight: 800;
+    color: var(--brand-orange);
+}
+
+.perf-box .lbl {
+    font-size: 0.75rem;
+    color: #94a3b8;
+    margin-top: 0.2rem;
+}
+
+/* ── Navigation sidebar ─────────────────────────────────────────────────────── */
+.sidebar-tree .current > .reference {
+    color: var(--brand-orange) !important;
+}
+
+/* ── Mobile ─────────────────────────────────────────────────────────────────── */
+@media (max-width: 768px) {
+    .mermaid svg { width: 100% !important; }
+    table.docutils { font-size: 0.78rem; }
+}

docs/source/algorithms/differential_evolution.md

@@ -169,46 +169,18 @@ $$
 
 ## Complete Algorithm
 
-```
+```{mermaid}
-Algorithm: Differential Evolution
+flowchart TD
-─────────────────────────────────
+    A["🎲 Initialize Population\nx_i = l + rand · (u − l)"] --> B["📊 Evaluate Fitness\nf_i = f(x_i) for all i"]
-Input: objective f, bounds [l, u], pop_size N_P, F, CR, max_iter
+    B --> C{{"g < max_iter?"}}
-
+    C -->|Yes| D["Mutation\nv = x_r1 + F · (x_r2 − x_r3)"]
-1. Initialize population:
+    D --> E["Crossover\nu_j = v_j if rand ≤ CR or j = j_rand\nelse u_j = x_j"]
-   For i = 1 to N_P:
+    E --> F["Clip to bounds [l, u]"]
-       x_{i,0} = l + rand(0,1) · (u - l)    # uniform in bounds
+    F --> G{{"f(trial) ≤ f(target)?"}}
-   
+    G -->|"Yes — better"| H["✅ Accept trial\nx_i ← u_i"]
-2. Evaluate fitness:
+    G -->|"No — worse"| I["Keep current\nx_i unchanged"]
-   f_i = f(x_{i,0}) for all i
+    H & I --> C
-
+    C -->|No| J["🏆 Return x_best, f(x_best)"]
-3. While g < max_iter and not converged:
-   
-   a. For i = 1 to N_P:
-      
-      i. Mutation:
-         Select r_1, r_2, r_3 distinct and ≠ i
-         v_{i,g+1} = x_{r_1,g} + F · (x_{r_2,g} - x_{r_3,g})
-      
-      ii. Crossover:
-          j_rand = randint(1, D)
-          For j = 1 to D:
-              if rand(0,1) ≤ CR or j = j_rand:
-                  u_{i,j,g+1} = v_{i,j,g+1}
-              else:
-                  u_{i,j,g+1} = x_{i,j,g}
-      
-      iii. Boundary handling:
-           Clip u_{i,g+1} to [l, u]
-      
-      iv. Selection:
-          if f(u_{i,g+1}) ≤ f(x_{i,g}):
-              x_{i,g+1} = u_{i,g+1}
-          else:
-              x_{i,g+1} = x_{i,g}
-   
-   b. g = g + 1
-
-4. Return x_best and f(x_best)
 ```
 
 ---

docs/source/algorithms/hmm.md

@@ -458,6 +458,24 @@ For complete examples with regime detection on real market data, see the [HMM Tu
 
 ### Transition Diagram
 
+```{mermaid}
+stateDiagram-v2
+    direction LR
+    [*] --> Normal
+    Bull : 📈 Bull Market
+    Normal : 📊 Normal Regime
+    Bear : 📉 Bear Market
+    Bull --> Bull : a₁₁ (self)
+    Bull --> Normal : a₁₂
+    Normal --> Bull : a₂₁
+    Normal --> Normal : a₂₂ (self)
+    Normal --> Bear : a₂₃
+    Bear --> Normal : a₃₂
+    Bear --> Bear : a₃₃ (self)
+```
+
+**Code example** — build and visualize the transition graph programmatically:
+
 ```python
 import networkx as nx
 

docs/source/api/point_processes.md

@@ -406,25 +406,32 @@ plt.show()
 
 ## Module Architecture
 
-```
+```{mermaid}
-optimizr/point_processes/
+graph TD
-├── mod.rs              # Module root, public API re-exports
+    MOD["📦 mod.rs\nPublic API re-exports"]
-├── kernels.rs          # ExcitationKernel trait + implementations
+    K["🔧 kernels.rs\nExcitationKernel trait"]
-│   ├── ExponentialKernel    (φ = αe^{-βt})
+    H["⚡ hawkes.rs\nSimulation & fitting"]
-│   ├── PowerLawKernel       (φ = K₀(1+t)^{-1-α₀})
+    ML["🔢 mittag_leffler.rs\nSpecial functions"]
-│   └── CompletelyMonotoneKernel (Mittag-Leffler)
+    FBM["〰️ mixed_fbm.rs\nFractional Brownian motion"]
-├── hawkes.rs           # Hawkes process simulation & fitting
+    PY["🐍 python_bindings.rs\nPyO3 bindings"]
-│   ├── HawkesProcess<K>     (univariate, Ogata thinning)
+
-│   └── BivariateHawkes<K>   (buy/sell reaction flow)
+    MOD --> K
-├── mittag_leffler.rs   # Special functions
+    MOD --> H
-│   ├── mittag_leffler()     (E_{α,β}(z))
+    MOD --> ML
-│   ├── f_alpha_lambda()     (Theorem 3.1 scaling fn)
+    MOD --> FBM
-│   ├── gamma()              (Lanczos Γ function)
+    MOD --> PY
-│   └── incomplete_gamma*()  (upper/lower)
+
-├── mixed_fbm.rs        # Fractional Brownian motion
+    K --> EK["ExponentialKernel\nφ = α·e^−βt"]
-│   ├── FractionalBM         (Cholesky & Hosking simulation)
+    K --> PLK["PowerLawKernel\nφ = K₀(1+t)^−1−α"]
-│   └── MixedFractionalBM    (a·B + b·B^H)
+    K --> CMK["CompletelyMonotoneKernel\nMittag-Leffler"]
-└── python_bindings.rs  # PyO3 bindings for all functions
+
+    H --> HP["HawkesProcess K\nunivariate · Ogata thinning"]
+    H --> BH["BivariateHawkes K\nbuy/sell reaction flow"]
+
+    ML --> mf["mittag_leffler · f_alpha_lambda\ngamma · incomplete_gamma"]
+
+    FBM --> fbm1["FractionalBM\nCholesky & Hosking"]
+    FBM --> fbm2["MixedFractionalBM\na·B + b·B^H"]
 ```
 
 ---

docs/source/conf.py

@@ -20,6 +20,7 @@ extensions = [
     'sphinx.ext.intersphinx',
     'sphinx.ext.mathjax',
     'myst_parser',
+    'sphinxcontrib.mermaid',
 ]
 
 # Add any paths that contain templates here, relative to this directory.
@@ -76,3 +77,10 @@ myst_enable_extensions = [
     "deflist",
     "dollarmath",
 ]
+
+# Custom CSS
+html_css_files = ["custom.css"]
+
+# Mermaid configuration
+mermaid_version = "10.9.0"
+mermaid_init_js = "mermaid.initialize({startOnLoad:true, theme:'dark', themeVariables:{primaryColor:'#f97316',primaryTextColor:'#fff',primaryBorderColor:'#ea6a0a',lineColor:'#fb923c',secondaryColor:'#1e293b',tertiaryColor:'#0f172a'}});"