CSS_Project/docs/experiments.html at kimon · codegithubka/CSS_Project

History

3410 lines (3172 loc) · 569 KB

Raw

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

301

302

303

304

305

306

307

308

309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356

357

358

359

360

361

362

363

364

365

366

367

368

369

370

371

372

373

374

375

376

377

378

379

380

381

382

383

384

385

386

387

388

389

390

391

392

393

394

395

396

397

398

399

400

401

402

403

404

405

406

407

408

409

410

411

412

413

414

415

416

417

418

419

420

421

422

423

424

425

426

427

428

429

430

431

432

433

434

435

436

437

438

439

440

441

442

443

444

445

446

447

448

449

450

451

452

453

454

455

456

457

458

459

460

461

462

463

464

465

466

467

468

469

470

471

472

473

474

475

476

477

478

479

480

481

482

483

484

485

486

487

488

489

490

491

492

493

494

495

496

497

498

499

500

501

502

503

504

505

506

507

508

509

510

511

512

513

514

515

516

517

518

519

520

521

522

523

524

525

526

527

528

529

530

531

532

533

534

535

536

537

538

539

540

541

542

543

544

545

546

547

548

549

550

551

552

553

554

555

556

557

558

559

560

561

562

563

564

565

566

567

568

569

570

571

572

573

574

575

576

577

578

579

580

581

582

583

584

585

586

587

588

589

590

591

592

593

594

595

596

597

598

599

600

601

602

603

604

605

606

607

608

609

610

611

612

613

614

615

616

617

618

619

620

621

622

623

624

625

626

627

628

629

630

631

632

633

634

635

636

637

638

639

640

641

642

643

644

645

646

647

648

649

650

651

652

653

654

655

656

657

658

659

660

661

662

663

664

665

666

667

668

669

670

671

672

673

674

675

676

677

678

679

680

681

682

683

684

685

686

687

688

689

690

691

692

693

694

695

696

697

698

699

700

701

702

703

704

705

706

707

708

709

710

711

712

713

714

715

716

717

718

719

720

721

722

723

724

725

726

727

728

729

730

731

732

733

734

735

736

737

738

739

740

741

742

743

744

745

746

747

748

749

750

751

752

753

754

755

756

757

758

759

760

761

762

763

764

765

766

767

768

769

770

771

772

773

774

775

776

777

778

779

780

781

782

783

784

785

786

787

788

789

790

791

792

793

794

795

796

797

798

799

800

801

802

803

804

805

806

807

808

809

810

811

812

813

814

815

816

817

818

819

820

821

822

823

824

825

826

827

828

829

830

831

832

833

834

835

836

837

838

839

840

841

842

843

844

845

846

847

848

849

850

851

852

853

854

855

856

857

858

859

860

861

862

863

864

865

866

867

868

869

870

871

872

873

874

875

876

877

878

879

880

881

882

883

884

885

886

887

888

889

890

891

892

893

894

895

896

897

898

899

900

901

902

903

904

905

906

907

908

909

910

911

912

913

914

915

916

917

918

919

920

921

922

923

924

925

926

927

928

929

930

931

932

933

934

935

936

937

938

939

940

941

942

943

944

945

946

947

948

949

950

951

952

953

954

955

956

957

958

959

960

961

962

963

964

965

966

967

968

969

970

971

972

973

974

975

976

977

978

979

980

981

982

983

984

985

986

987

988

989

990

991

992

993

994

995

996

997

998

999

1000

<!doctype html>

<head>

<title>experiments API documentation</title>

<body>

</svg>

Module Index

</a>

<input type="search" placeholder="Search..." role="searchbox" aria-label="search"

pattern=".+" required>

<h2>Contents</h2>

<ul>

<li><a href="#predator-prey-hydra-effect-experiments">Predator-Prey Hydra Effect Experiments</a></li>

</ul>

<h2>API Documentation</h2>

<li>

<a class="function" href="#generate_unique_seed">generate_unique_seed</a>

</li>

<li>

<a class="function" href="#count_populations">count_populations</a>

</li>

<li>

<a class="function" href="#get_evolved_stats">get_evolved_stats</a>

</li>

<li>

<a class="function" href="#average_pcfs">average_pcfs</a>

</li>

<li>

<a class="function" href="#save_results_jsonl">save_results_jsonl</a>

</li>

<li>

<a class="function" href="#save_results_npz">save_results_npz</a>

</li>

<li>

<a class="function" href="#load_results_jsonl">load_results_jsonl</a>

</li>

<li>

<a class="function" href="#run_single_simulation">run_single_simulation</a>

</li>

<li>

<a class="function" href="#run_phase1">run_phase1</a>

</li>

<li>

<a class="function" href="#run_phase2">run_phase2</a>

</li>

<li>

<a class="function" href="#run_phase3">run_phase3</a>

</li>

<li>

<a class="function" href="#run_phase4">run_phase4</a>

</li>

<li>

<a class="function" href="#run_phase5">run_phase5</a>

</li>

<li>

</li>

</ul>

built with pdoc<img

alt="pdoc logo"

src="data:image/svg+xml,%3Csvg%20xmlns%3D%22http%3A//www.w3.org/2000/svg%22%20role%3D%22img%22%20aria-label%3D%22pdoc%20logo%22%20width%3D%22300%22%20height%3D%22160%22%20viewBox%3D%220%200%20150%2080%22%3E%3Ctitle%3Epdoc%3C/title%3E%3Cpath%20d%3D%22M132.316%2048.886c.276-4.679%202.342-6.698%204.409-7.982s4.27-1.165%206.751-1.055c1.586.07%203.044.156%204.222-.482%201.142-.619%202.026-1.932%202.162-3.739.268-3.576-1.929-5.368-5.006-5.551s-7.599.524-10.517%201.606c-4.455%201.652-8.588%206.606-9.552%208.992s-2.342%206.193-1.745%2010.873%202.664%209.221%205.878%2011.79%205.878%203.808%2010.103%204.312%203.444.229%206.062.229%205.006-2.202%204.914-4.909-2.296-5.001-4.501-4.863-3.077.505-5.281.229-7.715-2.064-7.899-9.451z%22%20fill%3D%22%23198754%22/%3E%3Ccircle%20cx%3D%22101.504%22%20cy%3D%2248.943%22%20r%3D%2214.208%22%20fill%3D%22none%22%20stroke%3D%22%23198754%22%20stroke-width%3D%229.354%22/%3E%3Cpath%20d%3D%22M87.81.002c-3.637.065-5.001.454-7.014%201.232s-3.443%201.363-6.3%204.282c-1.723%201.76-3.148%205.019-3.776%207.329-.413%201.521-.316%202.63-.316%202.63l-.195%2034.612c.065%205.774-6.755%208.305-9.612%208.37s-9.678-1.038-9.743-9.408%207.128-9.521%208.362-9.521c1.413-.13%202.526-.021%203.718-.016%202.071.009%204.157-.778%204.092-4.671s-4.157-4.736-4.157-4.736c-6.3-.843-11.43%202.206-11.43%202.206S40.917%2038.15%2041.372%2049.634%2051.568%2068.19%2061.311%2068.125s18.316-7.007%2018.445-17.193l.13-22.772c.046-2.291%202.683-3.644%204.476-4.203.745-.232%201.694-.274%201.694-.274l10.457-.13s4.871-.324%207.729-3.114%204.352-6.294%204.352-6.294.974-3.049.13-4.606-.195-1.233-2.792-3.309-8.573-4.477-8.573-4.477S91.447-.063%2087.81.002zM0%2047.169l.065%2028.417S0%2080.127%204.481%2079.997s5.072-3.866%205.049-4.152l-.113-28.482s1.624-7.656%209.937-7.721%2010.002%206.942%2010.002%208.499-.909%2010.51-9.093%2010.51c-.948%200-2.99-.567-4.145-.272-3.919%201-3.194%204.554-3.194%204.554s.065%205.061%207.404%204.996%2018.575-6.034%2018.575-19.074S26.953%2030.04%2019.549%2029.91%201.234%2035.296%200%2047.169z%22%20fill%3D%22%23198754%22/%3E%3Cg%20transform%3D%22matrix%28.325601%200%200%20.325256%20-10.32669%20-45.802786%29%22%3E%3Ccircle%20cx%3D%22297.554%22%20cy%3D%22172.286%22%20r%3D%2216.5%22%20fill%3D%22%23fff%22/%3E%3Cellipse%20cx%3D%22297.709%22%20cy%3D%22172.642%22%20rx%3D%2211.071%22%20ry%3D%2210.871%22%20fill%3D%22%23105a48%22/%3E%3Ccircle%20cx%3D%22304.104%22%20cy%3D%22167.667%22%20r%3D%224.5%22%20fill%3D%22%23fff%22/%3E%3C/g%3E%3Cpath%20d%3D%22M94.661%2017.032l.893-1.476s.99.714%201.916.925%201.575.114%202.955.114l14.565-.162c1.283-.032%203.085-.762%203.02-3.293s-.373-3.503-.373-3.503l1.283-.487s.52.503.877%201.573.309%201.995.292%202.66-.227%201.541-.227%201.541%201.564-.308%202.359-1.038.823-.779%201.489-1.508.812-.86.812-.86.552-.13.877.26.341.957.065%201.46-1.672%202.206-3.247%203.066-2.76%201.427-3.929%201.768-3.848.73-7.063.714l-10.944-.114s-2.143-.081-3.02-.373-2.241-.973-2.598-1.265z%22%20fill%3D%22%23d36d49%22/%3E%3Cg%20fill%3D%22%23105a48%22%3E%3Cellipse%20cx%3D%2293.052%22%20cy%3D%2243.567%22%20rx%3D%22.869%22%20ry%3D%221.014%22%20transform%3D%22rotate%28341.022%29%22/%3E%3Cellipse%20cx%3D%22104.3%22%20cy%3D%22-16.184%22%20rx%3D%22.865%22%20ry%3D%221.009%22%20transform%3D%22rotate%2814.786%29%22/%3E%3C/g%3E%3C/svg%3E"/>

</a>

</div>

</nav>

experiments </h1>

<div class="docstring"><h1 id="predator-prey-hydra-effect-experiments">Predator-Prey Hydra Effect Experiments</h1>

HPC-ready experiment runner for investigating the Hydra effect in

predator-prey cellular automata.

<h6 id="experimental-phases">Experimental Phases</h6>

<ul>

<li>Phase 1: Parameter sweep to find critical point (bifurcation + cluster analysis)</li>

<li>Phase 2: Self-organization analysis (evolution toward criticality)</li>

<li>Phase 3: Finite-size scaling at critical point</li>

<li>Phase 4: Sensitivity analysis across parameter regimes</li>

<li>Phase 5: Model extensions (directed hunting comparison)</li>

</ul>

<h6 id="functions">Functions</h6>

<pre><code>run_single_simulation # Execute one simulation run and collect metrics.

run_phase1, run_phase2, run_phase3, run_phase4, run_phase5 # Phase-specific experiment runners.

</code></pre>

</div>

<h6 id="utilities">Utilities</h6>

<pre><code>generate_unique_seed # Deterministic seed generation from parameters.

count_populations # Count species populations on grid.

get_evolved_stats # Statistics for evolved parameters.

average_pcfs # Average multiple PCF measurements.

save_results_jsonl, load_results_jsonl, save_results_npz # I/O functions for experiment results.

</code></pre>

</div>

<h6 id="command-line-usage">Command Line Usage</h6>

<pre><code>python experiments.py --phase 1 # Run phase 1

python experiments.py --phase 1 --dry-run # Estimate runtime

python experiments.py --phase all # Run all phases

python experiments.py --phase 1 --output results/ # Custom output

</code></pre>

</div>

<h6 id="programmatic-usage">Programmatic Usage</h6>

<pre><code>from experiments import run_single_simulation, run_phase1

from <a href="models/config.html">models.config</a> import PHASE1_CONFIG

# Single simulation

result = run_single_simulation(

prey_birth=0.2,

prey_death=0.05,

predator_birth=0.8,

predator_death=0.1,

grid_size=100,

seed=42,

cfg=PHASE1_CONFIG,

)

# Full phase (writes to output directory)

import logging

results = run_phase1(PHASE1_CONFIG, Path("results/"), logging.getLogger())

</code></pre>

</div>

<label class="view-source-button" for="mod-experiments-view-source">View Source</label>

<div class="pdoc-code codehilite"><pre><a href="#L-1"> 1</a>#!/usr/bin/env python3

<a href="#L-2"> 2</a>"""

<a href="#L-3"> 3</a>Predator-Prey Hydra Effect Experiments

<a href="#L-4"> 4</a>======================================

<a href="#L-6"> 6</a>HPC-ready experiment runner for investigating the Hydra effect in

<a href="#L-7"> 7</a>predator-prey cellular automata.

<a href="#L-9"> 9</a>Experimental Phases

<a href="#L-10"> 10</a>-------------------

<a href="#L-11"> 11</a>- **Phase 1**: Parameter sweep to find critical point (bifurcation + cluster analysis)

<a href="#L-12"> 12</a>- **Phase 2**: Self-organization analysis (evolution toward criticality)

<a href="#L-13"> 13</a>- **Phase 3**: Finite-size scaling at critical point

<a href="#L-14"> 14</a>- **Phase 4**: Sensitivity analysis across parameter regimes

<a href="#L-15"> 15</a>- **Phase 5**: Model extensions (directed hunting comparison)

<a href="#L-17"> 17</a>Functions

<a href="#L-18"> 18</a>---------

<a href="#L-19"> 19</a>```python

<a href="#L-20"> 20</a>run_single_simulation # Execute one simulation run and collect metrics.

<a href="#L-21"> 21</a>run_phase1, run_phase2, run_phase3, run_phase4, run_phase5 # Phase-specific experiment runners.

<a href="#L-22"> 22</a>```

<a href="#L-24"> 24</a>Utilities

<a href="#L-25"> 25</a>---------

<a href="#L-26"> 26</a>```python

<a href="#L-27"> 27</a>generate_unique_seed # Deterministic seed generation from parameters.

<a href="#L-28"> 28</a>count_populations # Count species populations on grid.

<a href="#L-29"> 29</a>get_evolved_stats # Statistics for evolved parameters.

<a href="#L-30"> 30</a>average_pcfs # Average multiple PCF measurements.

<a href="#L-31"> 31</a>save_results_jsonl, load_results_jsonl, save_results_npz # I/O functions for experiment results.

<a href="#L-32"> 32</a>```

<a href="#L-34"> 34</a>Command Line Usage

<a href="#L-35"> 35</a>------------------

<a href="#L-36"> 36</a>```bash

<a href="#L-37"> 37</a>python experiments.py --phase 1 # Run phase 1

<a href="#L-38"> 38</a>python experiments.py --phase 1 --dry-run # Estimate runtime

<a href="#L-39"> 39</a>python experiments.py --phase all # Run all phases

<a href="#L-40"> 40</a>python experiments.py --phase 1 --output results/ # Custom output

<a href="#L-41"> 41</a>```

<a href="#L-43"> 43</a>Programmatic Usage

<a href="#L-44"> 44</a>------------------

<a href="#L-45"> 45</a>```python

<a href="#L-46"> 46</a>from experiments import run_single_simulation, run_phase1

<a href="#L-47"> 47</a>from models.config import PHASE1_CONFIG

<a href="#L-49"> 49</a># Single simulation

<a href="#L-50"> 50</a>result = run_single_simulation(

<a href="#L-51"> 51</a> prey_birth=0.2,

<a href="#L-52"> 52</a> prey_death=0.05,

<a href="#L-53"> 53</a> predator_birth=0.8,

<a href="#L-54"> 54</a> predator_death=0.1,

<a href="#L-55"> 55</a> grid_size=100,

<a href="#L-56"> 56</a> seed=42,

<a href="#L-57"> 57</a> cfg=PHASE1_CONFIG,

<a href="#L-58"> 58</a>)

<a href="#L-60"> 60</a># Full phase (writes to output directory)

<a href="#L-61"> 61</a>import logging

<a href="#L-62"> 62</a>results = run_phase1(PHASE1_CONFIG, Path("results/"), logging.getLogger())

<a href="#L-63"> 63</a>```

<a href="#L-64"> 64</a>"""

<a href="#L-66"> 66</a>import argparse

<a href="#L-67"> 67</a>import hashlib

<a href="#L-68"> 68</a>import json

<a href="#L-69"> 69</a>import logging

<a href="#L-70"> 70</a>import os

<a href="#L-71"> 71</a>import sys

<a href="#L-72"> 72</a>import time

<a href="#L-73"> 73</a>from dataclasses import asdict

<a href="#L-74"> 74</a>from pathlib import Path

<a href="#L-75"> 75</a>from typing import Dict, List, Tuple, Optional

<a href="#L-76"> 76</a>import warnings

<a href="#L-78"> 78</a>import numpy as np

<a href="#L-79"> 79</a>from tqdm import tqdm

<a href="#L-81"> 81</a>warnings.filterwarnings("ignore")

<a href="#L-83"> 83</a># Project imports

<a href="#L-84"> 84</a>project_root = str(Path(__file__).parent.parent)

<a href="#L-85"> 85</a>if project_root not in sys.path:

<a href="#L-86"> 86</a> sys.path.insert(0, project_root)

<a href="#L-88"> 88</a>from models.config import Config, get_phase_config, PHASE_CONFIGS

<a href="#L-90"> 90</a># Numba imports

<a href="#L-91"> 91</a>try:

<a href="#L-92"> 92</a> from models.numba_optimized import (

<a href="#L-93"> 93</a> compute_all_pcfs_fast,

<a href="#L-94"> 94</a> get_cluster_stats_fast,

<a href="#L-95"> 95</a> warmup_numba_kernels,

<a href="#L-96"> 96</a> set_numba_seed,

<a href="#L-97"> 97</a> NUMBA_AVAILABLE,

<a href="#L-98"> 98</a> )

<a href="#L-100"> 100</a> USE_NUMBA = NUMBA_AVAILABLE

<a href="#L-101"> 101</a>except ImportError:

<a href="#L-102"> 102</a> USE_NUMBA = False

<a href="#L-104"> 104</a> def warmup_numba_kernels(size, **kwargs):

<a href="#L-105"> 105</a> pass

<a href="#L-107"> 107</a> def set_numba_seed(seed):

<a href="#L-108"> 108</a> pass

<a href="#L-111"> 111</a># =============================================================================

<a href="#L-112"> 112</a># Utility Functions

<a href="#L-113"> 113</a># =============================================================================

<a href="#L-116"> 116</a>def generate_unique_seed(params: dict, rep: int) -> int:

<a href="#L-117"> 117</a> """

<a href="#L-118"> 118</a> Create a deterministic seed from a dictionary of parameters and a repetition index.

<a href="#L-120"> 120</a> This function serializes the input dictionary into a sorted JSON string,

<a href="#L-121"> 121</a> appends the repetition count, and hashes the resulting string using SHA-256.

<a href="#L-122"> 122</a> The first 8 characters of the hex digest are then converted to an integer

<a href="#L-123"> 123</a> to provide a stable, unique seed for random number generators.

<a href="#L-125"> 125</a> Parameters

<a href="#L-126"> 126</a> ----------

<a href="#L-127"> 127</a> params : dict

<a href="#L-128"> 128</a> A dictionary of configuration parameters. Keys are sorted to ensure

<a href="#L-129"> 129</a> determinism regardless of insertion order.

<a href="#L-130"> 130</a> rep : int

<a href="#L-131"> 131</a> The repetition or iteration index, used to ensure different seeds

<a href="#L-132"> 132</a> are generated for the same parameter set across multiple runs.

<a href="#L-134"> 134</a> Returns

<a href="#L-135"> 135</a> -------

<a href="#L-136"> 136</a> int

<a href="#L-137"> 137</a> A unique integer seed derived from the input parameters.

<a href="#L-139"> 139</a> Examples

<a href="#L-140"> 140</a> --------

<a href="#L-141"> 141</a> >>> params = {'learning_rate': 0.01, 'batch_size': 32}

<a href="#L-142"> 142</a> >>> generate_unique_seed(params, 1)

<a href="#L-143"> 143</a> 3432571217

<a href="#L-144"> 144</a> >>> generate_unique_seed(params, 2)

<a href="#L-145"> 145</a> 3960013583

<a href="#L-146"> 146</a> """

<a href="#L-147"> 147</a> identifier = json.dumps(params, sort_keys=True) + f"_{rep}"

<a href="#L-148"> 148</a> return int(hashlib.sha256(identifier.encode()).hexdigest()[:8], 16)

<a href="#L-151"> 151</a>def count_populations(grid: np.ndarray) -> Tuple[int, int, int]:

<a href="#L-152"> 152</a> """

<a href="#L-153"> 153</a> Count the number of empty, prey, and predator cells in the simulation grid.

<a href="#L-155"> 155</a> Parameters

<a href="#L-156"> 156</a> ----------

<a href="#L-157"> 157</a> grid : np.ndarray

<a href="#L-158"> 158</a> A 2D NumPy array representing the simulation environment, where:

<a href="#L-159"> 159</a> - 0: Empty cell

<a href="#L-160"> 160</a> - 1: Prey

<a href="#L-161"> 161</a> - 2: Predator

<a href="#L-163"> 163</a> Returns

<a href="#L-164"> 164</a> -------

<a href="#L-165"> 165</a> empty_count : int

<a href="#L-166"> 166</a> Total number of cells with a value of 0.

<a href="#L-167"> 167</a> prey_count : int

<a href="#L-168"> 168</a> Total number of cells with a value of 1.

<a href="#L-169"> 169</a> predator_count : int

<a href="#L-170"> 170</a> Total number of cells with a value of 2.

<a href="#L-172"> 172</a> Examples

<a href="#L-173"> 173</a> --------

<a href="#L-174"> 174</a> >>> grid = np.array([[0, 1], [2, 1]])

<a href="#L-175"> 175</a> >>> count_populations(grid)

<a href="#L-176"> 176</a> (1, 2, 1)

<a href="#L-177"> 177</a> """

<a href="#L-178"> 178</a> return int(np.sum(grid == 0)), int(np.sum(grid == 1)), int(np.sum(grid == 2))

<a href="#L-181"> 181</a>def get_evolved_stats(model, param: str) -> Dict:

<a href="#L-182"> 182</a> """

<a href="#L-183"> 183</a> Get statistics of an evolved parameter from the model.

<a href="#L-185"> 185</a> This function retrieves parameter values from the model's internal storage,

<a href="#L-186"> 186</a> filters out NaN values, and calculates basic descriptive statistics.

<a href="#L-188"> 188</a> Parameters

<a href="#L-189"> 189</a> ----------

<a href="#L-190"> 190</a> model : object

<a href="#L-191"> 191</a> The simulation model instance containing a `cell_params` attribute

<a href="#L-192"> 192</a> with a `.get()` method.

<a href="#L-193"> 193</a> param : str

<a href="#L-194"> 194</a> The name of the parameter to calculate statistics for.

<a href="#L-196"> 196</a> Returns

<a href="#L-197"> 197</a> -------

<a href="#L-198"> 198</a> stats : dict

<a href="#L-199"> 199</a> A dictionary containing the following keys:

<a href="#L-200"> 200</a> - 'mean': Arithmetic mean of valid values.

<a href="#L-201"> 201</a> - 'std': Standard deviation of valid values.

<a href="#L-202"> 202</a> - 'min': Minimum valid value.

<a href="#L-203"> 203</a> - 'max': Maximum valid value.

<a href="#L-204"> 204</a> - 'n': Count of non-NaN values.

<a href="#L-205"> 205</a> If no valid data is found, all stats return NaN and n returns 0.

<a href="#L-207"> 207</a> Examples

<a href="#L-208"> 208</a> --------

<a href="#L-209"> 209</a> >>> stats = get_evolved_stats(my_model, "speed")

<a href="#L-210"> 210</a> >>> print(stats['mean'])

<a href="#L-211"> 211</a> 1.25

<a href="#L-212"> 212</a> """

<a href="#L-213"> 213</a> arr = model.cell_params.get(param)

<a href="#L-214"> 214</a> if arr is None:

<a href="#L-215"> 215</a> return {"mean": np.nan, "std": np.nan, "min": np.nan, "max": np.nan, "n": 0}

<a href="#L-216"> 216</a> valid = arr[~np.isnan(arr)]

<a href="#L-217"> 217</a> if len(valid) == 0:

<a href="#L-218"> 218</a> return {"mean": np.nan, "std": np.nan, "min": np.nan, "max": np.nan, "n": 0}

<a href="#L-219"> 219</a> return {

<a href="#L-220"> 220</a> "mean": float(np.mean(valid)),

<a href="#L-221"> 221</a> "std": float(np.std(valid)),

<a href="#L-222"> 222</a> "min": float(np.min(valid)),

<a href="#L-223"> 223</a> "max": float(np.max(valid)),

<a href="#L-224"> 224</a> "n": len(valid),

<a href="#L-225"> 225</a> }

<a href="#L-228"> 228</a>def average_pcfs(

<a href="#L-229"> 229</a> pcf_list: List[Tuple[np.ndarray, np.ndarray, int]],

<a href="#L-230"> 230</a>) -> Tuple[np.ndarray, np.ndarray, np.ndarray]:

<a href="#L-231"> 231</a> """

<a href="#L-232"> 232</a> Average multiple Pair Correlation Function (PCF) measurements and calculate standard error.

<a href="#L-234"> 234</a> Parameters

<a href="#L-235"> 235</a> ----------

<a href="#L-236"> 236</a> pcf_list : list of tuple

<a href="#L-237"> 237</a> A list where each element is a tuple containing:

<a href="#L-238"> 238</a> - distances (np.ndarray): The radial distances (r).

<a href="#L-239"> 239</a> - pcf_values (np.ndarray): The correlation values g(r).

<a href="#L-240"> 240</a> - count (int): Metadata or weight (not used in current calculation).

<a href="#L-242"> 242</a> Returns

<a href="#L-243"> 243</a> -------

<a href="#L-244"> 244</a> distances : np.ndarray

<a href="#L-245"> 245</a> The radial distances from the first entry in the list.

<a href="#L-246"> 246</a> pcf_mean : np.ndarray

<a href="#L-247"> 247</a> The element-wise mean of the PCF values across all measurements.

<a href="#L-248"> 248</a> pcf_se : np.ndarray

<a href="#L-249"> 249</a> The standard error of the mean for the PCF values.

<a href="#L-251"> 251</a> Examples

<a href="#L-252"> 252</a> --------

<a href="#L-253"> 253</a> >>> data = [(np.array([0, 1]), np.array([1.0, 2.0]), 10),

<a href="#L-254"> 254</a> ... (np.array([0, 1]), np.array([1.2, 1.8]), 12)]

<a href="#L-255"> 255</a> >>> dist, mean, se = average_pcfs(data)

<a href="#L-256"> 256</a> >>> mean

<a href="#L-257"> 257</a> array([1.1, 1.9])

<a href="#L-258"> 258</a> """

<a href="#L-259"> 259</a> if len(pcf_list) == 0:

<a href="#L-260"> 260</a> return np.array([]), np.array([]), np.array([])

<a href="#L-262"> 262</a> distances = pcf_list[0][0]

<a href="#L-263"> 263</a> pcfs = np.array([p[1] for p in pcf_list])

<a href="#L-265"> 265</a> pcf_mean = np.mean(pcfs, axis=0)

<a href="#L-266"> 266</a> pcf_se = np.std(pcfs, axis=0) / np.sqrt(len(pcfs))

<a href="#L-268"> 268</a> return distances, pcf_mean, pcf_se

<a href="#L-271"> 271</a>def save_results_jsonl(results: List[Dict], output_path: Path):

<a href="#L-272"> 272</a> """

<a href="#L-273"> 273</a> Save a list of dictionaries to a file in JSON Lines (JSONL) format.

<a href="#L-275"> 275</a> Each dictionary in the list is serialized into a single JSON string and

<a href="#L-276"> 276</a> written as a new line. Non-serializable objects are converted to strings

<a href="#L-277"> 277</a> using the default string representation.

<a href="#L-279"> 279</a> Parameters

<a href="#L-280"> 280</a> ----------

<a href="#L-281"> 281</a> results : list of dict

<a href="#L-282"> 282</a> The collection of result dictionaries to be saved.

<a href="#L-283"> 283</a> output_path : Path

<a href="#L-284"> 284</a> The file system path (pathlib.Path) where the JSONL file will be created.

<a href="#L-286"> 286</a> Returns

<a href="#L-287"> 287</a> -------

<a href="#L-288"> 288</a> None

<a href="#L-290"> 290</a> Notes

<a href="#L-291"> 291</a> -----

<a href="#L-292"> 292</a> The file is opened in 'w' (write) mode, which will overwrite any existing

<a href="#L-293"> 293</a> content at the specified path.

<a href="#L-295"> 295</a> Examples

<a href="#L-296"> 296</a> --------

<a href="#L-297"> 297</a> >>> data = [{"id": 1, "score": 0.95}, {"id": 2, "score": 0.88}]

<a href="#L-298"> 298</a> >>> save_results_jsonl(data, Path("results.jsonl"))

<a href="#L-299"> 299</a> """

<a href="#L-300"> 300</a> with open(output_path, "w", encoding="utf-8") as f:

<a href="#L-301"> 301</a> for result in results:

<a href="#L-302"> 302</a> f.write(json.dumps(result, default=str) + "\n")

<a href="#L-305"> 305</a>def save_results_npz(results: List[Dict], output_path: Path):

<a href="#L-306"> 306</a> """

<a href="#L-307"> 307</a> Save simulation results to a compressed NumPy (.npz) binary file.

<a href="#L-309"> 309</a> This function flattens a list of result dictionaries into a single

<a href="#L-310"> 310</a> dictionary of NumPy arrays, prefixing keys with the run index to

<a href="#L-311"> 311</a> maintain data separation. The resulting file is compressed to

<a href="#L-312"> 312</a> reduce storage space.

<a href="#L-314"> 314</a> Parameters

<a href="#L-315"> 315</a> ----------

<a href="#L-316"> 316</a> results : list of dict

<a href="#L-317"> 317</a> A list where each dictionary contains key-value pairs of

<a href="#L-318"> 318</a> simulation data (e.g., arrays, lists, or scalars).

<a href="#L-319"> 319</a> output_path : Path

<a href="#L-320"> 320</a> The file system path (pathlib.Path) where the compressed

<a href="#L-321"> 321</a> NPZ file will be saved.

<a href="#L-323"> 323</a> Returns

<a href="#L-324"> 324</a> -------

<a href="#L-325"> 325</a> None

<a href="#L-327"> 327</a> Notes

<a href="#L-328"> 328</a> -----

<a href="#L-329"> 329</a> The keys in the saved file follow the format 'run_{index}_{original_key}'.

<a href="#L-330"> 330</a> Values are automatically converted to NumPy arrays if they are not

<a href="#L-331"> 331</a> already.

<a href="#L-333"> 333</a> Examples

<a href="#L-334"> 334</a> --------

<a href="#L-335"> 335</a> >>> results = [{"energy": [1, 2]}, {"energy": [3, 4]}]

<a href="#L-336"> 336</a> >>> save_results_npz(results, Path("output.npz"))

<a href="#L-337"> 337</a> """

<a href="#L-338"> 338</a> data = {}

<a href="#L-339"> 339</a> for i, res in enumerate(results):

<a href="#L-340"> 340</a> for key, val in res.items():

<a href="#L-341"> 341</a> data[f"run_{i}_{key}"] = np.array(val)

<a href="#L-342"> 342</a> np.savez_compressed(output_path, **data)

<a href="#L-345"> 345</a>def load_results_jsonl(input_path: Path) -> List[Dict]:

<a href="#L-346"> 346</a> """

<a href="#L-347"> 347</a> Load simulation results from a JSON Lines (JSONL) formatted file.

<a href="#L-349"> 349</a> This function reads a file line-by-line, parsing each line as an

<a href="#L-350"> 350</a> independent JSON object and aggregating them into a list of dictionaries.

<a href="#L-352"> 352</a> Parameters

<a href="#L-353"> 353</a> ----------

<a href="#L-354"> 354</a> input_path : Path

<a href="#L-355"> 355</a> The file system path (pathlib.Path) to the JSONL file.

<a href="#L-357"> 357</a> Returns

<a href="#L-358"> 358</a> -------

<a href="#L-359"> 359</a> results : list of dict

<a href="#L-360"> 360</a> A list of dictionaries reconstructed from the file content.

<a href="#L-362"> 362</a> Raises

<a href="#L-363"> 363</a> ------

<a href="#L-364"> 364</a> FileNotFoundError

<a href="#L-365"> 365</a> If the specified input path does not exist.

<a href="#L-366"> 366</a> json.JSONDecodeError

<a href="#L-367"> 367</a> If a line in the file is not valid JSON.

<a href="#L-369"> 369</a> Examples

<a href="#L-370"> 370</a> --------

<a href="#L-371"> 371</a> >>> data = load_results_jsonl(Path("results.jsonl"))

<a href="#L-372"> 372</a> >>> len(data)

<a href="#L-373"> 373</a> 2

<a href="#L-374"> 374</a> """

<a href="#L-375"> 375</a> results = []

<a href="#L-376"> 376</a> with open(input_path, "r", encoding="utf-8") as f:

<a href="#L-377"> 377</a> for line in f:

<a href="#L-378"> 378</a> results.append(json.loads(line.strip()))

<a href="#L-379"> 379</a> return results

<a href="#L-382"> 382</a># =============================================================================

<a href="#L-383"> 383</a># Simulation Functionality

<a href="#L-384"> 384</a># =============================================================================

<a href="#L-387"> 387</a>def run_single_simulation(

<a href="#L-388"> 388</a> prey_birth: float,

<a href="#L-389"> 389</a> prey_death: float,

<a href="#L-390"> 390</a> predator_birth: float,

<a href="#L-391"> 391</a> predator_death: float,

<a href="#L-392"> 392</a> grid_size: int,

<a href="#L-393"> 393</a> seed: int,

<a href="#L-394"> 394</a> cfg: Config,

<a href="#L-395"> 395</a> with_evolution: bool = False,

<a href="#L-396"> 396</a> compute_pcf: Optional[bool] = None,

<a href="#L-397"> 397</a>) -> Dict:

<a href="#L-398"> 398</a> """

<a href="#L-399"> 399</a> Run a single Predator-Prey (PP) simulation and collect comprehensive metrics.

<a href="#L-401"> 401</a> This function initializes a Cellular Automata model, executes a warmup phase

<a href="#L-402"> 402</a> to reach steady state, and then performs a measurement phase to track

<a href="#L-403"> 403</a> population dynamics, spatial clustering, and evolutionary changes.

<a href="#L-405"> 405</a> Parameters

<a href="#L-406"> 406</a> ----------

<a href="#L-407"> 407</a> prey_birth : float

<a href="#L-408"> 408</a> The probability or rate of prey reproduction.

<a href="#L-409"> 409</a> prey_death : float

<a href="#L-410"> 410</a> The base probability or rate of prey mortality.

<a href="#L-411"> 411</a> predator_birth : float

<a href="#L-412"> 412</a> The probability or rate of predator reproduction upon consuming prey.

<a href="#L-413"> 413</a> predator_death : float

<a href="#L-414"> 414</a> The probability or rate of predator mortality.

<a href="#L-415"> 415</a> grid_size : int

<a href="#L-416"> 416</a> The side length of the square simulation grid.

<a href="#L-417"> 417</a> seed : int

<a href="#L-418"> 418</a> Random seed for ensuring reproducibility of the simulation run.

<a href="#L-419"> 419</a> cfg : Config

<a href="#L-420"> 420</a> A configuration object containing simulation hyperparameters (densities,

<a href="#L-421"> 421</a> sampling rates, timing, etc.).

<a href="#L-422"> 422</a> with_evolution : bool, optional

<a href="#L-423"> 423</a> If True, enables the evolution of the 'prey_death' parameter within

<a href="#L-424"> 424</a> the model (default is False).

<a href="#L-425"> 425</a> compute_pcf : bool, optional

<a href="#L-426"> 426</a> Explicit toggle for Pair Correlation Function calculation. If None,

<a href="#L-427"> 427</a> it is determined by `cfg.pcf_sample_rate` (default is None).

<a href="#L-429"> 429</a> Returns

<a href="#L-430"> 430</a> -------

<a href="#L-431"> 431</a> result : dict

<a href="#L-432"> 432</a> A dictionary containing simulation results including:

<a href="#L-433"> 433</a> - Input parameters and survival flags.

<a href="#L-434"> 434</a> - Population mean and standard deviation for both species.

<a href="#L-435"> 435</a> - Cluster statistics (number of clusters, sizes, largest fractions).

<a href="#L-436"> 436</a> - Evolutionary statistics (mean, std, min, max, and final values).

<a href="#L-437"> 437</a> - PCF data and spatial indices (segregation and clustering).

<a href="#L-438"> 438</a> - Optional time series for populations and evolved parameters.

<a href="#L-440"> 440</a> Notes

<a href="#L-441"> 441</a> -----

<a href="#L-442"> 442</a> The function relies on several external utilities: `count_populations`,

<a href="#L-443"> 443</a> `get_evolved_stats`, `get_cluster_stats_fast`, `compute_all_pcfs_fast`,

<a href="#L-444"> 444</a> and `average_pcfs`.

<a href="#L-445"> 445</a> """

<a href="#L-447"> 447</a> from models.CA import PP

<a href="#L-449"> 449</a> if USE_NUMBA:

<a href="#L-450"> 450</a> set_numba_seed(seed)

<a href="#L-452"> 452</a> if compute_pcf is None:

<a href="#L-453"> 453</a> compute_pcf = cfg.collect_pcf and (np.random.random() < cfg.pcf_sample_rate)

<a href="#L-455"> 455</a> # Initialize model

<a href="#L-456"> 456</a> model = PP(

<a href="#L-457"> 457</a> rows=grid_size,

<a href="#L-458"> 458</a> cols=grid_size,

<a href="#L-459"> 459</a> densities=cfg.densities,

<a href="#L-460"> 460</a> neighborhood="moore", # NOTE: Default neighborhood

<a href="#L-461"> 461</a> params={

<a href="#L-462"> 462</a> "prey_birth": prey_birth,

<a href="#L-463"> 463</a> "prey_death": prey_death,

<a href="#L-464"> 464</a> "predator_death": predator_death,

<a href="#L-465"> 465</a> "predator_birth": predator_birth,

<a href="#L-466"> 466</a> },

<a href="#L-467"> 467</a> seed=seed,

<a href="#L-468"> 468</a> directed_hunting=cfg.directed_hunting,

<a href="#L-469"> 469</a> )

<a href="#L-471"> 471</a> if with_evolution:

<a href="#L-472"> 472</a> model.evolve(

<a href="#L-473"> 473</a> "prey_death",

<a href="#L-474"> 474</a> sd=cfg.evolve_sd,

<a href="#L-475"> 475</a> min_val=cfg.evolve_min,

<a href="#L-476"> 476</a> max_val=cfg.evolve_max,

<a href="#L-477"> 477</a> )

<a href="#L-479"> 479</a> # Scale timing with grid size

<a href="#L-480"> 480</a> warmup_steps = cfg.get_warmup_steps(grid_size)

<a href="#L-481"> 481</a> measurement_steps = cfg.get_measurement_steps(grid_size)

<a href="#L-483"> 483</a> # Warmup phase

<a href="#L-484"> 484</a> for _ in range(warmup_steps):

<a href="#L-485"> 485</a> model.update()

<a href="#L-487"> 487</a> # Measurement phase: start collecting our mertics

<a href="#L-488"> 488</a> prey_pops, pred_pops = [], [] # Prey populations and predator populations

<a href="#L-489"> 489</a> evolved_means, evolved_stds = [], [] # Evolution stats over time

<a href="#L-490"> 490</a> cluster_sizes_prey, cluster_sizes_pred = [], [] # Cluster sizes

<a href="#L-491"> 491</a> largest_fractions_prey, largest_fractions_pred = (

<a href="#L-492"> 492</a> [],

<a href="#L-493"> 493</a> [],

<a href="#L-494"> 494</a> ) # Largest cluster fractions = size of largest cluster / total population

<a href="#L-495"> 495</a> pcf_samples = {"prey_prey": [], "pred_pred": [], "prey_pred": []}

<a href="#L-497"> 497</a> # Determine minimum count for analysis

<a href="#L-498"> 498</a> min_count = int(cfg.min_density_for_analysis * (grid_size**2))

<a href="#L-500"> 500</a> for step in range(measurement_steps):

<a href="#L-501"> 501</a> model.update()

<a href="#L-503"> 503</a> _, prey, pred = count_populations(model.grid)

<a href="#L-504"> 504</a> prey_pops.append(prey)

<a href="#L-505"> 505</a> pred_pops.append(pred)

<a href="#L-507"> 507</a> # Track evolution

<a href="#L-508"> 508</a> if with_evolution:

<a href="#L-509"> 509</a> stats = get_evolved_stats(model, "prey_death")

<a href="#L-510"> 510</a> evolved_means.append(stats["mean"])

<a href="#L-511"> 511</a> evolved_stds.append(stats["std"])

<a href="#L-513"> 513</a> # Cluster analysis (at end of measurement)

<a href="#L-514"> 514</a> if step == measurement_steps - 1:

<a href="#L-515"> 515</a> prey_survived = prey_pops[-1] > min_count

<a href="#L-516"> 516</a> pred_survived = pred_pops[-1] > (min_count // 4)

<a href="#L-518"> 518</a> if prey_survived:

<a href="#L-519"> 519</a> prey_stats = get_cluster_stats_fast(model.grid, 1)

<a href="#L-520"> 520</a> cluster_sizes_prey = prey_stats["sizes"].tolist()

<a href="#L-521"> 521</a> largest_fractions_prey.append(prey_stats["largest_fraction"])

<a href="#L-523"> 523</a> if pred_survived:

<a href="#L-524"> 524</a> pred_stats = get_cluster_stats_fast(model.grid, 2)

<a href="#L-525"> 525</a> cluster_sizes_pred = pred_stats["sizes"].tolist()

<a href="#L-526"> 526</a> largest_fractions_pred.append(pred_stats["largest_fraction"])

<a href="#L-528"> 528</a> # PCF requires both

<a href="#L-529"> 529</a> if compute_pcf and prey_survived and pred_survived:

<a href="#L-530"> 530</a> max_dist = min(grid_size / 2, cfg.pcf_max_distance)

<a href="#L-531"> 531</a> pcf_data = compute_all_pcfs_fast(model.grid, max_dist, cfg.pcf_n_bins)

<a href="#L-532"> 532</a> pcf_samples["prey_prey"].append(pcf_data["prey_prey"])

<a href="#L-533"> 533</a> pcf_samples["pred_pred"].append(pcf_data["pred_pred"])

<a href="#L-534"> 534</a> pcf_samples["prey_pred"].append(pcf_data["prey_pred"])

<a href="#L-536"> 536</a> # Compile results

<a href="#L-537"> 537</a> result = {

<a href="#L-538"> 538</a> # Parameters

<a href="#L-539"> 539</a> "prey_birth": prey_birth,

<a href="#L-540"> 540</a> "prey_death": prey_death,

<a href="#L-541"> 541</a> "predator_birth": predator_birth,

<a href="#L-542"> 542</a> "predator_death": predator_death,

<a href="#L-543"> 543</a> "grid_size": grid_size,

<a href="#L-544"> 544</a> "with_evolution": with_evolution,

<a href="#L-545"> 545</a> "seed": seed,

<a href="#L-546"> 546</a> # Population dynamics

<a href="#L-547"> 547</a> "prey_mean": float(np.mean(prey_pops)),

<a href="#L-548"> 548</a> "prey_std": float(np.std(prey_pops)),

<a href="#L-549"> 549</a> "pred_mean": float(np.mean(pred_pops)),

<a href="#L-550"> 550</a> "pred_std": float(np.std(pred_pops)),

<a href="#L-551"> 551</a> "prey_survived": prey_pops[-1] > min_count,

<a href="#L-552"> 552</a> "pred_survived": pred_pops[-1] > (min_count // 4),

<a href="#L-553"> 553</a> # Cluster statistics

<a href="#L-554"> 554</a> "prey_n_clusters": len(cluster_sizes_prey),

<a href="#L-555"> 555</a> "pred_n_clusters": len(cluster_sizes_pred),

<a href="#L-556"> 556</a> "prey_cluster_sizes": cluster_sizes_prey,

<a href="#L-557"> 557</a> "pred_cluster_sizes": cluster_sizes_pred,

<a href="#L-558"> 558</a> # Order parameters

<a href="#L-559"> 559</a> "prey_largest_fraction": (

<a href="#L-560"> 560</a> float(np.mean(largest_fractions_prey)) if largest_fractions_prey else np.nan

<a href="#L-561"> 561</a> ),

<a href="#L-562"> 562</a> "pred_largest_fraction": (

<a href="#L-563"> 563</a> float(np.mean(largest_fractions_pred)) if largest_fractions_pred else np.nan

<a href="#L-564"> 564</a> ),

<a href="#L-565"> 565</a> }

<a href="#L-567"> 567</a> # Time series (if requested)

<a href="#L-568"> 568</a> if cfg.save_timeseries:

<a href="#L-569"> 569</a> subsample = cfg.timeseries_subsample

<a href="#L-570"> 570</a> result["prey_timeseries"] = prey_pops[

<a href="#L-571"> 571</a> ::subsample

<a href="#L-572"> 572</a> ] # NOTE: Sample temporal data every 'subsample' steps

<a href="#L-573"> 573</a> result["pred_timeseries"] = pred_pops[::subsample]

<a href="#L-575"> 575</a> # Evolution statistics

<a href="#L-576"> 576</a> if with_evolution and evolved_means:

<a href="#L-577"> 577</a> valid_means = [v for v in evolved_means if not np.isnan(v)]

<a href="#L-578"> 578</a> result["evolved_prey_death_mean"] = (

<a href="#L-579"> 579</a> float(np.mean(valid_means)) if valid_means else np.nan

<a href="#L-580"> 580</a> )

<a href="#L-581"> 581</a> result["evolved_prey_death_std"] = (

<a href="#L-582"> 582</a> float(np.mean([v for v in evolved_stds if not np.isnan(v)]))

<a href="#L-583"> 583</a> if evolved_stds

<a href="#L-584"> 584</a> else np.nan

<a href="#L-585"> 585</a> )

<a href="#L-586"> 586</a> result["evolved_prey_death_final"] = valid_means[-1] if valid_means else np.nan

<a href="#L-587"> 587</a> result["evolved_prey_death_min"] = (

<a href="#L-588"> 588</a> float(np.min(valid_means)) if valid_means else np.nan

<a href="#L-589"> 589</a> )

<a href="#L-590"> 590</a> result["evolved_prey_death_max"] = (

<a href="#L-591"> 591</a> float(np.max(valid_means)) if valid_means else np.nan

<a href="#L-592"> 592</a> )

<a href="#L-593"> 593</a> result["evolve_sd"] = cfg.evolve_sd

<a href="#L-595"> 595</a> if cfg.save_timeseries:

<a href="#L-596"> 596</a> result["evolved_prey_death_timeseries"] = evolved_means[

<a href="#L-597"> 597</a> :: cfg.timeseries_subsample

<a href="#L-598"> 598</a> ]

<a href="#L-600"> 600</a> # PCF statistics

<a href="#L-601"> 601</a> if pcf_samples["prey_prey"]:

<a href="#L-602"> 602</a> dist, pcf_rr, _ = average_pcfs(pcf_samples["prey_prey"])

<a href="#L-603"> 603</a> _, pcf_cc, _ = average_pcfs(pcf_samples["pred_pred"])

<a href="#L-604"> 604</a> _, pcf_cr, _ = average_pcfs(pcf_samples["prey_pred"])

<a href="#L-606"> 606</a> result["pcf_distances"] = dist.tolist()

<a href="#L-607"> 607</a> result["pcf_prey_prey"] = pcf_rr.tolist()

<a href="#L-608"> 608</a> result["pcf_pred_pred"] = pcf_cc.tolist()

<a href="#L-609"> 609</a> result["pcf_prey_pred"] = pcf_cr.tolist()

<a href="#L-611"> 611</a> # Short-range indices

<a href="#L-612"> 612</a> short_mask = dist < 3.0

<a href="#L-613"> 613</a> if np.any(short_mask):

<a href="#L-614"> 614</a> result["segregation_index"] = float(np.mean(pcf_cr[short_mask]))

<a href="#L-615"> 615</a> result["prey_clustering_index"] = float(np.mean(pcf_rr[short_mask]))

<a href="#L-616"> 616</a> result["pred_clustering_index"] = float(np.mean(pcf_cc[short_mask]))

<a href="#L-618"> 618</a> return result

<a href="#L-621"> 621</a># =============================================================================

<a href="#L-622"> 622</a># Experiment Phases

<a href="#L-623"> 623</a># =============================================================================

<a href="#L-626"> 626</a>def run_phase1(cfg: Config, output_dir: Path, logger: logging.Logger) -> List[Dict]:

<a href="#L-627"> 627</a> """

<a href="#L-628"> 628</a> Execute Phase 1 of the simulation: a parameter sweep to identify critical points.

<a href="#L-630"> 630</a> This function performs a 1D sweep across varying prey mortality rates while

<a href="#L-631"> 631</a> keeping other parameters fixed. It utilizes parallel execution via joblib

<a href="#L-632"> 632</a> and saves results incrementally to a JSONL file to ensure data integrity

<a href="#L-633"> 633</a> during long-running batches.

<a href="#L-635"> 635</a> Parameters

<a href="#L-636"> 636</a> ----------

<a href="#L-637"> 637</a> cfg : Config

<a href="#L-638"> 638</a> Configuration object containing simulation hyperparameters, sweep

<a href="#L-639"> 639</a> ranges, and execution settings (n_jobs, grid_size, etc.).

<a href="#L-640"> 640</a> output_dir : Path

<a href="#L-641"> 641</a> Directory where result files (JSONL) and metadata (JSON) will be stored.

<a href="#L-642"> 642</a> logger : logging.Logger

<a href="#L-643"> 643</a> Logger instance for tracking simulation progress and recording

<a href="#L-644"> 644</a> operational metadata.

<a href="#L-646"> 646</a> Returns

<a href="#L-647"> 647</a> -------

<a href="#L-648"> 648</a> all_results : list of dict

<a href="#L-649"> 649</a> A list of dictionaries containing the metrics collected from every

<a href="#L-650"> 650</a> individual simulation run in the sweep.

<a href="#L-652"> 652</a> Notes

<a href="#L-653"> 653</a> -----

<a href="#L-654"> 654</a> The function performs the following steps:

<a href="#L-655"> 655</a> 1. Pre-warms Numba kernels for performance.

<a href="#L-656"> 656</a> 2. Generates a deterministic set of simulation jobs using unique seeds.

<a href="#L-657"> 657</a> 3. Executes simulations in parallel using a generator for memory efficiency.

<a href="#L-658"> 658</a> 4. Records metadata including a timestamp and a serialized snapshot of

<a href="#L-659"> 659</a> the configuration.

<a href="#L-660"> 660</a> """

<a href="#L-661"> 661</a> from joblib import Parallel, delayed

<a href="#L-663"> 663</a> warmup_numba_kernels(cfg.grid_size, directed_hunting=cfg.directed_hunting)

<a href="#L-665"> 665</a> prey_deaths = cfg.get_prey_deaths()

<a href="#L-667"> 667</a> # Build job list

<a href="#L-668"> 668</a> jobs = []

<a href="#L-669"> 669</a> # Sweep through prey_death only (prey_birth is fixed)

<a href="#L-670"> 670</a> for pd in prey_deaths:

<a href="#L-671"> 671</a> for rep in range(cfg.n_replicates):

<a href="#L-672"> 672</a> params = {"pd": pd}

<a href="#L-674"> 674</a> seed = generate_unique_seed(params, rep)

<a href="#L-675"> 675</a> jobs.append(

<a href="#L-676"> 676</a> (

<a href="#L-677"> 677</a> cfg.prey_birth,

<a href="#L-678"> 678</a> pd,

<a href="#L-679"> 679</a> cfg.predator_birth,

<a href="#L-680"> 680</a> cfg.predator_death,

<a href="#L-681"> 681</a> cfg.grid_size,

<a href="#L-682"> 682</a> seed,

<a href="#L-683"> 683</a> cfg,

<a href="#L-684"> 684</a> False,

<a href="#L-685"> 685</a> )

<a href="#L-686"> 686</a> )

<a href="#L-688"> 688</a> logger.info(f"Phase 1: {len(jobs):,} simulations")

<a href="#L-689"> 689</a> logger.info(

<a href="#L-690"> 690</a> f" Grid: {cfg.n_prey_death} prey_death values × {cfg.n_replicates} reps (prey_birth={cfg.prey_birth})"

<a href="#L-691"> 691</a> )

<a href="#L-692"> 692</a> # Run with incremental saving

<a href="#L-693"> 693</a> output_jsonl = output_dir / "phase1_results.jsonl"

<a href="#L-694"> 694</a> all_results = []

<a href="#L-696"> 696</a> with open(output_jsonl, "w", encoding="utf-8") as f:

<a href="#L-697"> 697</a> executor = Parallel(n_jobs=cfg.n_jobs, return_as="generator")

<a href="#L-698"> 698</a> tasks = (delayed(run_single_simulation)(*job) for job in jobs)

<a href="#L-700"> 700</a> for result in tqdm(executor(tasks), total=len(jobs), desc="Phase 1"):

<a href="#L-701"> 701</a> f.write(json.dumps(result, default=str) + "\n")

<a href="#L-702"> 702</a> f.flush()

<a href="#L-703"> 703</a> all_results.append(result)

<a href="#L-705"> 705</a> # Save metadata

<a href="#L-706"> 706</a> meta = {

<a href="#L-707"> 707</a> "phase": 1,

<a href="#L-708"> 708</a> "description": "Parameter sweep for critical point",

<a href="#L-709"> 709</a> "n_sims": len(all_results),

<a href="#L-710"> 710</a> "timestamp": time.strftime("%Y-%m-%d %H:%M:%S"),

<a href="#L-711"> 711</a> "config": asdict(cfg),

<a href="#L-712"> 712</a> }

<a href="#L-713"> 713</a> with open(output_dir / "phase1_metadata.json", "w") as f:

<a href="#L-714"> 714</a> json.dump(meta, f, indent=2, default=str)

<a href="#L-716"> 716</a> logger.info(f"Phase 1 complete. Results: {output_jsonl}")

<a href="#L-717"> 717</a> return all_results

<a href="#L-720"> 720</a>def run_phase2(cfg: Config, output_dir: Path, logger: logging.Logger) -> List[Dict]:

<a href="#L-721"> 721</a> """

<a href="#L-722"> 722</a> Execute Phase 2 of the simulation: self-organization and criticality analysis.

<a href="#L-724"> 724</a> This phase tests the Self-Organized Criticality (SOC) hypothesis by

<a href="#L-725"> 725</a> initializing simulations at different points in the parameter space and

<a href="#L-726"> 726</a> observing whether evolutionary pressure drives the system toward a

<a href="#L-727"> 727</a> common critical point, regardless of initial prey mortality rates.

<a href="#L-729"> 729</a> Parameters

<a href="#L-730"> 730</a> ----------

<a href="#L-731"> 731</a> cfg : Config

<a href="#L-732"> 732</a> Configuration object containing simulation hyperparameters, evolution

<a href="#L-733"> 733</a> settings, and execution constraints.

<a href="#L-734"> 734</a> output_dir : Path

<a href="#L-735"> 735</a> Directory where result files (JSONL) and metadata (JSON) will be stored.

<a href="#L-736"> 736</a> logger : logging.Logger

<a href="#L-737"> 737</a> Logger instance for tracking progress and evolutionary convergence.

<a href="#L-739"> 739</a> Returns

<a href="#L-740"> 740</a> -------

<a href="#L-741"> 741</a> all_results : list of dict

<a href="#L-742"> 742</a> A list of dictionaries containing metrics from the evolutionary

<a href="#L-743"> 743</a> simulation runs.

<a href="#L-745"> 745</a> Notes

<a href="#L-746"> 746</a> -----

<a href="#L-747"> 747</a> The function captures:

<a href="#L-748"> 748</a> 1. Convergence of 'prey_death' across multiple replicates.

<a href="#L-749"> 749</a> 2. Final steady-state population distributions.

<a href="#L-750"> 750</a> 3. Incremental saving of results to prevent data loss.

<a href="#L-751"> 751</a> """

<a href="#L-752"> 752</a> from joblib import Parallel, delayed

<a href="#L-754"> 754</a> warmup_numba_kernels(cfg.grid_size, directed_hunting=cfg.directed_hunting)

<a href="#L-756"> 756</a> # Test at multiple prey_birth values

<a href="#L-757"> 757</a> pb = 0.2

<a href="#L-758"> 758</a> # Vary intial prey_death

<a href="#L-759"> 759</a> initial_prey_deaths = np.linspace(

<a href="#L-760"> 760</a> cfg.prey_death_range[0], cfg.prey_death_range[1], cfg.n_prey_death

<a href="#L-761"> 761</a> )

<a href="#L-763"> 763</a> jobs = []

<a href="#L-764"> 764</a> for initial_pd in initial_prey_deaths:

<a href="#L-765"> 765</a> for rep in range(cfg.n_replicates):

<a href="#L-766"> 766</a> params = {"pb": pb, "initial_pd": initial_pd, "phase": 2}

<a href="#L-767"> 767</a> seed = generate_unique_seed(params, rep)

<a href="#L-768"> 768</a> jobs.append(

<a href="#L-769"> 769</a> (

<a href="#L-770"> 770</a> pb,

<a href="#L-771"> 771</a> initial_pd,

<a href="#L-772"> 772</a> cfg.predator_birth,

<a href="#L-773"> 773</a> cfg.predator_death,

<a href="#L-774"> 774</a> cfg.grid_size,

<a href="#L-775"> 775</a> seed,

<a href="#L-776"> 776</a> cfg,

<a href="#L-777"> 777</a> True,

<a href="#L-778"> 778</a> )

<a href="#L-779"> 779</a> )

<a href="#L-781"> 781</a> logger.info(f"Phase 2: {len(jobs):,} simulations")

<a href="#L-782"> 782</a> logger.info(f" prey_birth value: {pb}")

<a href="#L-783"> 783</a> logger.info(f" initial prey_death values: {len(initial_prey_deaths)}")

<a href="#L-784"> 784</a> logger.info(f" Replicates: {cfg.n_replicates}")

<a href="#L-786"> 786</a> output_jsonl = output_dir / "phase2_results.jsonl"

<a href="#L-787"> 787</a> all_results = []

<a href="#L-789"> 789</a> with open(output_jsonl, "w", encoding="utf-8") as f:

<a href="#L-790"> 790</a> executor = Parallel(n_jobs=cfg.n_jobs, return_as="generator")

<a href="#L-791"> 791</a> tasks = (delayed(run_single_simulation)(*job) for job in jobs)

<a href="#L-793"> 793</a> for result in tqdm(executor(tasks), total=len(jobs), desc="Phase 2"):

<a href="#L-794"> 794</a> f.write(json.dumps(result, default=str) + "\n")

<a href="#L-795"> 795</a> f.flush()

<a href="#L-796"> 796</a> all_results.append(result)

<a href="#L-798"> 798</a> meta = {

<a href="#L-799"> 799</a> "phase": 2,

<a href="#L-800"> 800</a> "description": "Self-organization toward criticality",

<a href="#L-801"> 801</a> "n_sims": len(all_results),

<a href="#L-802"> 802</a> "initial_prey_deaths": initial_prey_deaths.tolist(),

<a href="#L-803"> 803</a> "timestamp": time.strftime("%Y-%m-%d %H:%M:%S"),

<a href="#L-804"> 804</a> }

<a href="#L-805"> 805</a> with open(output_dir / "phase2_metadata.json", "w") as f:

<a href="#L-806"> 806</a> json.dump(meta, f, indent=2, default=str)

<a href="#L-808"> 808</a> logger.info(f"Phase 2 complete. Results: {output_jsonl}")

<a href="#L-809"> 809</a> return all_results

<a href="#L-812"> 812</a>def run_phase3(cfg: Config, output_dir: Path, logger: logging.Logger) -> List[Dict]:

<a href="#L-813"> 813</a> """

<a href="#L-814"> 814</a> Phase 3: Finite-size scaling at critical point.

<a href="#L-816"> 816</a> - Multiple grid sizes at (critical_prey_birth, critical_prey_death)

<a href="#L-817"> 817</a> - Analyze cluster size cutoffs vs L

<a href="#L-818"> 818</a> """

<a href="#L-819"> 819</a> from joblib import Parallel, delayed

<a href="#L-821"> 821</a> # NOTE: Tuned to critical points from phase 1

<a href="#L-822"> 822</a> pb = cfg.critical_prey_birth

<a href="#L-823"> 823</a> pd = cfg.critical_prey_death

<a href="#L-825"> 825</a> logger.info(f"Phase 3: FSS at critical point (pb={pb}, pd={pd})")

<a href="#L-827"> 827</a> for L in cfg.grid_sizes:

<a href="#L-828"> 828</a> warmup_numba_kernels(L, directed_hunting=cfg.directed_hunting)

<a href="#L-830"> 830</a> jobs = []

<a href="#L-831"> 831</a> for L in cfg.grid_sizes: # Sweep through grid sizes

View remainder of file in raw view

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

FilesExpand file tree

experiments.html

Latest commit

History

experiments.html

File metadata and controls