Coverage for /opt/hostedtoolcache/Python/3.11.10/x64/lib/python3.11/site-packages/hypervehicle/utilities.py: 20%

1import os 

2import sys 

3import glob 

4import numpy as np 

5import pandas as pd 

6from stl import mesh 

7from tqdm import tqdm 

8from art import tprint, art 

9import xml.etree.ElementTree as ET 

10from typing import Dict, List, Optional, TYPE_CHECKING 

11 

12if TYPE_CHECKING: 

13 from .vehicle import Vehicle 

14 

15 

16def surfce_to_stl( 

17 parametric_surface, 

18 triangles_per_edge_r: int, 

19 triangles_per_edge_s: int, 

20 si: float = 1.0, 

21 sj: float = 1.0, 

22 mirror_y: bool = False, 

23 i_clustering_func: callable = None, 

24 j_clustering_func: callable = None, 

25 verbosity=0, 

26) -> mesh.Mesh: 

27 """ 

28 Function to convert parametric_surface generated using the Eilmer Geometry 

29 Package into a stl mesh object. 

30 

31 Parameters 

32 ---------- 

33 parametric_surface : Any 

34 The parametric surface object. 

35 

36 si : float, optional 

37 The clustering in the i-direction. The default is 1.0. 

38 

39 sj : float, optional 

40 The clustering in the j-direction. The default is 1.0. 

41 

42 triangles_per_edge_r : int 

43 The resolution for the stl object - x/i direction. 

44 

45 triangles_per_edge_s : int 

46 The resolution for the stl object - y/j direction. 

47 

48 mirror_y : bool, optional 

49 Create mirror image about x-z plane. The default is False. 

50 

51 i_clustering_func : callable, optional 

52 A custom clustering function to apply in the i direction. 

53 The default is None. 

54 

55 j_clustering_func : callable, optional 

56 A custom clustering function to apply in the j direction. 

57 The default is None. 

58 

59 Returns 

60 ---------- 

61 stl_mesh : Mesh 

62 The numpy-stl mesh. 

63 """ 

64 ni = triangles_per_edge_r 

65 nj = triangles_per_edge_s 

66 

67 if verbosity > 1: 

68 print(f"Triangles per edge: (ni, nj) = ({ni}, {nj})") 

69 

70 # Create list of vertices 

71 if i_clustering_func: 

72 r_list = [i_clustering_func(i) for i in np.linspace(0, 1, ni + 1)] 

73 else: 

74 r_list = default_vertex_func(lb=0.0, ub=1.0, steps=ni + 1, spacing=si) 

75 

76 if j_clustering_func: 

77 s_list = [j_clustering_func(i) for i in np.linspace(0, 1, ni + 1)] 

78 else: 

79 s_list = default_vertex_func(lb=0.0, ub=1.0, steps=nj + 1, spacing=sj) 

80 

81 y_mult: int = -1 if mirror_y else 1 

82 

83 if verbosity > 1: 

84 print(f"r_list = {r_list}") 

85 print(f"s_list = {s_list}") 

86 

87 # For stl generation we split each cell into 4x triangles with indece [0, 3] 

88 # p01-------p11 

89 # | \ 2 / | 

90 # | \ / | 

91 # | 3 pc 1 | 

92 # | / \ | 

93 # | / 0 \ | 

94 # p00-------p10 

95 

96 N_triangles = 4 * (ni) * (nj) 

97 stl_mesh = mesh.Mesh(np.zeros(N_triangles, dtype=mesh.Mesh.dtype)) 

98 

99 # mesh.vectors contains a list defining three corners of each triangle. 

100 t = 0 

101 

102 # For vertices along the x direction (i) 

103 for i in range(ni): 

104 # For vertices along the y direction (j) 

105 for j in range(nj): 

106 # Get corner points for current cell (quad) 

107 pos00 = parametric_surface(r_list[i], s_list[j]) 

108 pos10 = parametric_surface(r_list[i + 1], s_list[j]) 

109 pos01 = parametric_surface(r_list[i], s_list[j + 1]) 

110 pos11 = parametric_surface(r_list[i + 1], s_list[j + 1]) 

111 

112 # Get centre for current cell (quad) 

113 pos_x = 0.25 * (pos00.x + pos10.x + pos01.x + pos11.x) 

114 pos_y = 0.25 * (pos00.y + pos10.y + pos01.y + pos11.y) 

115 pos_z = 0.25 * (pos00.z + pos10.z + pos01.z + pos11.z) 

116 

117 # Add triangle 0 [p00, p10, pc] 

118 stl_mesh.vectors[t][0] = np.array([pos00.x, y_mult * pos00.y, pos00.z]) 

119 stl_mesh.vectors[t][1] = np.array([pos10.x, y_mult * pos10.y, pos10.z]) 

120 stl_mesh.vectors[t][2] = np.array([pos_x, y_mult * pos_y, pos_z]) 

121 t += 1 

122 

123 # Add triangle 1 [p10, p11, pc] 

124 stl_mesh.vectors[t][0] = np.array([pos10.x, y_mult * pos10.y, pos10.z]) 

125 stl_mesh.vectors[t][1] = np.array([pos11.x, y_mult * pos11.y, pos11.z]) 

126 stl_mesh.vectors[t][2] = np.array([pos_x, y_mult * pos_y, pos_z]) 

127 t += 1 

128 

129 # Add triangle 2 [p11, p01, pc] 

130 stl_mesh.vectors[t][0] = np.array([pos11.x, y_mult * pos11.y, pos11.z]) 

131 stl_mesh.vectors[t][1] = np.array([pos01.x, y_mult * pos01.y, pos01.z]) 

132 stl_mesh.vectors[t][2] = np.array([pos_x, y_mult * pos_y, pos_z]) 

133 t += 1 

134 

135 # Add triangle 3 [p01, p00, pc] 

136 stl_mesh.vectors[t][0] = np.array([pos01.x, y_mult * pos01.y, pos01.z]) 

137 stl_mesh.vectors[t][1] = np.array([pos00.x, y_mult * pos00.y, pos00.z]) 

138 stl_mesh.vectors[t][2] = np.array([pos_x, y_mult * pos_y, pos_z]) 

139 t += 1 

140 

141 return stl_mesh 

142 

143 

144def default_vertex_func(lb, ub, steps, spacing=1.0): 

145 span = ub - lb 

146 dx = 1.0 / (steps - 1) 

147 return np.array([lb + (i * dx) ** spacing * span for i in range(steps)]) 

148 

149 

150def assess_inertial_properties( 

151 vehicle: "Vehicle", component_densities: Dict[str, float] 

152): 

153 """ 

154 

155 Parameters 

156 ---------- 

157 vehicle : Vehicle 

158 A hypervehicle Vehicle instance. 

159 

160 component_densities : Dict[str, float] 

161 A dictionary containing the effective densities for each component. 

162 Note that the keys of the dict must match the keys of 

163 vehicle._named_components. 

164 

165 Returns 

166 ------- 

167 vehicle_properties : dict 

168 A dictionary containing the vehicle's mass, volume, location of 

169 CoG and moment of inertia matrix. 

170 

171 component_properties : dict 

172 A dictionary containing the same keys as vehicle_properties, but 

173 the values are now dictionaries for each component of the vehicle. 

174 """ 

175 # Check if vehicle has been generated 

176 if not vehicle._generated: 

177 vehicle.generate() 

178 

179 volumes = {} 

180 areas = {} 

181 masses = {} 

182 cgs = {} 

183 inertias = {} 

184 total_mass = 0 

185 total_volume = 0 

186 total_area = 0 

187 

188 for name, component in vehicle._named_components.items(): 

189 # Estiamte inertial properties 

190 volume, vmass, cog, inertia = component.mesh.get_mass_properties_with_density( 

191 component_densities[name] 

192 ) 

193 

194 # Also calculate surface area 

195 area = float(sum(component.mesh.areas)) 

196 

197 # Save component result 

198 volumes[name] = volume 

199 areas[name] = area 

200 masses[name] = vmass 

201 cgs[name] = cog 

202 inertias[name] = inertia 

203 total_mass += vmass 

204 total_volume += volume 

205 total_area += area 

206 

207 # Composite centre of mass 

208 composite_cog = 0 

209 for component in vehicle._named_components: 

210 m = masses[component] 

211 composite_cog += m * cgs[component] 

212 

213 composite_cog *= 1 / total_mass 

214 

215 # Parallel axis theorem 

216 shifted_inertias = {} 

217 composite_inertia = 0 

218 for component in vehicle._named_components: 

219 m = masses[component] 

220 r = cgs[component] - composite_cog 

221 I_adj = inertias[component] + m * r**2 

222 

223 shifted_inertias[component] = I_adj 

224 composite_inertia += I_adj 

225 

226 # Prepare output 

227 vehicle_properties = { 

228 "mass": total_mass, 

229 "volume": total_volume, 

230 "area": total_area, 

231 "cog": composite_cog, 

232 "moi": composite_inertia, 

233 } 

234 component_properties = { 

235 "mass": masses, 

236 "volume": volumes, 

237 "area": areas, 

238 "cog": cgs, 

239 "moi": inertias, 

240 } 

241 

242 return vehicle_properties, component_properties 

243 

244 

245class SensitivityStudy: 

246 """ 

247 Computes the geometric sensitivities using finite differencing. 

248 """ 

249 

250 def __init__(self, vehicle_constructor, verbosity: Optional[int] = 1): 

251 """Vehicle geometry sensitivity constructor. 

252 

253 Parameters 

254 ---------- 

255 vehicle_constructor : AbstractGenerator 

256 The Vehicle instance constructor. 

257 

258 verbosity : int, optional 

259 The code verbosity. The default is 1. 

260 

261 Returns 

262 ------- 

263 VehicleSensitivity object. 

264 """ 

265 self.vehicle_constructor = vehicle_constructor 

266 self.verbosity = verbosity 

267 

268 # Parameter sensitivities 

269 self.parameter_sensitivities = None 

270 self.component_sensitivities = None 

271 self.component_scalar_sensitivities = None 

272 self.scalar_sensitivities = None 

273 self.property_sensitivities = None 

274 

275 # Nominal vehicle instance 

276 self.nominal_vehicle_instance = None 

277 

278 # Combined data file name 

279 self.combined_fn = "all_components_sensitivity.csv" 

280 

281 def __repr__(self): 

282 return "HyperVehicle sensitivity study" 

283 

284 def dvdp( 

285 self, 

286 parameter_dict: dict[str, any], 

287 overrides: Optional[dict[str, any]] = None, 

288 perturbation: Optional[float] = 5, 

289 write_nominal_stl: Optional[bool] = True, 

290 nominal_stl_prefix: Optional[str] = None, 

291 ): 

292 """Computes the sensitivity of the geometry with respect to the 

293 parameters. 

294 

295 Parameters 

296 ---------- 

297 parameter_dict : dict 

298 A dictionary of the design parameters to perturb, and their 

299 nominal values. 

300 

301 overrides : dict, optional 

302 Optional vehicle generator overrides to provide along with the 

303 parameter dictionary without variation. The default is None. 

304 

305 perturbation : float, optional 

306 The design parameter perturbation amount, specified as percentage. 

307 The default is 20. 

308 

309 vehicle_creator_method : str, optional 

310 The name of the method which returns a hypervehicle.Vehicle 

311 instance, ready for generation. The default is 'create_instance'. 

312 

313 write_nominal_stl : bool, optional 

314 A boolean flag to write the nominal geometry STL(s) to file. The 

315 default is True. 

316 

317 nominal_stl_prefix : str, optional 

318 The prefix to append when writing STL files for the nominal geometry. 

319 If None, no prefix will be used. The default is None. 

320 

321 Returns 

322 ------- 

323 sensitivities : dict 

324 A dictionary containing the sensitivity information for all 

325 components of the geometry, relative to the nominal geometry. 

326 """ 

327 # Print banner 

328 if self.verbosity > 0: 

329 print_banner() 

330 print("Running geometric sensitivity study.") 

331 

332 # TODO - return perturbed instances? After generatation to allow 

333 # quickly writing to STL 

334 from hypervehicle.generator import AbstractGenerator 

335 

336 # Check overrides 

337 overrides = overrides if overrides else {} 

338 

339 # Create Vehicle instance with nominal parameters 

340 if self.verbosity > 0: 

341 print(" Generating nominal geometry...") 

342 

343 constructor_instance: AbstractGenerator = self.vehicle_constructor( 

344 **parameter_dict, **overrides 

345 ) 

346 nominal_instance = constructor_instance.create_instance() 

347 nominal_instance.verbosity = 0 

348 

349 # Generate components 

350 nominal_instance.generate() 

351 nominal_meshes = { 

352 name: component.mesh 

353 for name, component in nominal_instance._named_components.items() 

354 } 

355 

356 if self.verbosity > 0: 

357 print(" Done.") 

358 

359 if write_nominal_stl: 

360 # Write nominal instance to STL files 

361 nominal_instance.to_stl(prefix=nominal_stl_prefix) 

362 

363 # Generate meshes for each parameter 

364 if self.verbosity > 0: 

365 print(" Generating perturbed geometries...") 

366 print(" Parameters: ", parameter_dict.keys()) 

367 

368 sensitivities = {} 

369 analysis_sens = {} 

370 component_analysis_sens = {} 

371 property_sens = {} 

372 for parameter, value in parameter_dict.items(): 

373 if self.verbosity > 0: 

374 print(f" Generating for {parameter}.") 

375 

376 sensitivities[parameter] = {} 

377 

378 # Create copy 

379 adjusted_parameters = parameter_dict.copy() 

380 

381 # Adjust current parameter for sensitivity analysis 

382 adjusted_parameters[parameter] *= 1 + perturbation / 100 

383 dp = adjusted_parameters[parameter] - value 

384 

385 # Create Vehicle instance with perturbed parameter 

386 constructor_instance = self.vehicle_constructor( 

387 **adjusted_parameters, **overrides 

388 ) 

389 parameter_instance = constructor_instance.create_instance() 

390 parameter_instance.verbosity = 0 

391 

392 # Generate stl meshes 

393 parameter_instance.generate() 

394 parameter_meshes = { 

395 name: component.mesh 

396 for name, component in parameter_instance._named_components.items() 

397 } 

398 

399 # Generate sensitivities for geometric analysis results 

400 if nominal_instance.analysis_results: 

401 analysis_sens[parameter] = {} 

402 for r, v in nominal_instance.analysis_results.items(): 

403 analysis_sens[parameter][r] = ( 

404 parameter_instance.analysis_results[r] - v 

405 ) / dp 

406 

407 # Repeat for components 

408 component_analysis_sens[parameter] = ( 

409 parameter_instance._volmass - nominal_instance._volmass 

410 ) / dp 

411 

412 # Generate sensitivities for vehicle properties 

413 if nominal_instance.properties: 

414 property_sens[parameter] = {} 

415 for property, v in nominal_instance.properties.items(): 

416 property_sens[parameter][property] = ( 

417 parameter_instance.properties[property] - v 

418 ) / dp 

419 

420 # Generate sensitivities 

421 for component, nominal_mesh in nominal_meshes.items(): 

422 parameter_mesh = parameter_meshes[component] 

423 sensitivity_df = self._compare_meshes( 

424 nominal_mesh, 

425 parameter_mesh, 

426 dp, 

427 parameter, 

428 ) 

429 

430 sensitivities[parameter][component] = sensitivity_df 

431 

432 if self.verbosity > 0: 

433 print(" Done.") 

434 

435 # Return output 

436 self.parameter_sensitivities = sensitivities 

437 self.scalar_sensitivities = analysis_sens 

438 self.component_scalar_sensitivities = component_analysis_sens 

439 self.property_sensitivities = property_sens 

440 self.component_sensitivities = self._combine(nominal_instance, sensitivities) 

441 self.nominal_vehicle_instance = nominal_instance 

442 

443 if self.verbosity > 0: 

444 print("Sensitivity study complete.") 

445 

446 return sensitivities 

447 

448 def to_csv(self, outdir: Optional[str] = None): 

449 """Writes the sensitivity information to CSV file. 

450 

451 Parameters 

452 ---------- 

453 outdir : str, optional 

454 The output directory to write the sensitivity files to. If 

455 None, the current working directory will be used. The default 

456 is None. 

457 

458 Returns 

459 ------- 

460 combined_data_filepath : str 

461 The filepath to the combined sensitivity data. 

462 """ 

463 # Check if sensitivities have been generated 

464 if self.component_sensitivities is None: 

465 raise Exception("Sensitivities have not yet been generated.") 

466 

467 else: 

468 # Check output directory 

469 if outdir is None: 

470 outdir = os.getcwd() 

471 

472 if not os.path.exists(outdir): 

473 # Make the directory 

474 os.mkdir(outdir) 

475 

476 # Save sensitivity data for each component 

477 all_sens_data = pd.DataFrame() 

478 for component, df in self.component_sensitivities.items(): 

479 df.to_csv( 

480 os.path.join(outdir, f"{component}_sensitivity.csv"), index=False 

481 ) 

482 all_sens_data = pd.concat([all_sens_data, df]) 

483 

484 # Also save the combined sensitivity data 

485 combined_data_path = os.path.join(outdir, self.combined_fn) 

486 all_sens_data.to_csv(combined_data_path, index=False) 

487 

488 # Also save analysis sensitivities 

489 if self.scalar_sensitivities: 

490 # Make analysis results directory 

491 properties_dir = os.path.join(outdir, f"scalar_sensitivities") 

492 if not os.path.exists(properties_dir): 

493 os.mkdir(properties_dir) 

494 

495 # Save volume and mass 

496 reformatted_results = {} 

497 for p, s in self.component_scalar_sensitivities.items(): 

498 labels = [] 

499 values = [] 

500 s: pd.DataFrame 

501 for component, comp_sens in s.iterrows(): 

502 comp_sens: pd.Series 

503 for i, j in comp_sens.items(): 

504 labels.append(f"{component}_{i}") 

505 values.append(j) 

506 

507 reformatted_results[p] = values 

508 

509 # Convert to DataFrame and save 

510 comp_sens = pd.DataFrame(data=reformatted_results, index=labels) 

511 comp_sens.to_csv( 

512 os.path.join(properties_dir, "volmass_sensitivity.csv") 

513 ) 

514 

515 # Save others 

516 for param in self.scalar_sensitivities: 

517 self.scalar_sensitivities[param]["cog"].tofile( 

518 os.path.join(properties_dir, f"{param}_cog_sensitivity.txt"), 

519 sep=", ", 

520 ) 

521 self.scalar_sensitivities[param]["moi"].tofile( 

522 os.path.join(properties_dir, f"{param}_moi_sensitivity.txt"), 

523 sep=", ", 

524 ) 

525 

526 # Also save user-defined property sensitivities 

527 if self.property_sensitivities: 

528 properties_dir = os.path.join(outdir, f"scalar_sensitivities") 

529 if not os.path.exists(properties_dir): 

530 os.mkdir(properties_dir) 

531 

532 pd.DataFrame(self.property_sensitivities).to_csv( 

533 os.path.join(properties_dir, "property_sensitivity.csv") 

534 ) 

535 

536 return combined_data_path 

537 

538 @staticmethod 

539 def _compare_meshes(mesh1, mesh2, dp, parameter_name: str) -> pd.DataFrame: 

540 """Compares two meshes with each other and applies finite differencing 

541 to quantify their differences. 

542 

543 Parameters 

544 ---------- 

545 mesh1 : Mesh 

546 The reference mesh. 

547 

548 mesh1 : Mesh 

549 The perturbed mesh. 

550 

551 dp : float 

552 The parameter perturbation. 

553 

554 parameter_name : str 

555 The name of the parameter. 

556 

557 Returns 

558 -------- 

559 df : pd.DataFrame 

560 A DataFrame of the finite difference results. 

561 """ 

562 # Take the vector difference 

563 diff = mesh2.vectors - mesh1.vectors 

564 

565 # Resize difference array to flatten 

566 shape = diff.shape 

567 flat_diff = diff.reshape((shape[0] * shape[2], shape[1])) 

568 

569 # Also flatten the reference mesh vectors 

570 vectors = mesh1.vectors.reshape((shape[0] * shape[2], shape[1])) 

571 

572 # Concatenate all data column-wise 

573 all_data = np.zeros((shape[0] * shape[2], shape[1] * 2)) 

574 all_data[:, 0:3] = vectors # Reference locations 

575 all_data[:, 3:6] = flat_diff # Location deltas 

576 

577 # Create DataFrame 

578 df = pd.DataFrame(data=all_data, columns=["x", "y", "z", "dx", "dy", "dz"]) 

579 df["magnitude"] = np.sqrt(np.square(df[["dx", "dy", "dz"]]).sum(axis=1)) 

580 

581 # Sensitivity calculations 

582 sensitivities = df[["dx", "dy", "dz"]] / dp 

583 sensitivities.rename( 

584 columns={ 

585 "dx": f"dxd{parameter_name}", 

586 "dy": f"dyd{parameter_name}", 

587 "dz": f"dzd{parameter_name}", 

588 }, 

589 inplace=True, 

590 ) 

591 

592 # Merge dataframes 

593 df = df.merge(sensitivities, left_index=True, right_index=True) 

594 

595 # Delete duplicate vertices 

596 df = df[~df.duplicated()] 

597 

598 return df 

599 

600 @staticmethod 

601 def _combine(nominal_instance, sensitivities): 

602 """Combines the sensitivity information for multiple parameters.""" 

603 component_names = nominal_instance._named_components.keys() 

604 params = list(sensitivities.keys()) 

605 

606 allsens = {} 

607 for component in component_names: 

608 df = sensitivities[params[0]][component][["x", "y", "z"]] 

609 for param in params: 

610 p_s = sensitivities[param][component][ 

611 [f"dxd{param}", f"dyd{param}", f"dzd{param}"] 

612 ] 

613 df = pd.concat([df, p_s], axis=1) 

614 

615 allsens[component] = df 

616 

617 return allsens 

618 

619 

620def append_sensitivities_to_tri( 

621 dp_filenames: List[str], 

622 components_filepath: Optional[str] = "Components.i.tri", 

623 match_tolerance: Optional[float] = 1e-5, 

624 rounding_tolerance: Optional[float] = 1e-8, 

625 combined_sens_fn: Optional[str] = "all_components_sensitivity.csv", 

626 outdir: Optional[str] = None, 

627 verbosity: Optional[int] = 1, 

628) -> float: 

629 """Appends shape sensitivity data to .i.tri file, and writes the sensitivity 

630 data to csv file too. This step is required for geometries with multiple 

631 components. The .tri file is used to match individual sensitivity files 

632 (dp_filenames) to the geometry. The combined sensitivity file is required 

633 to calculate flow sensitivities with the .plt file, which has local flow 

634 conditions attached. 

635 

636 Parameters 

637 ---------- 

638 dp_files : list[str] 

639 A list of the file names of the sensitivity data. 

640 

641 components_filepath : str, optional 

642 The filepath to the .tri file to be appended to. The default is 

643 'Components.i.tri'. 

644 

645 match_tolerance : float, optional 

646 The precision tolerance for matching point coordinates. The 

647 default is 1e-5. 

648 

649 rounding_tolerance : float, optional 

650 The tolerance to round data off to. The default is 1e-8. 

651 

652 combined_sens_fn : str, optional 

653 The filename of the combined geometry sensitivity data. The default 

654 is "all_components_sensitivity.csv". 

655 

656 outdir : str, optional 

657 The output directory to write the combined sensitivity file to. If 

658 None, the current working directory will be used. The default 

659 is None. 

660 

661 verbosity : int, optional 

662 The verbosity of the code. The defualt is 1. 

663 

664 Returns 

665 --------- 

666 match_fraction : float 

667 The fraction of cells which got matched. If this is below 100%, try 

668 decreasing the match tolerance and run again. 

669 

670 Examples 

671 --------- 

672 >>> dp_files = ['wing_0_body_width_sensitivity.csv', 

673 'wing_1_body_width_sensitivity.csv'] 

674 """ 

675 # Check outdir 

676 if outdir is None: 

677 outdir = os.getcwd() 

678 

679 else: 

680 # Desired outdir provided, check it exists 

681 if not os.path.exists(outdir): 

682 # Make the directory 

683 os.mkdir(outdir) 

684 

685 # TODO - rename to 'combine sensitivity' or "combine_comp_sens" 

686 # Parse .tri file 

687 tree = ET.parse(components_filepath) 

688 root = tree.getroot() 

689 grid = root[0] 

690 piece = grid[0] 

691 points = piece[0] 

692 

693 points_data = points[0].text 

694 

695 points_data_list = [el.split() for el in points_data.splitlines()] 

696 points_data_list = [[float(j) for j in i] for i in points_data_list] 

697 

698 points_df = pd.DataFrame(points_data_list, columns=["x", "y", "z"]).dropna() 

699 

700 # Ensure previous components sensitivity file is not included 

701 try: 

702 del dp_filenames[dp_filenames.index(combined_sens_fn)] 

703 except ValueError: 

704 # It is not in there 

705 pass 

706 

707 # Load and concatenate sensitivity data across components 

708 dp_df = pd.DataFrame() 

709 for filename in dp_filenames: 

710 df = pd.read_csv(filename) 

711 dp_df = pd.concat([dp_df, df]) 

712 

713 # Extract parameters 

714 parameters = [] 

715 param_cols = dp_df.columns[3:] 

716 for i in range(int(len(param_cols) / 3)): 

717 parameters.append(param_cols[int(i * 3)].split("dxd")[-1]) 

718 

719 # Match points_df to sensitivity df 

720 if verbosity > 0: 

721 print("Running coordinate matching algorithm for sensitivities...") 

722 pbar = tqdm( 

723 total=len(points_df), position=0, leave=True, desc="Point matching progress" 

724 ) 

725 data_str = "\n " 

726 param_data = dict(zip(parameters, ["\n " for _ in parameters])) 

727 all_data = np.zeros((len(points_df), len(param_cols)), dtype=float) 

728 matched_points = 0 

729 for i in range(len(points_df)): 

730 match_x = (points_df["x"].iloc[i] - dp_df["x"]).abs() < match_tolerance 

731 match_y = (points_df["y"].iloc[i] - dp_df["y"]).abs() < match_tolerance 

732 match_z = (points_df["z"].iloc[i] - dp_df["z"]).abs() < match_tolerance 

733 

734 match = match_x & match_y & match_z 

735 try: 

736 # TODO - what if there are multiple matches? (due to intersect perturbations) 

737 matched_data = dp_df[match].iloc[0][param_cols].values 

738 

739 # Round off infinitesimally small values 

740 matched_data[abs(matched_data) < rounding_tolerance] = 0 

741 

742 # Update data 

743 all_data[i, :] = matched_data 

744 p_n = 0 

745 for parameter, data_str in param_data.items(): 

746 line = "" 

747 for j in range(3): 

748 line += f"\t{matched_data[j+3*p_n]:.14e}" 

749 line += "\n " 

750 

751 data_str += line 

752 param_data[parameter] = data_str 

753 p_n += 1 

754 

755 # Count matched points 

756 matched_points += 1 

757 

758 except IndexError: 

759 # No match found, append zeros to maintain order 

760 line = f"\t{0:.14e}\t{0:.14e}\t{0:.14e}\n " 

761 

762 # Update data string for each parameter 

763 p_n = 0 

764 for parameter, data_str in param_data.items(): 

765 param_data[parameter] = data_str + line 

766 p_n += 1 

767 

768 # Update progress bar 

769 if verbosity > 0: 

770 pbar.update(1) 

771 

772 match_fraction = matched_points / len(points_df) 

773 if verbosity > 0: 

774 pbar.close() 

775 print(f"Done - matched {100*match_fraction:.2f}% of points.") 

776 

777 # Write combined sensitivity data to CSV 

778 combined_sense = pd.merge( 

779 left=points_df.reset_index(), 

780 right=pd.DataFrame(all_data, columns=param_cols), 

781 left_index=True, 

782 right_index=True, 

783 ).drop("index", axis=1) 

784 combined_sense.to_csv(os.path.join(outdir, combined_sens_fn), index=False) 

785 

786 # Write the matched sensitivity df to i.tri file as new xml element 

787 # NumberOfComponents is how many sensitivity components there are (3 for x,y,z) 

788 for parameter in parameters: 

789 attribs = { 

790 "Name": f"{parameter}_sens", 

791 "NumberOfComponents": "3", 

792 "type": "Float64", 

793 "format": "ascii", 

794 "TRIXtype": "SHAPE_LINEARIZATION", 

795 } 

796 PointData = ET.SubElement(piece, "PointData") 

797 PointDataArray = ET.SubElement(PointData, "DataArray", attribs) 

798 PointDataArray.text = param_data[parameter] 

799 

800 # Save to file 

801 tree.write(components_filepath) 

802 

803 return match_fraction 

804 

805 

806def csv_to_delaunay(filepath: str): 

807 """Converts a csv file of points to a Delaunay3D surface. 

808 

809 Parameters 

810 ------------ 

811 filepath : str 

812 The filepath to the CSV file. 

813 """ 

814 # TODO - rename 

815 from paraview.simple import CSVReader, TableToPoints, Delaunay3D, SaveData # type: ignore 

816 

817 root_dir = os.path.dirname(filepath) 

818 prefix = filepath.split(os.sep)[-1].split(".csv")[0] 

819 savefilename = os.path.join(root_dir, f"{prefix}.vtu") 

820 

821 # create a new 'CSV Reader' 

822 fin_0_L_b_sensitivitycsv = CSVReader(FileName=[filepath]) 

823 

824 # create a new 'Table To Points' 

825 tableToPoints1 = TableToPoints(Input=fin_0_L_b_sensitivitycsv) 

826 

827 # Properties modified on tableToPoints1 

828 tableToPoints1.XColumn = "x" 

829 tableToPoints1.YColumn = "y" 

830 tableToPoints1.ZColumn = "z" 

831 

832 # create a new 'Delaunay 3D' 

833 delaunay3D1 = Delaunay3D(Input=tableToPoints1) 

834 

835 # save data 

836 SaveData(savefilename, proxy=delaunay3D1) 

837 

838 

839def convert_all_csv_to_delaunay(directory: str = ""): 

840 # TODO - rename 

841 # TODO - docstrings 

842 # TODO - specify outdir 

843 files = glob.glob(os.path.join(directory, "*.csv")) 

844 

845 if len(files) == 0: 

846 print(f"No CSV files in directory {directory}.") 

847 sys.exit() 

848 

849 for file in files: 

850 csv_to_delaunay(file) 

851 

852 

853def merge_stls( 

854 stl_files: List[str], name: Optional[str] = None, verbosity: Optional[int] = 1 

855) -> str: 

856 """Merge STL files into a single file. Note that this function 

857 depends on having PyMesh installed. 

858 

859 Parameters 

860 ---------- 

861 stl_files : list[str] 

862 A list of the STL file names to be merged. 

863 

864 name : str, optional 

865 The prefix of the combined STL filename output. 

866 

867 verbosity : int, optional 

868 The function verbosity. The default is 1. 

869 

870 Returns