[8a20be5] | 1 | #!/usr/bin/env python |
---|
| 2 | # -*- coding: utf-8 -*- |
---|
| 3 | |
---|
[87985ca] | 4 | import sys |
---|
| 5 | import math |
---|
[d547f16] | 6 | from os.path import basename, dirname, join as joinpath |
---|
| 7 | import glob |
---|
[7cf2cfd] | 8 | import datetime |
---|
[87985ca] | 9 | |
---|
[1726b21] | 10 | import numpy as np |
---|
[473183c] | 11 | |
---|
[29fc2a3] | 12 | ROOT = dirname(__file__) |
---|
| 13 | sys.path.insert(0, ROOT) # Make sure sasmodels is first on the path |
---|
| 14 | |
---|
| 15 | |
---|
[e922c5d] | 16 | from . import core |
---|
| 17 | from . import kerneldll |
---|
| 18 | from . import models |
---|
| 19 | from .data import plot_theory, empty_data1D, empty_data2D |
---|
| 20 | from .direct_model import DirectModel |
---|
| 21 | from .convert import revert_model |
---|
[750ffa5] | 22 | kerneldll.ALLOW_SINGLE_PRECISION_DLLS = True |
---|
[87985ca] | 23 | |
---|
[d547f16] | 24 | # List of available models |
---|
| 25 | MODELS = [basename(f)[:-3] |
---|
[e922c5d] | 26 | for f in sorted(glob.glob(joinpath(ROOT,"models","[a-zA-Z]*.py")))] |
---|
[d547f16] | 27 | |
---|
[7cf2cfd] | 28 | # CRUFT python 2.6 |
---|
| 29 | if not hasattr(datetime.timedelta, 'total_seconds'): |
---|
| 30 | def delay(dt): |
---|
| 31 | """Return number date-time delta as number seconds""" |
---|
| 32 | return dt.days * 86400 + dt.seconds + 1e-6 * dt.microseconds |
---|
| 33 | else: |
---|
| 34 | def delay(dt): |
---|
| 35 | """Return number date-time delta as number seconds""" |
---|
| 36 | return dt.total_seconds() |
---|
| 37 | |
---|
| 38 | |
---|
| 39 | def tic(): |
---|
| 40 | """ |
---|
| 41 | Timer function. |
---|
| 42 | |
---|
| 43 | Use "toc=tic()" to start the clock and "toc()" to measure |
---|
| 44 | a time interval. |
---|
| 45 | """ |
---|
| 46 | then = datetime.datetime.now() |
---|
| 47 | return lambda: delay(datetime.datetime.now() - then) |
---|
| 48 | |
---|
| 49 | |
---|
| 50 | def set_beam_stop(data, radius, outer=None): |
---|
| 51 | """ |
---|
| 52 | Add a beam stop of the given *radius*. If *outer*, make an annulus. |
---|
| 53 | |
---|
| 54 | Note: this function does not use the sasview package |
---|
| 55 | """ |
---|
| 56 | if hasattr(data, 'qx_data'): |
---|
| 57 | q = np.sqrt(data.qx_data**2 + data.qy_data**2) |
---|
| 58 | data.mask = (q < radius) |
---|
| 59 | if outer is not None: |
---|
| 60 | data.mask |= (q >= outer) |
---|
| 61 | else: |
---|
| 62 | data.mask = (data.x < radius) |
---|
| 63 | if outer is not None: |
---|
| 64 | data.mask |= (data.x >= outer) |
---|
| 65 | |
---|
[8a20be5] | 66 | |
---|
[aa4946b] | 67 | def sasview_model(model_definition, **pars): |
---|
[87985ca] | 68 | """ |
---|
| 69 | Load a sasview model given the model name. |
---|
| 70 | """ |
---|
| 71 | # convert model parameters from sasmodel form to sasview form |
---|
| 72 | #print "old",sorted(pars.items()) |
---|
[aa4946b] | 73 | modelname, pars = revert_model(model_definition, pars) |
---|
[87985ca] | 74 | #print "new",sorted(pars.items()) |
---|
[87c722e] | 75 | sas = __import__('sas.models.'+modelname) |
---|
| 76 | ModelClass = getattr(getattr(sas.models,modelname,None),modelname,None) |
---|
[8a20be5] | 77 | if ModelClass is None: |
---|
[87c722e] | 78 | raise ValueError("could not find model %r in sas.models"%modelname) |
---|
[8a20be5] | 79 | model = ModelClass() |
---|
| 80 | |
---|
| 81 | for k,v in pars.items(): |
---|
| 82 | if k.endswith("_pd"): |
---|
| 83 | model.dispersion[k[:-3]]['width'] = v |
---|
| 84 | elif k.endswith("_pd_n"): |
---|
| 85 | model.dispersion[k[:-5]]['npts'] = v |
---|
| 86 | elif k.endswith("_pd_nsigma"): |
---|
| 87 | model.dispersion[k[:-10]]['nsigmas'] = v |
---|
[87985ca] | 88 | elif k.endswith("_pd_type"): |
---|
| 89 | model.dispersion[k[:-8]]['type'] = v |
---|
[8a20be5] | 90 | else: |
---|
| 91 | model.setParam(k, v) |
---|
| 92 | return model |
---|
| 93 | |
---|
[87985ca] | 94 | def randomize(p, v): |
---|
| 95 | """ |
---|
| 96 | Randomizing parameter. |
---|
| 97 | |
---|
| 98 | Guess the parameter type from name. |
---|
| 99 | """ |
---|
| 100 | if any(p.endswith(s) for s in ('_pd_n','_pd_nsigma','_pd_type')): |
---|
| 101 | return v |
---|
| 102 | elif any(s in p for s in ('theta','phi','psi')): |
---|
| 103 | # orientation in [-180,180], orientation pd in [0,45] |
---|
| 104 | if p.endswith('_pd'): |
---|
| 105 | return 45*np.random.rand() |
---|
| 106 | else: |
---|
| 107 | return 360*np.random.rand() - 180 |
---|
| 108 | elif 'sld' in p: |
---|
| 109 | # sld in in [-0.5,10] |
---|
| 110 | return 10.5*np.random.rand() - 0.5 |
---|
| 111 | elif p.endswith('_pd'): |
---|
| 112 | # length pd in [0,1] |
---|
| 113 | return np.random.rand() |
---|
| 114 | else: |
---|
[b89f519] | 115 | # values from 0 to 2*x for all other parameters |
---|
| 116 | return 2*np.random.rand()*(v if v != 0 else 1) |
---|
[87985ca] | 117 | |
---|
[216a9e1] | 118 | def randomize_model(name, pars, seed=None): |
---|
| 119 | if seed is None: |
---|
| 120 | seed = np.random.randint(1e9) |
---|
| 121 | np.random.seed(seed) |
---|
| 122 | # Note: the sort guarantees order of calls to random number generator |
---|
| 123 | pars = dict((p,randomize(p,v)) for p,v in sorted(pars.items())) |
---|
| 124 | # The capped cylinder model has a constraint on its parameters |
---|
| 125 | if name == 'capped_cylinder' and pars['cap_radius'] < pars['radius']: |
---|
| 126 | pars['radius'],pars['cap_radius'] = pars['cap_radius'],pars['radius'] |
---|
| 127 | return pars, seed |
---|
| 128 | |
---|
[87985ca] | 129 | def parlist(pars): |
---|
| 130 | return "\n".join("%s: %s"%(p,v) for p,v in sorted(pars.items())) |
---|
| 131 | |
---|
| 132 | def suppress_pd(pars): |
---|
| 133 | """ |
---|
| 134 | Suppress theta_pd for now until the normalization is resolved. |
---|
| 135 | |
---|
| 136 | May also suppress complete polydispersity of the model to test |
---|
| 137 | models more quickly. |
---|
| 138 | """ |
---|
| 139 | for p in pars: |
---|
| 140 | if p.endswith("_pd"): pars[p] = 0 |
---|
| 141 | |
---|
[7cf2cfd] | 142 | def eval_sasview(model_definition, pars, data, Nevals=1): |
---|
[346bc88] | 143 | from sas.models.qsmearing import smear_selection |
---|
[7cf2cfd] | 144 | model = sasview_model(model_definition, **pars) |
---|
[346bc88] | 145 | smearer = smear_selection(data, model=model) |
---|
[0763009] | 146 | value = None # silence the linter |
---|
[216a9e1] | 147 | toc = tic() |
---|
[0763009] | 148 | for _ in range(max(Nevals, 1)): # make sure there is at least one eval |
---|
[216a9e1] | 149 | if hasattr(data, 'qx_data'): |
---|
[346bc88] | 150 | q = np.sqrt(data.qx_data**2 + data.qy_data**2) |
---|
| 151 | index = ((~data.mask) & (~np.isnan(data.data)) |
---|
| 152 | & (q >= data.qmin) & (q <= data.qmax)) |
---|
| 153 | if smearer is not None: |
---|
| 154 | smearer.model = model # because smear_selection has a bug |
---|
[3e6aaad] | 155 | smearer.accuracy = data.accuracy |
---|
[346bc88] | 156 | smearer.set_index(index) |
---|
| 157 | value = smearer.get_value() |
---|
| 158 | else: |
---|
| 159 | value = model.evalDistribution([data.qx_data[index], data.qy_data[index]]) |
---|
[216a9e1] | 160 | else: |
---|
| 161 | value = model.evalDistribution(data.x) |
---|
[346bc88] | 162 | if smearer is not None: |
---|
| 163 | value = smearer(value) |
---|
[216a9e1] | 164 | average_time = toc()*1000./Nevals |
---|
| 165 | return value, average_time |
---|
| 166 | |
---|
[0763009] | 167 | def eval_opencl(model_definition, pars, data, dtype='single', Nevals=1, cutoff=0.): |
---|
[216a9e1] | 168 | try: |
---|
[aa4946b] | 169 | model = core.load_model(model_definition, dtype=dtype, platform="ocl") |
---|
[216a9e1] | 170 | except Exception,exc: |
---|
| 171 | print exc |
---|
| 172 | print "... trying again with single precision" |
---|
[aa4946b] | 173 | model = core.load_model(model_definition, dtype='single', platform="ocl") |
---|
[7cf2cfd] | 174 | calculator = DirectModel(data, model, cutoff=cutoff) |
---|
[0763009] | 175 | value = None # silence the linter |
---|
[216a9e1] | 176 | toc = tic() |
---|
[0763009] | 177 | for _ in range(max(Nevals, 1)): # force at least one eval |
---|
[7cf2cfd] | 178 | value = calculator(**pars) |
---|
[216a9e1] | 179 | average_time = toc()*1000./Nevals |
---|
| 180 | return value, average_time |
---|
| 181 | |
---|
[7cf2cfd] | 182 | |
---|
[0763009] | 183 | def eval_ctypes(model_definition, pars, data, dtype='double', Nevals=1, cutoff=0.): |
---|
[aa4946b] | 184 | model = core.load_model(model_definition, dtype=dtype, platform="dll") |
---|
[7cf2cfd] | 185 | calculator = DirectModel(data, model, cutoff=cutoff) |
---|
[0763009] | 186 | value = None # silence the linter |
---|
[216a9e1] | 187 | toc = tic() |
---|
[0763009] | 188 | for _ in range(max(Nevals, 1)): # force at least one eval |
---|
[7cf2cfd] | 189 | value = calculator(**pars) |
---|
[216a9e1] | 190 | average_time = toc()*1000./Nevals |
---|
| 191 | return value, average_time |
---|
| 192 | |
---|
[7cf2cfd] | 193 | |
---|
[3e6aaad] | 194 | def make_data(qmax, is2D, Nq=128, resolution=0.0, accuracy='Low', view='log'): |
---|
[216a9e1] | 195 | if is2D: |
---|
[346bc88] | 196 | data = empty_data2D(np.linspace(-qmax, qmax, Nq), resolution=resolution) |
---|
[3e6aaad] | 197 | data.accuracy = accuracy |
---|
[87985ca] | 198 | set_beam_stop(data, 0.004) |
---|
| 199 | index = ~data.mask |
---|
[216a9e1] | 200 | else: |
---|
[b89f519] | 201 | if view == 'log': |
---|
| 202 | qmax = math.log10(qmax) |
---|
| 203 | q = np.logspace(qmax-3, qmax, Nq) |
---|
| 204 | else: |
---|
| 205 | q = np.linspace(0.001*qmax, qmax, Nq) |
---|
[346bc88] | 206 | data = empty_data1D(q, resolution=resolution) |
---|
[216a9e1] | 207 | index = slice(None, None) |
---|
| 208 | return data, index |
---|
| 209 | |
---|
[a503bfd] | 210 | def compare(name, pars, Ncpu, Nocl, opts, set_pars): |
---|
[b89f519] | 211 | view = 'linear' if '-linear' in opts else 'log' if '-log' in opts else 'q4' if '-q4' in opts else 'log' |
---|
| 212 | |
---|
[216a9e1] | 213 | opt_values = dict(split |
---|
| 214 | for s in opts for split in ((s.split('='),)) |
---|
| 215 | if len(split) == 2) |
---|
| 216 | # Sort out data |
---|
[29f5536] | 217 | qmax = 10.0 if '-exq' in opts else 1.0 if '-highq' in opts else 0.2 if '-midq' in opts else 0.05 |
---|
[216a9e1] | 218 | Nq = int(opt_values.get('-Nq', '128')) |
---|
[346bc88] | 219 | res = float(opt_values.get('-res', '0')) |
---|
[3e6aaad] | 220 | accuracy = opt_values.get('-accuracy', 'Low') |
---|
[73a3e22] | 221 | is2D = "-2d" in opts |
---|
[3e6aaad] | 222 | data, index = make_data(qmax, is2D, Nq, res, accuracy, view=view) |
---|
[216a9e1] | 223 | |
---|
[87985ca] | 224 | |
---|
| 225 | # modelling accuracy is determined by dtype and cutoff |
---|
| 226 | dtype = 'double' if '-double' in opts else 'single' |
---|
[216a9e1] | 227 | cutoff = float(opt_values.get('-cutoff','1e-5')) |
---|
[87985ca] | 228 | |
---|
| 229 | # randomize parameters |
---|
[7cf2cfd] | 230 | #pars.update(set_pars) # set value before random to control range |
---|
[216a9e1] | 231 | if '-random' in opts or '-random' in opt_values: |
---|
| 232 | seed = int(opt_values['-random']) if '-random' in opt_values else None |
---|
| 233 | pars, seed = randomize_model(name, pars, seed=seed) |
---|
[87985ca] | 234 | print "Randomize using -random=%i"%seed |
---|
[7cf2cfd] | 235 | pars.update(set_pars) # set value after random to control value |
---|
[87985ca] | 236 | |
---|
| 237 | # parameter selection |
---|
| 238 | if '-mono' in opts: |
---|
| 239 | suppress_pd(pars) |
---|
| 240 | if '-pars' in opts: |
---|
| 241 | print "pars",parlist(pars) |
---|
| 242 | |
---|
[aa4946b] | 243 | model_definition = core.load_model_definition(name) |
---|
[87985ca] | 244 | # OpenCl calculation |
---|
[a503bfd] | 245 | if Nocl > 0: |
---|
[aa4946b] | 246 | ocl, ocl_time = eval_opencl(model_definition, pars, data, |
---|
| 247 | dtype=dtype, cutoff=cutoff, Nevals=Nocl) |
---|
[346bc88] | 248 | print "opencl t=%.1f ms, intensity=%.0f"%(ocl_time, sum(ocl)) |
---|
[a503bfd] | 249 | #print max(ocl), min(ocl) |
---|
[87985ca] | 250 | |
---|
| 251 | # ctypes/sasview calculation |
---|
| 252 | if Ncpu > 0 and "-ctypes" in opts: |
---|
[aa4946b] | 253 | cpu, cpu_time = eval_ctypes(model_definition, pars, data, |
---|
| 254 | dtype=dtype, cutoff=cutoff, Nevals=Ncpu) |
---|
[87985ca] | 255 | comp = "ctypes" |
---|
[346bc88] | 256 | print "ctypes t=%.1f ms, intensity=%.0f"%(cpu_time, sum(cpu)) |
---|
[87985ca] | 257 | elif Ncpu > 0: |
---|
[7cf2cfd] | 258 | try: |
---|
| 259 | cpu, cpu_time = eval_sasview(model_definition, pars, data, Ncpu) |
---|
| 260 | comp = "sasview" |
---|
| 261 | #print "ocl/sasview", (ocl-pars['background'])/(cpu-pars['background']) |
---|
| 262 | print "sasview t=%.1f ms, intensity=%.0f"%(cpu_time, sum(cpu)) |
---|
| 263 | except ImportError: |
---|
| 264 | Ncpu = 0 |
---|
[87985ca] | 265 | |
---|
| 266 | # Compare, but only if computing both forms |
---|
[a503bfd] | 267 | if Nocl > 0 and Ncpu > 0: |
---|
| 268 | #print "speedup %.2g"%(cpu_time/ocl_time) |
---|
| 269 | #print "max |ocl/cpu|", max(abs(ocl/cpu)), "%.15g"%max(abs(ocl)), "%.15g"%max(abs(cpu)) |
---|
| 270 | #cpu *= max(ocl/cpu) |
---|
[346bc88] | 271 | resid = (ocl - cpu) |
---|
| 272 | relerr = resid/cpu |
---|
[ba69383] | 273 | #bad = (relerr>1e-4) |
---|
[a503bfd] | 274 | #print relerr[bad],cpu[bad],ocl[bad],data.qx_data[bad],data.qy_data[bad] |
---|
[0763009] | 275 | _print_stats("|ocl-%s|"%comp+(" "*(3+len(comp))), resid) |
---|
| 276 | _print_stats("|(ocl-%s)/%s|"%(comp,comp), relerr) |
---|
[87985ca] | 277 | |
---|
| 278 | # Plot if requested |
---|
| 279 | if '-noplot' in opts: return |
---|
[1726b21] | 280 | import matplotlib.pyplot as plt |
---|
[87985ca] | 281 | if Ncpu > 0: |
---|
[a503bfd] | 282 | if Nocl > 0: plt.subplot(131) |
---|
[7cf2cfd] | 283 | plot_theory(data, cpu, view=view, plot_data=False) |
---|
[87985ca] | 284 | plt.title("%s t=%.1f ms"%(comp,cpu_time)) |
---|
[7cf2cfd] | 285 | #cbar_title = "log I" |
---|
[a503bfd] | 286 | if Nocl > 0: |
---|
[87985ca] | 287 | if Ncpu > 0: plt.subplot(132) |
---|
[7cf2cfd] | 288 | plot_theory(data, ocl, view=view, plot_data=False) |
---|
[a503bfd] | 289 | plt.title("opencl t=%.1f ms"%ocl_time) |
---|
[7cf2cfd] | 290 | #cbar_title = "log I" |
---|
[a503bfd] | 291 | if Ncpu > 0 and Nocl > 0: |
---|
[87985ca] | 292 | plt.subplot(133) |
---|
[29f5536] | 293 | if '-abs' in opts: |
---|
[b89f519] | 294 | err,errstr,errview = resid, "abs err", "linear" |
---|
[29f5536] | 295 | else: |
---|
[b89f519] | 296 | err,errstr,errview = abs(relerr), "rel err", "log" |
---|
[a503bfd] | 297 | #err,errstr = ocl/cpu,"ratio" |
---|
[7cf2cfd] | 298 | plot_theory(data, None, resid=err, view=errview, plot_data=False) |
---|
[346bc88] | 299 | plt.title("max %s = %.3g"%(errstr, max(abs(err)))) |
---|
[7cf2cfd] | 300 | #cbar_title = errstr if errview=="linear" else "log "+errstr |
---|
| 301 | #if is2D: |
---|
| 302 | # h = plt.colorbar() |
---|
| 303 | # h.ax.set_title(cbar_title) |
---|
[ba69383] | 304 | |
---|
[a503bfd] | 305 | if Ncpu > 0 and Nocl > 0 and '-hist' in opts: |
---|
[ba69383] | 306 | plt.figure() |
---|
[346bc88] | 307 | v = relerr |
---|
[ba69383] | 308 | v[v==0] = 0.5*np.min(np.abs(v[v!=0])) |
---|
| 309 | plt.hist(np.log10(np.abs(v)), normed=1, bins=50); |
---|
| 310 | plt.xlabel('log10(err), err = | F(q) single - F(q) double| / | F(q) double |'); |
---|
| 311 | plt.ylabel('P(err)') |
---|
| 312 | plt.title('Comparison of single and double precision models for %s'%name) |
---|
| 313 | |
---|
[8a20be5] | 314 | plt.show() |
---|
| 315 | |
---|
[0763009] | 316 | def _print_stats(label, err): |
---|
| 317 | sorted_err = np.sort(abs(err)) |
---|
| 318 | p50 = int((len(err)-1)*0.50) |
---|
| 319 | p98 = int((len(err)-1)*0.98) |
---|
| 320 | data = [ |
---|
| 321 | "max:%.3e"%sorted_err[-1], |
---|
| 322 | "median:%.3e"%sorted_err[p50], |
---|
| 323 | "98%%:%.3e"%sorted_err[p98], |
---|
| 324 | "rms:%.3e"%np.sqrt(np.mean(err**2)), |
---|
| 325 | "zero-offset:%+.3e"%np.mean(err), |
---|
| 326 | ] |
---|
| 327 | print label," ".join(data) |
---|
| 328 | |
---|
| 329 | |
---|
| 330 | |
---|
[87985ca] | 331 | # =========================================================================== |
---|
| 332 | # |
---|
| 333 | USAGE=""" |
---|
| 334 | usage: compare.py model [Nopencl] [Nsasview] [options...] [key=val] |
---|
| 335 | |
---|
| 336 | Compare the speed and value for a model between the SasView original and the |
---|
| 337 | OpenCL rewrite. |
---|
| 338 | |
---|
| 339 | model is the name of the model to compare (see below). |
---|
| 340 | Nopencl is the number of times to run the OpenCL model (default=5) |
---|
| 341 | Nsasview is the number of times to run the Sasview model (default=1) |
---|
| 342 | |
---|
| 343 | Options (* for default): |
---|
| 344 | |
---|
| 345 | -plot*/-noplot plots or suppress the plot of the model |
---|
[2d0aced] | 346 | -single*/-double uses double precision for comparison |
---|
[29f5536] | 347 | -lowq*/-midq/-highq/-exq use q values up to 0.05, 0.2, 1.0, 10.0 |
---|
[216a9e1] | 348 | -Nq=128 sets the number of Q points in the data set |
---|
[73a3e22] | 349 | -1d*/-2d computes 1d or 2d data |
---|
[2d0aced] | 350 | -preset*/-random[=seed] preset or random parameters |
---|
| 351 | -mono/-poly* force monodisperse/polydisperse |
---|
| 352 | -ctypes/-sasview* whether cpu is tested using sasview or ctypes |
---|
[3e6aaad] | 353 | -cutoff=1e-5* cutoff value for including a point in polydispersity |
---|
[2d0aced] | 354 | -pars/-nopars* prints the parameter set or not |
---|
| 355 | -abs/-rel* plot relative or absolute error |
---|
[b89f519] | 356 | -linear/-log/-q4 intensity scaling |
---|
[ba69383] | 357 | -hist/-nohist* plot histogram of relative error |
---|
[346bc88] | 358 | -res=0 sets the resolution width dQ/Q if calculating with resolution |
---|
[3e6aaad] | 359 | -accuracy=Low resolution accuracy Low, Mid, High, Xhigh |
---|
[87985ca] | 360 | |
---|
| 361 | Key=value pairs allow you to set specific values to any of the model |
---|
| 362 | parameters. |
---|
| 363 | |
---|
| 364 | Available models: |
---|
| 365 | """ |
---|
| 366 | |
---|
[7cf2cfd] | 367 | |
---|
[216a9e1] | 368 | NAME_OPTIONS = set([ |
---|
[87985ca] | 369 | 'plot','noplot', |
---|
| 370 | 'single','double', |
---|
[29f5536] | 371 | 'lowq','midq','highq','exq', |
---|
[87985ca] | 372 | '2d','1d', |
---|
| 373 | 'preset','random', |
---|
| 374 | 'poly','mono', |
---|
| 375 | 'sasview','ctypes', |
---|
| 376 | 'nopars','pars', |
---|
| 377 | 'rel','abs', |
---|
[b89f519] | 378 | 'linear', 'log', 'q4', |
---|
[ba69383] | 379 | 'hist','nohist', |
---|
[216a9e1] | 380 | ]) |
---|
| 381 | VALUE_OPTIONS = [ |
---|
| 382 | # Note: random is both a name option and a value option |
---|
[3e6aaad] | 383 | 'cutoff', 'random', 'Nq', 'res', 'accuracy', |
---|
[87985ca] | 384 | ] |
---|
| 385 | |
---|
[7cf2cfd] | 386 | def columnize(L, indent="", width=79): |
---|
| 387 | column_width = max(len(w) for w in L) + 1 |
---|
| 388 | num_columns = (width - len(indent)) // column_width |
---|
| 389 | num_rows = len(L) // num_columns |
---|
| 390 | L = L + [""] * (num_rows*num_columns - len(L)) |
---|
| 391 | columns = [L[k*num_rows:(k+1)*num_rows] for k in range(num_columns)] |
---|
| 392 | lines = [" ".join("%-*s"%(column_width, entry) for entry in row) |
---|
| 393 | for row in zip(*columns)] |
---|
| 394 | output = indent + ("\n"+indent).join(lines) |
---|
| 395 | return output |
---|
| 396 | |
---|
| 397 | |
---|
[373d1b6] | 398 | def get_demo_pars(name): |
---|
| 399 | __import__('sasmodels.models.'+name) |
---|
[e922c5d] | 400 | model = getattr(models, name) |
---|
[373d1b6] | 401 | pars = getattr(model, 'demo', None) |
---|
| 402 | if pars is None: pars = dict((p[0],p[2]) for p in model.parameters) |
---|
| 403 | return pars |
---|
| 404 | |
---|
[87985ca] | 405 | def main(): |
---|
| 406 | opts = [arg for arg in sys.argv[1:] if arg.startswith('-')] |
---|
| 407 | args = [arg for arg in sys.argv[1:] if not arg.startswith('-')] |
---|
[d547f16] | 408 | models = "\n ".join("%-15s"%v for v in MODELS) |
---|
[87985ca] | 409 | if len(args) == 0: |
---|
[7cf2cfd] | 410 | print(USAGE) |
---|
| 411 | print(columnize(MODELS, indent=" ")) |
---|
[87985ca] | 412 | sys.exit(1) |
---|
| 413 | if args[0] not in MODELS: |
---|
| 414 | print "Model %r not available. Use one of:\n %s"%(args[0],models) |
---|
| 415 | sys.exit(1) |
---|
| 416 | |
---|
| 417 | invalid = [o[1:] for o in opts |
---|
[216a9e1] | 418 | if o[1:] not in NAME_OPTIONS |
---|
| 419 | and not any(o.startswith('-%s='%t) for t in VALUE_OPTIONS)] |
---|
[87985ca] | 420 | if invalid: |
---|
| 421 | print "Invalid options: %s"%(", ".join(invalid)) |
---|
| 422 | sys.exit(1) |
---|
| 423 | |
---|
[d547f16] | 424 | # Get demo parameters from model definition, or use default parameters |
---|
| 425 | # if model does not define demo parameters |
---|
| 426 | name = args[0] |
---|
[373d1b6] | 427 | pars = get_demo_pars(name) |
---|
[d547f16] | 428 | |
---|
[87985ca] | 429 | Nopencl = int(args[1]) if len(args) > 1 else 5 |
---|
[ba69383] | 430 | Nsasview = int(args[2]) if len(args) > 2 else 1 |
---|
[87985ca] | 431 | |
---|
| 432 | # Fill in default polydispersity parameters |
---|
| 433 | pds = set(p.split('_pd')[0] for p in pars if p.endswith('_pd')) |
---|
| 434 | for p in pds: |
---|
| 435 | if p+"_pd_nsigma" not in pars: pars[p+"_pd_nsigma"] = 3 |
---|
| 436 | if p+"_pd_type" not in pars: pars[p+"_pd_type"] = "gaussian" |
---|
| 437 | |
---|
| 438 | # Fill in parameters given on the command line |
---|
| 439 | set_pars = {} |
---|
| 440 | for arg in args[3:]: |
---|
| 441 | k,v = arg.split('=') |
---|
| 442 | if k not in pars: |
---|
| 443 | # extract base name without distribution |
---|
| 444 | s = set(p.split('_pd')[0] for p in pars) |
---|
| 445 | print "%r invalid; parameters are: %s"%(k,", ".join(sorted(s))) |
---|
| 446 | sys.exit(1) |
---|
| 447 | set_pars[k] = float(v) if not v.endswith('type') else v |
---|
| 448 | |
---|
| 449 | compare(name, pars, Nsasview, Nopencl, opts, set_pars) |
---|
| 450 | |
---|
[8a20be5] | 451 | if __name__ == "__main__": |
---|
[87985ca] | 452 | main() |
---|