gammapy /
gammapy
| @@ 377-407 (lines=31) @@ | ||
| 374 | ||
| 375 | return {"counts": np.array(counts, dtype=int).sum()} |
|
| 376 | ||
| 377 | def estimate_norm_ul(self, datasets): |
|
| 378 | """Estimate upper limit for a flux point. |
|
| 379 | ||
| 380 | Parameters |
|
| 381 | ---------- |
|
| 382 | datasets : `~gammapy.modeling.Datasets` |
|
| 383 | the list of dataset object |
|
| 384 | ||
| 385 | Returns |
|
| 386 | ------- |
|
| 387 | result : dict |
|
| 388 | Dict with upper limit for the flux point norm. |
|
| 389 | """ |
|
| 390 | norm = self.model.norm |
|
| 391 | ||
| 392 | # TODO: the minuit backend has convergence problems when the fit statistic is not |
|
| 393 | # of parabolic shape, which is the case, when there are zero counts in the |
|
| 394 | # energy bin. For this case we change to the scipy backend. |
|
| 395 | counts = self.estimate_counts(datasets)["counts"] |
|
| 396 | ||
| 397 | if np.all(counts == 0): |
|
| 398 | result = self.fit.confidence( |
|
| 399 | parameter=norm, |
|
| 400 | sigma=self.sigma_ul, |
|
| 401 | backend="scipy", |
|
| 402 | reoptimize=self.reoptimize, |
|
| 403 | ) |
|
| 404 | else: |
|
| 405 | result = self.fit.confidence(parameter=norm, sigma=self.sigma_ul) |
|
| 406 | ||
| 407 | return {"norm_ul": result["errp"] + norm.value} |
|
| 408 | ||
| 409 | def estimate_norm_ts(self, datasets): |
|
| 410 | """Estimate ts and sqrt(ts) for the flux point. |
|
| @@ 1047-1075 (lines=29) @@ | ||
| 1044 | ||
| 1045 | return {"counts": np.array(counts, dtype=int)} |
|
| 1046 | ||
| 1047 | def estimate_norm_ul(self): |
|
| 1048 | """Estimate upper limit for a flux point. |
|
| 1049 | ||
| 1050 | Returns |
|
| 1051 | ------- |
|
| 1052 | result : dict |
|
| 1053 | Dict with upper limit for the flux point norm. |
|
| 1054 | """ |
|
| 1055 | if not self._contribute_to_stat: |
|
| 1056 | return {"norm_ul": np.nan} |
|
| 1057 | ||
| 1058 | norm = self.model.norm |
|
| 1059 | ||
| 1060 | # TODO: the minuit backend has convergence problems when the fit statistic is not |
|
| 1061 | # of parabolic shape, which is the case, when there are zero counts in the |
|
| 1062 | # energy bin. For this case we change to the scipy backend. |
|
| 1063 | counts = self.estimate_counts()["counts"] |
|
| 1064 | ||
| 1065 | if np.all(counts == 0): |
|
| 1066 | result = self.fit.confidence( |
|
| 1067 | parameter=norm, |
|
| 1068 | sigma=self.sigma_ul, |
|
| 1069 | backend="scipy", |
|
| 1070 | reoptimize=self.reoptimize, |
|
| 1071 | ) |
|
| 1072 | else: |
|
| 1073 | result = self.fit.confidence(parameter=norm, sigma=self.sigma_ul) |
|
| 1074 | ||
| 1075 | return {"norm_ul": result["errp"] + norm.value} |
|
| 1076 | ||
| 1077 | def estimate_norm_ts(self): |
|
| 1078 | """Estimate ts and sqrt(ts) for the flux point. |
|
