Issues/43

README.md

@@ -36,12 +36,6 @@ git clone [email protected]:LSSTDESC/firecrown.git
 cd firecrown
 ```
 
-The current version of `augur` relies on firecrown v0.5 or lower so in the same directory you can do:
-
-```
-git checkout tags/v0.5
-pip install . -e
-```
 
 Now run a `pytest` to see if things work.
 
@@ -100,11 +94,20 @@ The user can create configuration files to fit their specific purposes following
 
 ```
 from augur.generate import generate
-lk = generate('./examples/config_test.yml', return_all_outputs=False, force_read=False)
+lk = generate('./examples/config_test.yml', return_all_outputs=False)
 ```
 
 This likelihood object can then be used by `cosmosis`, `cobaya` or `NumCosmo`. For more details follow the examples in the [`firecrown`](https://github.com/LSSTDESC/firecrown) repository.
 
+Additionally, we can compute the Fisher matrix and Fisher biases via numerical derivatives using the following commands:
+
+```
+from augur.analyze import Analyze
+ao = Analyze('./examples/config_test.yml')
+ao.get_fisher_bias()  # This command computes the derivates+Fisher matrices+fisher bias
+print(ao.Fij, ao.bi)  # These are the values of the Fisher matrix, Fij, and Fisher biases bi
+```
+
 ## Example run for SRD v1
 We also include example configuration files for `cosmosis` and `cobaya` to reproduce the results from the [LSST DESC Science Requirements Document](https://arxiv.org/pdf/1809.01669.pdf).
 
@@ -142,4 +145,4 @@ cd examples
 cobaya-run cobaya_mcmc.yaml
 ```
 
-By default the outputs will be saved at `./examples/cobaya_evaluate_output`.
+By default the outputs will be saved at `./examples/cobaya_evaluate_output`.

augur/_version.py

@@ -1 +1 @@
-__version__ = "0.3.0"
+__version__ = "0.4.0"

augur/generate.py

@@ -12,10 +12,18 @@
 from augur.tracers.two_point import ZDist, LensSRD2018, SourceSRD2018
 from augur.utils.cov_utils import get_gaus_cov, get_SRD_cov, get_noise_power
 from augur.utils.cov_utils import TJPCovGaus
-import firecrown.likelihood.gauss_family.statistic.source.weak_lensing as wl
-import firecrown.likelihood.gauss_family.statistic.source.number_counts as nc
-from firecrown.likelihood.gauss_family.statistic.two_point import TwoPoint
-from firecrown.likelihood.gauss_family.gaussian import ConstGaussian
+from packaging.version import Version
+import firecrown
+if Version(firecrown.__version__) >= Version('1.8'):
+    import firecrown.likelihood.weak_lensing as wl
+    import firecrown.likelihood.number_counts as nc
+    from firecrown.likelihood.two_point import TwoPoint
+    from firecrown.likelihood.gaussian import ConstGaussian
+elif Version(firecrown.__version__) >= Version('1.7.4'):
+    import firecrown.likelihood.gauss_family.statistic.source.weak_lensing as wl
+    import firecrown.likelihood.gauss_family.statistic.source.number_counts as nc
+    from firecrown.likelihood.gauss_family.statistic.two_point import TwoPoint
+    from firecrown.likelihood.gauss_family.gaussian import ConstGaussian
 from firecrown.modeling_tools import ModelingTools
 from firecrown.parameters import ParamsMap
 from augur.utils.config_io import parse_config
@@ -305,6 +313,14 @@ def generate(config, return_all_outputs=False, write_sacc=True):
     tools.prepare(cosmo)
     # Run the likelihood (to get the theory)
     lk.compute_loglike(tools)
+    # Get all bandpower windows before erasing the placeholder sacc
+    win_dict = {}
+    for st in lk.statistics:
+        st = st.statistic
+        tr1 = st.source0.sacc_tracer
+        tr2 = st.source1.sacc_tracer
+        idx = S.indices(tracers=(tr1, tr2))
+        win_dict[(tr1, tr2)] = S.get_bandpower_windows(idx)
     # Empty the placeholder Sacc's covariance and data vector so we can "overwrite"
     S.covariance = None
     S.data = []
@@ -313,10 +329,12 @@ def generate(config, return_all_outputs=False, write_sacc=True):
     for st in lk.statistics:
         # Hack to be able to reuse the statistics
         st = st.statistic
+        tr1 = st.source0.sacc_tracer
+        tr2 = st.source1.sacc_tracer
         st.ready = False
-        S.add_ell_cl(st.sacc_data_type, st.sacc_tracers[0], st.sacc_tracers[1],
-                     st.ell_or_theta_, st.get_theory_vector(),
-                     window=st.theory_window_function)
+        S.add_ell_cl(st.sacc_data_type, tr1, tr2,
+                     st.ells, st.get_theory_vector(),  # Only valid for harmonic space
+                     window=win_dict[(tr1, tr2)])
     if config['cov_options']['cov_type'] == 'gaus_internal':
         fsky = config['cov_options']['fsky']
         cov = get_gaus_cov(S, lk, cosmo, fsky, config)

environment.yml

@@ -2,7 +2,7 @@ name: forecasting
 channels:
   - conda-forge
 dependencies:
-  - firecrown>=1.7.1
+  - firecrown>=1.7.4
   - flake8
   - healpy
   - jinja2