msyriac / pyfisher Goto Github PK

We need to check if num_massive_neutrinos of 1 (CAMB default) vs. 3 (Planck analysis) makes any significant difference to derivatives and/or Fisher forecasts. Easiest thing to do first would be to make new derivs with num_massive_neutrinos = 3 and see if you get the same constraints for a set of experiments that includes lensing Auto information from high-ell.

I'm allowing for two ways to calculate the Fisher matrix, which under its current implementation should be identical for the case where no lensing data is added. The default is the compact = False version which uses FisherForecast.CovFromVecs. This is just an implementation of Eq.A.4 of 1509.07471 (or more explicitly, Eq 3 of 1109.6322). The compact = True option uses FisherForecast.CovFromVecsSmall, which is easier to extend to datasets involving cross-correlations with optical data (Eq.4 of 1402.4108). However, the numbers don't perfectly agree, so there's likely a bug somewhere.

The bot created this issue to inform you that pyup.io has been set up on this repo.
Once you have closed it, the bot will open pull requests for updates as soon as they are available.

Check that accuracy boost makes things better

Need to write an __init__.py

Add cosmic shear and CMB x shear spectra for forecasting.

We noticed that there were discrepancies between polarization estimator noise curves from pyfisher/lensInterface.lensNoise() and orphics.lensing.lensing_noise() functions.

Here the dashed lines are from lensInterface.lensNoise and the solid lines are from lensing.lensing_noise(). I am opening this issue to track the changes needed to be made to lensInterface.py

1. L min for Kappa is hard-coded (https://github.com/msyriac/pyfisher/blob/master/pyfisher/lensInterface.py#L51)

2. L max for Kappa set in param.ini is not being used (https://github.com/msyriac/pyfisher/blob/master/pyfisher/lensInterface.py#L55)

3. Wrong function signature for getNlIterative (https://github.com/msyriac/pyfisher/blob/master/pyfisher/lensInterface.py#L68)

• call from lensInterface.py
  myNls.getNlIterative(pols,kellmin,kellmax,tellmax,pellmin,pellmax,dell=dell,halo=True,plot=plot)

• function definition in orphics/lensing.py
  def getNlIterative(self,polCombs,pellmin,pellmax,dell=20,halo=True,dTolPercentage=1.,verbose=True,plot=False,max_iterations=np.inf,eff_at=60,kappa_min=0,kappa_max=np.inf):

4. Polarization l min and l max are not used in getNlIterative (It is probably better to open an issue about this on orphics repo).
   (https://github.com/msyriac/orphics/blob/master/orphics/lensing.py#L1218)

We should add derivatives w.r.t. tensor-scalar-ratio r and the equation of state of dark energy w, and also its dynamical component w_a. It would be nice to have a plot of C_ell around each parameter for a few representative ells, with the final chosen step size clearly shown (showing both linearity and numerical stability).

Add BB for r forecasts.