Lensing the C_l using the deflection angle from the computed lensing potential power spectrum. lensing_method=1: using an accurate curved-sky correlation function method lensing_method=2: using the flat-sky lower order result of astro-ph/9505109 and astro-ph/9803150 as in CMBFAST lensing_method=3: using inaccurate full sky harmonic method of astro-ph/0001303 The flat sky result is accurate to about 0.1% in TT, and 0.4% in EE and is about a factor of two faster than lensing_method=1. lensing_method=3 is only present for comparison and is not recommended in any regime Set accurate_BB=T if you want BB accurately by integrating the full angular range otherwise it saves a large amount of time by only integrating the small scales accute_BB only does *not* include any non-linear corrections or ensure you have chosen sufficiently high l_max and k_max, so does not neccessarily give an accurate result Uses the un-lensed Cls and the computed lensing potential power spectrum. Usual values of k_max are fine for all but the lensed BB Cls To get the lensed BB accurate around l=1000 you need to go to l_max >2000, and higher for higher l. Since this probes small scales in the lensing power spectrum you also need to go to higher k_max - for concordance models something like k_eta_max_scalar=10000. At l>1000 you can expect to need higher k_max, and for non-linear evolution to cause a significant error. Correlation function routines by AL+AC Nov 2004 with flat-sky borrowings from CMBFAST Curved sky results use the method of astro-ph/xxx. Full sky harmonic lensing routines by Gayoung Chon and AC. Ref: astro-ph/0001303 by W. Hu. For better derivations see also astro-ph/0301064 and astro-ph/0301031 Adapted for CAMB and optimized by AL. Uses f90 version of "J1-RECURSION OF 3J-COEFFICIENTS" by K. Schulten and R.G. Gordon obtainable from the CPC program library (www.cpc.cs.qub.ac.uk). March 2006: fixed problem with l_max when generating with tensors (thanks Chad Fendt)