From b6666c5103f9301e823aad485087b9c39442bfb0 Mon Sep 17 00:00:00 2001 From: Surhud More Date: Fri, 6 Apr 2018 18:16:42 +0900 Subject: [PATCH] DOC: ceff mode is faster but less accurate The constant efficiency mode is faster but less accurate. See e.g, the second last sentence in arXiv:1306.2144 . There is no free lunch. :) --- sphinx-documentation/nested.rst | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sphinx-documentation/nested.rst b/sphinx-documentation/nested.rst index 110f03fe..6ffdc9f2 100644 --- a/sphinx-documentation/nested.rst +++ b/sphinx-documentation/nested.rst @@ -156,7 +156,7 @@ General sampling options - :code:`nlive | n_live_points (400)` : number of points used in every iteration. - :code:`IS | importance_nested_sampling (True)` : whether to use Importance Nested Samplin (see `arXiv:1306.2144 `__). - :code:`efr | sampling_efficiency (0.8)` : defines the sampling efficiency (see 'Use cases' below). -- :code:`ceff | const_efficiency_mode (True)` : constant efficiency mode -- slower, but more accurate evidence estimation. +- :code:`ceff | const_efficiency_mode (True)` : constant efficiency mode -- faster, but less accurate evidence estimation. - :code:`seed | seed (-1)`: seed of the random number generator (if negative, uses system clock). - :code:`logZero | log_zero (-1e90)` : if the log-likelihood of a sample is smaller than this value, the sample is ignored. - :code:`updInt | n_iter_before_update (100)` : number of iteration after which the output files are updated.