From 78d0d576b47c392858b32c4dad46d9bf8f1fdf30 Mon Sep 17 00:00:00 2001 From: pelssers Date: Thu, 26 Mar 2020 11:34:12 +0100 Subject: [PATCH] Update README and HISTORY --- HISTORY.md | 6 ++++++ README.md | 4 ---- 2 files changed, 6 insertions(+), 4 deletions(-) diff --git a/HISTORY.md b/HISTORY.md index 7d037b4c0..ba51105d4 100644 --- a/HISTORY.md +++ b/HISTORY.md @@ -1,3 +1,9 @@ +1.0.0 / 2020-03-26 +------------------ +- Fiducial volume specification (#64) +- added default cS1 cut (#63) +- Cleanup and optimizations (#63, #64, #65) + 0.5.0 / 2020-01-31 ------------------ - Autographed Hessian; use Hessian in the optimizer (#62) diff --git a/README.md b/README.md index b3cb28c25..128cc187c 100644 --- a/README.md +++ b/README.md @@ -27,7 +27,3 @@ This has several advantages: - Each event has its "private" detector model computation at the observed (x, y, z, time), so it is easy and cheap to add time- and position dependences to the likelihood. - Since the likelihood for a dataset takes O(seconds) to compute, we can do this at each of optimizer's proposed points during inference. We thus remove a histogram precomputation step exponential in the number of parameters, and can thus fit a great deal more parameters. - By implementing the signal model in tensorflow, the likelihood becomes differentiable. Using the gradient during fitting drastically reducing the number of needed interactions for a fit or profile likelihood. - -Note this is still under construction / development, so it probably has some bugs and little documentation. - -