Horizon annealing: a collection-based approach to automated sequencing of the fossil record

Abstract

2012:Horizon annealing: a collection-based approach to automated sequencing of the fossil record. Lethaia, Vol. 45, A number of different approaches to quantitative biochronology have been proposed and used to construct high-resolution time-scales for a range of uses. We present a new approach, horizon annealing, which uses simulated annealing to optimize the sequencing of collection horizons. Temporal sequences of events produced by this method are compared with those produced by graphic correlation, CONOP and RASC for a series of previously studied exemplar data sets. Horizon annealing produces results similar to other methods, but it does have properties (the ordination of collections and the avoidance of some local minima) that make it useful for high-resolution studies, particularly those based on capture-mark-recapture methods requiring detailed presence-absence data for individual collections and taxa. h Chronostratigraphy, graphic correlation, graptolite, rate of evolution, CONOP9.H. David Sheets [sheets@canisius.edu], Zachary T. Izard [izard_z@yahoo.com] and John

A perspective on reconstructing the spatial and temporal patterns of Earth′s biodiversity in deep time

“…基于这一思路设计的单元组合法(Unitary Association method, UAM) [61] 虽然可以解决数据较少情况 下不同剖面化石对比中的矛盾问题, 但当数据集规模 较大时则难以应用. 另一条思路则是通过一些随机寻 优算法, 如模拟退火算法、遗传算法、蚁群算法等来 获得事件排序的相对最优解, 如约束最优化法(Constrained Optimization method, CONOP) [57,62] 和层位退火 法(Horizon annealing, HA) [63] , 并通过并行计算以及大 量处理器核心(如超级计算机)来尽可能缩短计算时 间 [21,39] . 这种方法摒弃基于带化石或标准化石来进行 传统生物地层对比的简单做法, 希望通过尝试使用所 有化石记录, 对每个物种的首末现进行计算排序, 并且 使最终得到的化石首末现事件的排序与实际剖面中观 察到的化石首末现的先后顺序尽可能一致, 以实现最 优的剖面对比及地质事件排序 [59,60] .…”

Stratigraphic ranking of selected invertebrate fossils: A quantitative approach at different temporal and geographic scales

“…Yet, analyzing large geographic or stratigraphic datasets manually is not possible; fortunately, numerous quantitative stratigraphic approaches have been developed in the past decades such as 'Unitary Association' (UA; Guex, 1979), 'Ranking & Scaling' (Agterberg and Gradstein, 1999), 'Constrained Optimization' (Kemple et al, 1989), and 'Horizon Annealing' (Sheets et al, 2012). For a detailed comparison among these biostratigraphic methods we here refer to Gradstein (2012).…”