万里挑一：ATLAS的触发系统 One in a million: The ATLAS ‘Trigger’ System
LHC观测到的质子-质子相互作用频率高达每秒10亿次，数据足以超载10万张光盘。但是这些碰撞中，只有一小部分包含值得研究的信息，因此ATLAS使用了分层触发系统来筛选有研究价值的数据。
The LHC sees a proton-proton interaction rate of magnitude of a billion events per second, providing enough data for 100,000 compact discs. Only a small fraction of those interactions contains interesting properties, thus a tiered trigger system is implemented to pick out interesting events.
第一层在几毫秒内会将碰撞数据量降低一万倍，达到100KHz。突颖而出的标准基于基本的电子技术，并会使用离散逻辑模块和定制集成电路进行校准。
The first tier brings the interaction rate down by ten thousand times to 100KHz within a few milliseconds. The criteria for an event to be flagged is based off of basic electronics and is calibrated using discrete logic modules and custom integrated circuits.
然后进行第二层筛选，将数据规模进一步减少一百倍，因为每次碰撞中都包含了数十个单独的交互。如果它能触发第二层筛选，那该数据将继续进行脱机分析。
A second trigger is then implemented to reduce the rate by a further 1 hundred times, since each event contains tens of individual interactions. If the second trigger is set off, then that data proceeds for offline analysis.
这样，在需要移动和分析数据的时候，它们的量才变得可控。而且因为技术的进步，第二层触发器的建造比原先设想的要容易，因为碰撞中的每个区域都会通过同一个处理器运行，而不是像刚开始的时候想的时候那样，每个区域都使用自己的单个处理器。
From this at any given time the data required to be moved and analyzed is at a manageable amount. Due to technological advances the second trigger became easier to implement than originally thought, since all regions of interest in an event could run through a single processor rather than having individual processors for each region of interest.
经过第二层触发器过滤器之后，每秒钟只有400个碰撞事件会进入初始存档并列为值得研究的事件。比如当时希格斯粒子衰变成了两个tau轻子。*
After the second trigger events are escalated to the event filter, where only 400 events per second will proceed into raw storage and categorized as potentially interesting. An example would be Higgs decaying into two tau leptons. *
但是还没结束，触发器分析团队会进一步使用cuts指定通过碰撞事件筛选器的数据。Cuts指的是背景噪音信号的百分比，从信号中去除任何不需要的噪音，才能创建更清晰的数据图。
However, that’s not the end yet, the trigger analysis team further specify data that makes it through the event filter using cuts. Cuts are a signal to background percentages to remove any unwanted noise from a signal to create a clearer data plot.
图为ATLAS中每个碰撞中每一喷流中特定衰减模式的数目
The number of particular decay modes in each event per number of jets
最后，从ATLAS探测器收集的数据会被分为开放数据和内部数据。开放数据会对大众开放，其中包括不同的最新希格斯粒子衰变成WW, Z玻色子和夸克的模型。
Data collected from ATLAS detector is split into open and closed data. Open data, which is open to the public, includes different decay modes of the Higgs particle into WW, Z bosons and top quarks.
尽管通过触发级别减少了数据量，但ATLAS检测器每天仍然会收集1千万亿字节的数据。需要长期存储的数据会保存在磁带上，而中期数据则不断转移到更新的技术上，如高密度磁带。
Although data quantity is reduced through trigger levels the ATLAS detector still gathers 1 petabyte of data every day. Long term storage is kept on magnetic tapes whereas medium term data is continually moved onto newer technology such as high-density tapes.
为了每天庞大的数据数量，CERN开发了自己的软件存储方案EOS，以满足ATLAS实验的数据存储需求。当然，EOS现在也被应用到了粒子物理学以外的其他领域，比如澳大利亚学术和研究网络(Australian Academic and Research Network)，或其他大数据系统。
CERN developed their own software storage solution known as EOS, in order to keep up with the data storage demand of the ATLAS experiment. However, EOS is now being used beyond particle physics, as other big data systems such as the Australian Academic and Research Network require CERN’s technology.
作为万维网的鼻祖，
CERN的计算机技术始终保持最前端,
期待未来大数据时代对EOS的应用，
以及实验过程中的新开发！
As founders of the World Wide Web,
CERN's has always been at
the forefront of computing technology.
Hope there will be more applications of EOS
in the era of big data and
new developments in future experiments!
图为CERN 的计算机中心，有超大的数据储存能量
Inside CERN's computer centre with very high capacity of data storage
图片和视频来自 ATLAS, CERN, IOP, Research Gate官网
文中除*部分英文信息应用于ATLAS, CERN, IOP官网，
其余中英文内容为原创
Pictures and video from official website of ATLAS, CERN, IOP, Research Gate.
The * parts of English information was cited from official websites of ATLAS, CERN, IOP,
the rest of the Chinese and English content is original
