Tuning of Monte Carlo Event Generators

大型强子对撞机（LHC）的物理学可能会涉及一系列令人眼花缭乱的新现象，这是由于人造粒子对撞机从未产生过的新重粒子。目前，我们还不知道这些粒子的特征是什么，但有许多理论模型可以预测它们的特定排列。其中包括一些令人兴奋的想法，如“超对称”，它至少需要已知基本粒子数量的两倍；额外维度的存在，这将是一个深刻的发现；或者是一系列其他的变化。其中任何一种如果被发现，都将是我们对宇宙理解的巨大飞跃。大型强子对撞机的大部分物理工作都将致力于找出这些模型中，如果有的话，是哪一个在自然界中实现的。大型强子对撞机物理学将会困难的部分原因是，粒子物理学在过去50年里以如此高的精度研究的现象——非常成功的“标准模型”（SM）——将成为测量新物理学的一个巨大的“背景噪音”来源。为了研究新的重粒子，我们必须首先对可能隐藏它们的SM物理有很好的控制。如果我们的SM模型不准确，就不可能提取出新粒子的清晰信号。或者，也许更糟的是，我们可能不小心声称发现了实际上并不存在的东西！理解SM如何在对撞机实验中表现出来并不容易：特别是，在进行计算时，强核力是有问题的。这种力将基本的夸克粒子“粘在一起”，而夸克粒子从来没有被孤立地看到过，从而形成了我们观察到的强子粒子，比如质子和中子。使用QED（电磁学的SM理论）进行的计算可以迅速收敛到准确的预测，但在QCD（强作用力理论）中进行的相应计算非常复杂，收敛速度要慢得多。QCD还有一些方面是不能用常规方法处理的：例如，夸克强结合成强子的过程还没有被很好地理解。为了解决这些计算问题，粒子物理学家使用计算机模拟，而不是从最基本的QCD思想开始计算所有东西，而是采用更近似的路径。这些被称为“事件生成器”的模拟模拟了粒子碰撞的各个方面，从最具能量的相互作用，到来自高能夸克的粒子喷射，再到夸克最终向强子的低能量转换。人们必须为这种计算捷径付出的代价是预测性：真正的QCD计算（尽管远远超出任何已知方法）只需要指定几个数字，而事件生成器则需要大量输入参数。对于给定的值是否“好”的唯一真正测试是事件生成器的预测与实际测量相比有多好。这对于大型强子对撞机物理学来说是非常重要的：除非以可靠的事件生成器调谐的形式理解新物理学的QCD背景，否则所有令人兴奋的新物理学都可能被隐藏起来。然而，有了这么多参数，找出哪些组合是好的本身就是一项艰巨的任务。该项目的目的是根据所有可用数据系统地调整LHC中使用的所有主要事件发生器，以便对新物理学的QCD背景进行良好的建模。重要的是，调整生成器所依据的数据包括来自大型强子对撞机本身的数据，因为在那里将遇到以前从未探测过的QCD区域。通过对QCD模型的良好控制，大型强子对撞机所要寻找的新物理可以被可靠地识别和研究，也许会永远改变我们对宇宙的看法。

Physics at the Large Hadron Collider (LHC) is likely to involve a bewildering array of new phenomena, due to new heavy particles which have never been produced before with a man-made particle collider. At present, we have no idea what the characteristics of these particles will be, but there are many theoretical models which predict particular arrangements. These include such exciting ideas as 'supersymmetry', which at least requires a doubling of the number of known fundamental particles; the existence of extra dimensions, which would be a deeply profound discovery; or a menagerie of other variations. Any of these, if discovered, would be a giant leap forward in our understanding of the universe. Much of LHC physics will be geared towards working out which, if any, of these models is realised in nature. Part of the reason that LHC physics will be difficult is that the phenomena that particle physics has spent the past 50 years investigating to such high accuracy - the extraordinarily successful 'Standard Model' (SM) - will be a huge source of 'background noise' in measurements of new physics. To study the new, heavy particles, we must first have excellent control of the SM physics which threatens to hide them. If our models of the SM are not accurate, it may be impossible to extract clean signals of the new particles. Or, perhaps worse, we could accidentally claim the discovery of something which is not actually there! Understanding how the SM manifests itself in collider experiments is not easy: in particular, the strong nuclear force is problematic when it comes to making calculations. This force 'glues together' fundamental 'quark' particles, which are never seen in isolation, to make the observed 'hadron' particles like protons and neutrons. Calculations made using QED, the SM theory of electromagnetism, rapidly converge to accurate predictions, but the corresponding calculations in QCD, the strong force theory, are extremely complex and the convergence rate is much slower. There are also aspects of QCD which cannot be handled by normal methods: for example, the process by which quarks bind strongly into hadrons is not well understood. To get around these calculational problems, particle physicists use computer simulations which, rather than calculating everything from the most fundamental QCD ideas upwards, take a more approximate path. These simulations, called 'event generators', simulate all aspects of particle collisions, from the most energetic interactions, to the spray of particles that come from a high energy quark, and to the eventual low-energy conversions of quarks into hadrons. The price that one must pay for such a calculational shortcut is predictivity: where a genuine QCD calculation, albeit one massively beyond any known methods, would only require a few numbers to be specified, event generators have a lot of input parameters. The only real test of whether or not the values given are 'good' is how well the event generator predictions compare to real measurements. This is of enormous importance to LHC physics: unless the QCD backgrounds to new physics are understood, in the form of reliable event generator tunings, all the new and exciting physics may remain hidden. With so many parameters, however, working out which combinations are good is a substantial task in itself. The aim of this project is to systematically tune all the major event generators to be used at the LHC against all the available data, so that QCD backgrounds to new physics are well-modelled. It is important that the data against which the generators are tuned includes that from the LHC itself, as areas of QCD which have never been probed before will be encountered there. With a good control of QCD models, the new physics which the LHC was built to find can be reliably identified and studied, perhaps changing our view of the universe forever.