Observational cosmology with multi-wavelength surveys

宇宙学研究宇宙的大尺度性质和演化。因此，宇宙学可以说是物理学中最完整的分支之一，因为它必须能够描述物质的大规模分布和运动，主要由引力控制，但也包括大爆炸期间和之后亚原子粒子之间的复杂相互作用，以及发生在星系和星系团中的暴力物理过程。然而，宇宙学与其他物理学分支的最大区别在于它不可能重复实验。我们只有一组数据：宇宙，我们不能重复它。这最初使宇宙学处于尴尬的境地，由于缺乏实验数据，进展主要是由基于基本前提的理论工作推动的。令人惊讶的是，随着天文观测的改进，许多理论预言被证明是正确的。在过去的几十年里，宇宙学已经发展成为一门由实验观测驱动的充实的科学。由于只有一个宇宙可以观测，宇宙学家的追求是尽可能多地观测：绘制出整个可观测宇宙中物质和能量的分布。这项工作的目的不仅仅是制图。由于光速的有限性，我们看到的远处的结构与我们观察到的光子发射时的情况相同。这样，宇宙学家也能够在时间中旅行，因此宇宙学家的理想地图不仅描述了宇宙的当前状态，而且还描述了自大爆炸以来的演变。到目前为止，我们只能收集这张地图的单独部分，涵盖了宇宙演化的早期阶段，从大爆炸后不久发射的宇宙微波背景辐射（CMB）的测量，以及在这个演化中的晚期步骤，在我们周围星系分布的观测方面。然而，在接下来的十年里，我们将朝着完成宇宙学家的理想地图迈出巨大的步伐：至少有一半的可观测天空将通过不同的实验在广泛的电磁波谱范围内联合绘制，这些观测将覆盖比迄今为止可获得的更大的体积。然而，宇宙学信息以一种吸收性难题的形式编码到这些数据集中：不同的实验涵盖不同的径向和角度尺度范围，以及不同的能量状态，某些部分的数据最终由非宇宙学来源和仪器效应主导。因此，美丽的宇宙学家的地图必须小心地从原始实验数据中解脱出来，以免它不可避免地受到污染。该项目的重点是确定这些数据集的区域和组合，这些数据集对重建这张地图有价值，并且包含最相关的宇宙学信息，利用最先进的统计和计算工具。例如，该项目的主要目标之一是探测原始引力波，即宇宙大爆炸期间产生的时空涟漪，这可以告诉我们很多关于早期宇宙的物理条件。这些波在宇宙微波背景辐射的偏振中留下了印记，其振幅明显小于我们银河系的辐射，因此在研究前者之前，必须小心地将后者从数据中删除。随着宇宙学即将进入“大数据”时代，就像大多数其他科学分支目前正在做的那样，为这一项目开发的许多算法和方法将对从大气物理学到社会科学的广泛学科有用，宇宙学研究的计算需求也将成为技术发展的驱动力。

Cosmology studies the large-scale properties and evolution of the Universe. As such, cosmology is arguably one of the most complete branches of physics in that it must be able to describe the large-scale distribution and motion of matter, governed primarily by the gravitational forces, but also the intricate interactions between subatomic particles that dictated the rules during and after the Big Bang, as well as the violent physical processes that take place in galaxies and clusters of galaxies. However, the most characteristic feature of cosmology that separates it from other branches of physics, is the impossibility to replicate experiments. We only have one set of data: the Universe, and we cannot repeat it. This originally put cosmology in an awkward position, where due to the lack of experimental data, progress was mainly driven by theoretical work based on fundamental premises. Astonishingly, as astronomical observations improved, many of these theoretical predictions were actually found to be valid, and the last couple of decades have seen cosmology grow into a fully fleshed science driven by experimental observations.Since there is only one Universe to observe, the cosmologist's quest is to observe as much of it as possible: to map out the distribution of matter and energy in the entire observable Universe. The aim of this endeavour is not merely cartographic. Due to the finiteness of the speed of light, we see distant structures the way they were at the time the photons we observe were emitted. This way the cosmologist is also able to travel in time, and therefore the cosmologist's ideal map describes not only the current state of the Universe, but also its evolution since the moment of the Big Bang. So far we have only been able to collect separate pieces of this map, covering the early stages in the evolution of the Universe from measurements of the cosmic microwave background (CMB) emitted shortly after the Big Bang, as well as the late-time steps in this evolution, in terms of observations of the distribution of galaxies around us. However, in the next decade, large steps will be taken towards the completion of the cosmologist's ideal map: at least half of the observable sky will be jointly mapped by different experiments in a wide range of the electromagnetic spectrum, and these observations will cover far larger volumes than have been accessible so far.However, the cosmological information is encoded into these datasets in the form of an absorbing puzzle: different experiments cover different ranges of radial and angular scales, as well as different energy regimes, and certain sections of the data end up being dominated by non-cosmological sources and instrumental effects. The beautiful cosmologist's map must therefore be carefully disentangled from the raw experimental data, lest it be inevitably contaminated. This project focuses on identifying the regions and combinations of these datasets that are valuable to reconstruct this map, and that contain the most relevant cosmological information, making use of state-of-the-art statistical and computational tools. As an example, one of the main objectives of this project is the detection of primordial gravitational waves, the ripples in space-time originated during the Big-Bang, which could teach us a lot about the physical conditions in the early Universe. These waves leave an imprint in the polarisation of the CMB with an amplitude significantly smaller than the emission of our own galaxy, and therefore the latter must be carefully removed from the data before the former can be studied.With cosmology soon entering the era of "big data", as most other branches of science are currently doing, many of the algorithms and methods developed for this project will be useful for a wide range of disciplines, from atmospheric physics to the social sciences, and the computing needs of cosmological studies will also act as a driver for technological development.