两年前的暑假,我在怀化转车去成都,在 C 兄那里停留了一天。当时天气炎热, C 兄提议去郊区的腾古远水库游泳,我非常赞同。可是我没带游泳装备,最关键的是没带泳镜。自从读大学起,游泳时都是装备齐全的,泳裤、泳镜、泳帽一个不少。如果没有泳镜,还真不知怎么游了。当时如果去买泳镜,还不知道在哪里能买到,而且会耽误时间, C 兄便帮我在他隔壁哥们那边借了一副。 最近,在租住的公寓楼下有一个配套的小泳池,我经常去游。游着游着就想起了两年前那件借泳镜的事。心里头又来了一个问号,我如果不戴泳镜,还能游泳吗?这看似是个傻问题。既然会游泳了,不戴泳镜又何访。之所以有这个问号,是由于我游泳时已经习惯了在水下换气,这时候如果戴着泳镜,即使在水下也可以睁着眼睛。而如果我不戴泳镜,依旧按照已经习惯了的方式游泳则要么不能睁开眼睛,看不到东西;要么眼睛进水,引起眼睛不适。总之,如果不戴泳镜,按照我已习惯的方式游泳是不方便的。 我又想到了小时候在家乡的小河里游泳的场景,那时哪有什么游泳装备?个个小伙伴都游得很开心,虽然泳姿是野路子,不如现在的正规,但好歹不需要依赖任何装备就能游。记忆中有一副毛主席游长江的画面,画面上他老人家也没戴泳镜,不也游得好好的?如果哪天真来洪水了,难道我还要先去找副泳镜才能游吗?想到这些,我觉得该试着摘掉泳镜游泳了。为了能看见周围物体,又不让眼睛进水,我选择了让头始终露出水面,在水面上换气的方式。刚开始这样游的时候还挺费力的,也呛了几口水。这样游过几次之后,就熟练了,也轻松了。虽然效率比不上戴着泳镜游标准泳姿,但这对外界环境的依赖小啊。换句话说,不需要戴泳镜的游泳方式比标准的全副装备的游泳方式的环境适应性更好。 信息科学中的“鲁棒性”可以用来形容这两种游泳方式的不同。在维基百科上,对计算机科学中的鲁棒性的定义如下:“ In computer science, robustness is the ability of a computer system to cope with errors during execution and cope with erroneous input. ”即:计算机科学中,鲁棒性是指计算机系统在运行过程中处理错误和错误的输入的能力。还是以上述两种游泳方式为例来说明鲁棒性,如果有两个人,其中 A 在不戴泳镜和戴泳镜两种条件下都能游泳 , 而 B 只有在戴了泳镜时才能游泳。那么在标准的游泳比赛上(要戴泳镜),两人都可以参加。但是,当情况发生变化时,比如规定参加比赛者不许戴泳镜,这时 B 就不能参加了。这个例子中, A, B 对正确的输入指令(戴泳镜的标准游泳)都可以应付,但 B 对异常的输入指令(不戴泳镜游泳)则无法处理,而 A 可以处理。可见 A 的鲁棒性更好。 编过程序的人都对鲁棒性有切身体会,鲁棒性好的程序对一些异常情况都能较好处理,整体上可靠性更高,而鲁棒性差的程序则经常出问题甚至崩溃。鲁棒性对于我们的工作、生活也有重要的意义。例如:在看高水平的乒乓球 / 羽毛球比赛时,冠军通常都有一个共性:失误较少,而且在出现失误时能很快调整过来。这其实是鲁棒性好的一种表现。再比如,在篮球比赛中,普通球员投进一个时,我们不会说他水平高,因为他很难再进一个。相反,顶级球员即使在有多人干扰的情况下也能接二连三的进球。这就是鲁棒性高低的区别。 作为高校师生,也要努力提高自己的鲁棒性。多媒体教学设备普及后,大学教师的课堂教学都是用多媒体课件进行的。时间长了以后,可能很多教师对多媒体课件产生了依赖。如果突然没有了多媒体,估计有不少老师不知如何上课了。也常听学生讲“我必须在没有任何干扰的情况下才能学习”。这其实也是鲁棒性不好的表现。 如何提高鲁棒性呢?这是一个很重要的问题。我想到的是要多进行测试,发现问题并解决问题,以使自己在不同条件下都能得到相对稳定的结果。计算机科学和其他学科中肯定还有很多好的方法,在此抛砖引玉,希望大家给出更多的建议。
“世上没有免费的午餐”一般是指一个有关国家福利的经济原理,跟中国的“天上不会掉馅饼”的道理很接近。在机器学习中有一个“没有免费午餐定理”,说的是要想在某些指标上得到性能的提高,必须在另一些指标上付出相应的代价。 在控制界没有正式的没有免费午餐定理,但大家都知道,一个控制器的控制品质(响应速度和抗干扰能力)和鲁棒性(对模型误差的容忍度)一般是相矛盾的,即一般是提高品质会降低鲁棒性,反之亦然。有人也把这个矛盾叫控制系统的没有免费午餐定理,听起来合情合理。 那么能否找到一种控制算法,既提高控制品质,又提高鲁棒性?或者说控制技术中会不会有免费午餐? 做过 MPC 项目的人都知道,传统的工业 MPC 控制器的抗干扰性不是很好。为了提高抗干扰能力,我们在 MPC 算法中加了一个递推(自适应) ARMA 模型来描述不可测干扰,用来提高模型预测的精度。在无数次仿真中,发现我们的方法不仅能提高控制品质,还大大提高了鲁棒稳定性!后一点令我们非常吃惊。难道这是控制中的免费午餐?我请控制界的一些大牛们帮助分析一下,至今没有得到任何回复,估计是太难了。 如果理论上能证明我们提出的自适应干扰模型能同时提高控制品质和鲁棒稳定性,那就证明了控制中确有免费的午餐,那就好玩了。如果能找到反例,也很有价值。请理论好的同学帮帮忙。文献【 1 】是我们方法的详细介绍,文献【 2 】再一次提到这个问题。 发现一个现象,提出一个理论难题,是很有意思、有刺激的事。这样的事在控制界已经很少发生了。 到底有没有免费的午餐,天上会掉馅饼吗? 感觉是,如果你朝正确方向努力上十年、二十年,天上会掉馅饼的。 没有严格证明。。。 【1】 Xu, Z.H., Zhu, Y.C., Han, K.,Zhao, J., Qian, J.X. (2010). A multi-iteration pseudo-linear regression method and an adaptive disturbance model for MPC. Journal of Process Control .Vol. 20, No. 4, pp 365-558. 【2】 Zhu, Y.C., R. Patwardhan, S. B.Wagner and J. Zhao (2012). Towards a low cost and high performance MPC: the role of system identification. Computers Chemical Engineering . Vol.51, pp. 124-135.
据不完全搜索(普通外网与中国知网),“鲁棒”与“气候变化”在中文世界中似乎缺少直接关联,但在国外气候科学研究中“鲁棒“与”气候”已经喜结良缘,不少文献出现,现略举数例: 用两个高分辨率气候变化模拟情景分析欧洲极端降水变化的鲁棒性 hillside 2014-1-18 0 0 hillside 15 秒前 试用集合气候模型对月气温降水概率分布函数的变化预测值进行鲁棒性估算——英文论文 hillside 2014-1-18 0 12 hillside 22 分钟前 空间鲁棒的气候极值聚集预测研究——英文论文介绍 hillside 2014-1-18 0 18 hillside 1 小时前 局地降水极值未来变化的鲁棒性研究——英文论文介绍 hillside 2014-1-18 0 14 hillside 1 小时前 我个人判断,“集合预报”及“鲁棒性”都与所谓“不确定性”有着不解之缘,假以时日,它们之间应当可以建立关联(后来找到一点国外研究信息,见于附件)。 附1: PNAS文章确定全球变暖多重影响下的热点区域 国家科学图书馆《科学研究动态监测快报》2013年7月15日第14期(总第128期)气候变化科学专辑 根据PNAS在线发布的一项题为《全球变暖的多行业气候影响热点区域》(Multisectoral Climate Impact Hotspots in a Warming World)的最新研究成果,如果温室气体排放量有增无减,本世纪末全球可能有1/10的人口生活在气候影响的热点区域。在最糟糕的情景下,还有更多的人会面临粮食减产、水资源缺乏、生态系统失衡和人体健康等多重风险。 该研究由德国波茨坦气候影响研究所(PIK)、国际应用系统分析研究所(IIASA)和其他单位的研究人员合作完成,模拟在全球不同地区,气候变化对水的供应、疾病的传播、农业生产和生态系统的影响会如何重叠,结果确定亚马逊地区、地中海和东非地区的多个行业可能会经历剧烈变化。研究本身是跨行业影响模型对比计划(ISI-MIP)的部分产出成果之一。 文章的第一组作者,PIK的Franziska Piontek指出,气候变化对不同部门的重叠影响可能会相互作用,并因此对受影响地区的人民的生计带来数倍的压力,这就是该研究关注全球多个行业的原因。中国科学院合著者汤秋鸿指出,研究将过去30年中最干旱的3年设定为阈值,如果未来情景下水资源量低于该阈值,则认为气候变化给该区域带来水资源缺乏的环境风险。结果表明,当前的极端情况在未来可能成为常态。 多个不同气候模型的组合增加了结果的鲁棒性 和传播效果。共同作者之一,NASA戈达德空间研究所的Alex Ruane提出,多个模型预测得出的热点区域比单个模型模拟得到的要更可靠。 这是首例基于计算机模型全面模拟气候变化及其影响,并确定气候变化影响的行业热点的研究,也是全球气候影响的研究人员首次共同努力阐明人类正在面临的风险。研究的目的是为未来分析全球变暖的后果奠定新基础。 (裴惠娟 编译) 原文题目:Identifying Climate Impact Hotspots Across Sectors 附2: 气候变化再审视———非政府国际气候变化研究组报告 (见本人帖子: 介绍«气候变化再审视——非政府国际气候变化研究组报告» (2013年5月出版) ) 2 模型的本质问题 Susan Solomon (IPCC 第四次评估报告发布时第一工作小组的联合主席) 和在«美国国家科学院院报» (Proceedings of the National Academy of Sciences of the United States of America) 上撰文的其他3个作者指出, 由大气CO2 浓度增加导致的气候变化在排放停止后的1000 年内是不可逆转的(Solomon etal,2009)。 这在虚拟世界的计算机运行的气候模型中可能如此,但是在现实世界里未必如此。 这4位科学家提出气候参数建模需要符合三个标准条件: ①观测到的变化已经出现且有证据表明是人为因素引起的; ②各种现象基于充分理解的物理原理;③预测是有效的,且在模型间具有足够的鲁棒性。 对于Solomon等提出的第三条标准,许多计算机模型预测确实是有效的,并且在模型间具有鲁棒性,但这些模型的假设和具体时空上的结论通常存在巨大的分歧,甚至并不能证明彼此的有效性,也不能将这些不一致的预测进行组合产生有意义的平均值。许多研究发现,实际数据与模型预测相悖。因此, 认为这些模型是鲁棒的想法只是痴心妄想而已 。 附3: http://journals.ametsoc.org/doi/abs/10.1175/JCLI3990.1 Held, Isaac M., Brian J. Soden, 2006: Robust Responses of the Hydrological Cycle to Global Warming. J. Climate , 19 , 5686–5699. doi: http://dx.doi.org/10.1175/JCLI3990.1 Robust Responses of the Hydrological Cycle to Global Warming (水文循环对于全球变暖的鲁棒响应) Isaac M. Held National Oceanic and Atmospheric Administration/Geophysical Fluid Dynamics Laboratory, Princeton, New Jersey Brian J. Soden Rosenstiel School for Marine and Atmospheric Science, University of Miami, Miami, Florida Abstract Using the climate change experiments generated for the Fourth Assessment of the Intergovernmental Panel on Climate Change, this study examines some aspects of the changes in the hydrological cycle that are robust across the models. These responses include the decrease in convective mass fluxes, the increase in horizontal moisture transport, the associated enhancement of the pattern of evaporation minus precipitation and its temporal variance, and the decrease in the horizontal sensible heat transport in the extratropics. A surprising finding is that a robust decrease in extratropical sensible heat transport is found only in the equilibrium climate response, as estimated in slab ocean responses to the doubling of CO 2 , and not in transient climate change scenarios. All of these robust responses are consequences of the increase in lower-tropospheric water vapor. Received: September 13, 2005; Final Form: March 17, 2006 附4:http://www.iansa.eu/papers/IANSA-2012-01-dreslerova.pdf(全文下载) Human Response to Potential Robust Climate Change around 5500 cal BP in the Territory of Bohemia (the Czech Republic) (公元前5500左右波希米亚地区人类对于潜在的鲁棒气候变化的响应) Interdisciplinaria archaeologica(《跨学科考古学》杂志),Volume III ● Issue 1/2012 Dagmar Dreslerováa等 1. Introduction This contribution is dedicated to Marek Zvelebil. The range of his interests was admirably wide, from the life of hunters and gatherers through the Mesolithic – Neolithic transition and the beginning of agriculture, to the study of the ancient landscape as a whole. He used various approaches to solve archaeological themes including ethnography, linguistics, or environmental studies. In the latter respect he did not fear being accused of environmental determinism, as the example of an explanatory model for the Mesolithic-Neolithic transition in Denmark, published together with Rowley-Conwy (Rowley-Conwy 1984, Zvelebil, Rowley-Conwy 1984) demonstrates. The expressions flux and transition were often used in his work but may also be used in order to characterise Marek Zvelebil himself. Landscape, environment and flux will also be the subject of this article. It attempts to answer the question as to whether the observed change in spatial distribution of archaeological evidence from the Neolithic and Eneolithic (in the sense of the Middle and Late Neolithic in NW Europe, ca. 4200–2200 BC) might have been caused by a change in climate or whether this phenomenon was independent from external forces and a result of cultural factors. The previous climate, as the most important agent influencing the alteration of all other parts of an environment, is the subject of many scientific disciplines, although the outcomes are, despite tremendous efforts, still somewhat unsatisfactory. The main reasons for this are: the complexity of the climate system as such, the regionality of the climate, the short history of its direct instrumental measurement, the evaluation of the climatic parameters in relative terms (e.g. wetter, drier), the varying sensitivities of the proxies, and the difficulties of their more precise dating. Previous allegations can be illustrated by comparing proxy data supported by warmer/drier and cooler/wetter climate phases at ca. 6000 cal BP in Britain and north-west Europe (Schulting 2010) or in the eastern Mediterranean and adjacent regions over the past 6000 years (Finné et al. 2011). In both cases the proxies from the same period of time vary enormously in spite of the relative geographical proximity of the areas. 附5: http://www.theccc.org.uk/blog/climate-science-remains-robust-despite-claims-in-the-mail/ 气候科学依然鲁棒 (摘录,全文可见链接) Climate science remains robust despite claims in the Mail Parliament has given the CCC the responsibility to constantly monitor the science upon which we depend to guide our responses to the threat of climate change. Like all scientists, we always take a sceptical stance, testing each assertion against the evidence and ensuring proper peer review of every important finding. Although we turn to the experts to find the best available facts, we assess material from other sources carefully, to ensure that we have missed no useful insights. The Mail on Sunday recently published an article by David Rose challenging fundamental aspects of climate science and suggesting that action now to reduce greenhouse gas emissions is unwarranted. In accordance with our responsibility, we considered the three main claims, which focus on:………… 附6: http://onlinelibrary.wiley.com/doi/10.1002/wea.543/pdf Robust adaptation to climate change ( 气候变化的鲁棒性适应) Robert L. Wilby 1 and Suraje Dessai 2,3 1 Department of Geography, Loughborough University,Leicestershire, UK 2 School of Geography, University of Exeter, Exeter, UK 3 Tyndall Centre for Climate Change Research, UK Introduction Coupled ocean/atmosphere general circulation models (OA/GCMs) have been instrumental in showing the need for global action to curb the anthropogenic emissions that cause climate change. It may be contested, however, that these tools have been less helpful in informing how to adapt at regional and local scales (Schiermeier, 2007). Given the legacy of past emissions and the prospect of unavoidable climate change, the case for committing more financial and technical resources to adaptation is gaining ground (UNDP, 2007; Parry et al., 2009). This poses a challenging question: how can we ensure that adaptation measures realize societal benefits now, and over coming decades, despite uncertainty about climate variability and change? The scientific community is developing regional climate downscaling (RCD) techniques to reconcile the scale mismatch between coarse-resolution OA/GCMs and location-specific information needs of adaptation planners. The resulting ‘scenarios’ are regarded as plausible descriptions of the future climate that reflect the influence of local topography and/or land-sea effects, and their interactions with changing synoptic-scale weather patterns under rising concentrations of greenhouse gases. Thanks to widely available and user-friendly tools, the volume of peer-reviewed research on RCD has grown dramatically over the last decade (Wilby et al., 2009). It is becoming apparent, however, that downscaling also has serious practical limitations, especially where the meteorological data scenarios. The value of high-resolution climate change scenarios for long-term planning may be questionable wherever climate variability is already stressing human and environmental systems. For instance, parts of North Africa and the Middle East are facing a water crisis due to rapid population growth, combined with weak governance, climate variability and limited renewable supplies. Under these circumstances, even achieving Millennium Development Goals (MDGs) by 2015 (such as access to safe drinking water) may seem a remote prospect, let alone sustaining progress under climate change to the 2020s and beyond. Hence, there have been calls to re-examine the ways in which climate risk information is used in adaptation and development planning (Dessai et al.,2005). This paper begins by comparing two different approaches to climate risk assessment in adaptation planning. We then describe a framework for robust adaptation decision-making that departs from traditional ‘predict and provide’ methods. We draw upon examples from the water sectors of developing and developed countries as evidence of how significant progress can be made in the majority of cases without climate change projections. Our views are also shaped by recent experiences of supporting adaptation in practice. 附7: http://www.ghanabusinessnews.com/2013/06/06/african-countries-need-robust-climate-change-policies-fao/ 非洲国家需要鲁棒的气候政策 African countries need robust climate change policies – FAO Page last updated at Thursday, June 6, 2013 14:14 PM // Leave Your Comment There is the need for African countries to formulate and implement robust policies, to mainstream climate change adaptation and mitigation into national and sub-national development plans. Dr Lamoudia Thiombiano, Country Representative, Food and Agricultural Organisation (FAO) of the United Nations, who made the call, also said there was the need for these countries to increase public and private investments in agriculture and rural transformation. He was speaking at the ongoing 2nd Climate Change and Population Conference being hosted by the Regional Institute of Population Studies (RIPS), University of Ghana, Accra. - See more at: http://www.ghanabusinessnews.com/2013/06/06/african-countries-need-robust-climate-change-policies-fao/#sthash.ey1rzCDS.dpuf 附8: http://www.nber.org/papers/w16440 关于非洲内战与气候变化之间存在鲁棒性的肯定性研究及否认其间鲁棒性存在的反驳论文 Climate and Civil War: Is the Relationship Robust? Marshall Burke, John Dykema, David Lobell, Edward Miguel, Shanker Satyanath NBER Working Paper No. 16440, Issued in October 2010 A recent paper by Burke et al. (henceforth “we”) finds a strong historical relationship between warmer- than-average temperatures and the incidence of civil war in Africa (Burke et al. 2009). These findings have recently been challenged by Buhaug (2010) who finds fault with how we controlled for other potential explanatory variables, how we coded civil wars, and with our choice of historical time period and climate dataset. We demonstrate that Buhaug’s proposed method of controlling for confounding variables has serious econometric shortcomings and show that our original findings are robust to the use of different climate data and to alternate codings of major war. Using Buhaug’s preferred climate data under sound econometric assumptions yields results that suggest an even stronger relationship between temperature and conflict for the 1981-2002 period than we originally reported. We do find that our historical relationship between temperature and conflict weakens over the last decade, a period of unprecedented African economic growth and very few large wars. This paper is available as PDF (250 K) or via email . 附9: http://www.jydoc.com/article/9151.html 鲁棒性——健康的系统 时间:2005年04月11日 作者:普通高中通用技术课程 点击:48次 控制系统的鲁棒性研究是现代控制理论研究中一个非常活跃的领域,鲁棒控制问题最早出现在上个世纪人们对于微分方程的研究中。Black首先在他的1927年的一项专利上应用了鲁棒控制。但是什么叫做鲁棒性呢?其实这个名字是一个音译,其英文拼写为Robust。也就是健壮和强壮的意思。控制专家用这个名字来表示当一个控制系统中的参数发生摄动时系统能否保持正常工作的一种特性或属性。就像人在受到外界病菌的感染后,是否能够通过自身的免疫系统恢复健康一样。