数学家传记
英格丽·多贝西是一位比利时数学家和物理学家,她在图像压缩中的小波方面做出了重要工作。她是“增强研究生教育多样性”组织的董事会成员,该组织帮助女性获得数学高级学位。
英格丽·多贝西的父母是Marcel和Simone多贝西。两人现在都已退休,但Marcel曾是一名土木采矿工程师,而Simone曾是一名犯罪学家,对历史有浓厚兴趣,退休后还回到大学攻读历史学本科。
多贝西的父亲鼓励她在上学期间对科学产生兴趣。然而,即使还是个孩子,她对数学的兴趣就已显而易见。她写道[4]:-
我一直对事物如何运作以及如何制作东西感兴趣。例如,我非常喜欢编织和陶艺,而且我从小就喜欢这类手工活动。但我也对观察机械如何运作感兴趣,或者对某些数学事实为何成立感兴趣(比如,一个数如果其各位数字相加得到的数能被9整除,那么这个数就能被9整除——用73512和8577试试,它们都是9的倍数;而对于像7这样的数,就没有这么简单的整除规则了)。
在[4]中,多贝西回忆了她童年的其他记忆:-
在我八九岁时,玩洋娃娃时我最喜欢的事情就是给它们缝衣服。我喜欢尝试制作合身的纸样——让我着迷的是,通过把平整的布料拼在一起,就能做出完全不平整、而是贴合曲面的东西。大约在同一时期,当我晚上睡不着时,我会在脑子里计算2的幂:1、2、4、8、16、……(每次都乘以2)。这些数字很快就变得非常大,但我会坚持算很久。同样让我着迷的是,看到这些数字增长得有多快。
在比利时完成中学教育后,多贝西进入布鲁塞尔自由大学攻读物理学学位。本档案中的许多数学家最初都是作为数学家起步,后来才将他们的数学技能应用于其他科学学科。然而,多贝西是少数朝相反方向发展的人之一,她从物理学训练开始。她于1975年获得物理学学士学位,并继续在布鲁塞尔自由大学进行物理学博士研究。
从她获得学士学位时起,多贝西 就成为布鲁塞尔自由大学理论物理系的研究助理。她担任此职从1975年到1984年,在此期间的中途,1980年她凭借题为 Representation of quantum mechanical operators by kernels on Hilbert spaces of analytic functions 的学位论文获得物理学博士学位。此时她已经发表了相当多的论文,写了大约十篇文章。1978年 An application of hyperdifferential operators to holomorphic quantization 问世,随后是与 Dirk Aerts 合写的若干论文:A characterization of subsystems in physics; Physical justification for using the tensor product to describe two quantum systems as one joint system; A mathematical condition for a sublattice of a propositional system to represent a physical subsystem, with a physical interpretation;以及 A connection between propositional systems in Hilbert spaces and von Neumann algebras。在这四篇论文中第三篇的一篇评论中,E G Beltrametti 解释了这项工作以及这些合写论文中第二篇的背景:-
大卫·希尔伯特 空间量子力学的基本规则之一是,当两个物理系统,比如 和 S,被视为复合系统 S 的组成部分时,那么与 S 相关联的 希尔伯特 空间是与 和 S 相关联的 希尔伯特 空间 和 H 的张量积。这条规则在传统的量子力学逻辑代数方法中几乎没有得到什么论证。作者们为其论证做出了重要贡献。
1981年 多贝西 前往美国,尽管她继续保留在布鲁塞尔自由大学的研究助理职位,在那里度过了两年从事博士后工作。回到比利时后,她于1984年被任命为布鲁塞尔自由大学理论物理系的终身助理教授职位。事实上,她在1984年获得了她的第一个重要奖项,当时她被授予路易·恩潘物理学奖。这一享有盛誉的奖项每五年颁发一次,授予一位比利时科学家,依据是其29岁之前所做的工作。
1987年,多贝西回到美国,担任新泽西州詹姆斯·D·默里希尔AT&T贝尔实验室数学研究中心的技术人员职位。同年,她与A Robert Calderbank结婚,他也是一位数学家。
除了我们已经提到的事件之外,从数学的角度来看,1987年对 多贝西 来说是重要的一年。让我们设定背景并将她的工作置于其中。在20世纪80年代初,J Morlet 发现了一种表示地球物理信号的新方法。他没有使用 约瑟夫·傅里叶 变换方法来分析信号,而是有了使用小波的直觉想法,后来与 Alex 马塞尔·格罗斯曼 合作,通过引入连续小波变换将他的直觉置于坚实的数学基础之上。大约在1985年,多贝西 与 Yves Meyer 和 Alex 格罗斯曼 合作,引入了一种离散方法,使函数能够从一组离散值中重构出来。正是 多贝西 在1987年的突破,当她构造出紧支撑连续小波时,带来了许多重要应用。多贝西 认为,她之所以被视为数学家,是因为这些应用在物理学之外:-
……即使作为一名物理学家,我的工作也非常理论化、非常数学化。我开始对数学在物理学之外的应用(特别是在工程学中)感兴趣,这就是为什么我现在被视为一名数学家。
尽管她仍然留在贝尔实验室的员工队伍中,多贝西还是请假去担任大学职位。1990年,她在密歇根大学度过了六个月,次年她被任命为罗格斯大学数学系的教授,在那里度过了两年。在罗格斯大学期间,她于1992年出版了Ten lectures on wavelets。这本重要的书使她在1994年被美国数学会授予凯瑟琳·斯蒂尔数学阐述奖。该奖项的引文写道:-
小波的概念起源于许多领域,而多贝西的部分成就在于找到了这一概念产生的那些地方,并展示了所有方法如何相互关联。将小波用作分析工具就像约瑟夫·傅里叶分析一样——简单却非常强大。事实上,小波是傅里叶分析在频率和空间上同时局部化情形的推广。并且像约瑟夫·傅里叶分析一样,它既有理论方面也有实际重要性。...
伊恩·斯图尔特描述了小波分析的若干应用。以下是其中两个描述:-
也许迄今为止小波分析最著名的应用源于美国联邦调查局在1993年决定使用小波变换来编码数字化指纹记录。小波变换占用的计算机内存比传统的图像存储方法少,其使用预计可将指纹记录所需的计算机内存量减少93%。...
在过去二十年中,医疗中心开始采用各种基于扫描仪的成像系统,如计算机断层扫描和磁共振成像,这些系统使用计算机将扫描仪收集的数字化数据组装成身体内部结构的二维或三维图像。...一个质量差的数字化图像可以通过对其进行小波变换、去除不需要的分量,并对小波表示进行“逆变换”以再次生成图像来平滑和清理。该方法减少了患者在扫描过程中暴露于辐射的时间,从而使成像技术更便宜、更快捷、更安全。
1992年1月,多贝西在马里兰州巴尔的摩举行的AMS-MAA联合会议上作了Wavelets making waves in mathematics and engineering讲座。该讲座由美国数学会以录像带形式出版。G·沃尔特对视频的评论告诉了我们很多关于多贝西作为讲师风格的信息:-
大多数数学家讲课的视频都不太成功。他们试图传达思想,但摄像机却把我们的注意力集中在演讲者的外表和他或她的怪癖上,例如一条鲜艳的领带或捏鼻子的习惯。在这个多贝西关于小波的讲座视频中,情况正好相反。她通过伸展双臂来说明伸缩和平移的方式有助于理解小波理论的主要概念。此外,她对这个主题明显的热情有助于把我们的注意力集中在主题上而不是她身上,并促成了一个非常成功的视频。
多贝西从1992年到1997年是弗瑞兹·约翰 D and Catherine T MacArthur基金会的会士,并于1993年当选为美国艺术与科学院的成员。次年,多贝西离开了贝尔实验室,担任了她在普林斯顿大学数学系和应用与计算数学项目的现任教授职位。1997年至2001年间,她担任应用与计算数学项目的主任。American Mathematical Society在1997年进一步授予多贝西荣誉,颁发Ruth Lyttle Satter数学奖,以表彰:-
...她对小波及其应用的深刻而优美的分析。
National Academy of Sciences于1998年选举多贝西为成员,然后在2000年授予她四年一度的数学奖:-
...因在小波和小波展开方面的基础性发现,以及她在使小波方法成为应用数学的实用基本工具方面所起的作用。
除了我们上面提到的那些之外,她还因其卓越成就获得了许多荣誉。例如,她于1998年当选为电气和电子工程师协会的会士,并于同年获得了该协会信息理论学会技术创新金禧奖。1999年,她当选为荷兰皇家艺术与科学院,2000年她获得了众多荣誉,如布鲁塞尔自由大学的荣誉学位和Eduard Rhein基金会2000年基础研究奖。
多贝西被瑞士洛桑联邦理工学院(2001年)、法国巴黎皮埃尔和玛丽·居里大学(2005年)、意大利热那亚大学(2006年)、比利时哈瑟尔特大学(2008年)和牛津大学(2013年)授予荣誉学位。她被授予比利时佛兰德斯皇家艺术与科学院金质奖章(2005年)和ICIAM先驱奖(2008年)。2004年,她被任命为普林斯顿大学William R Kenan Jr教授。
多贝西于2007年当选为伦敦数学会荣誉会员。2015年,她因[7]当选为美国国家工程院院士:-
……对小波数学及其应用的贡献。
她于2018年获得香港城市大学颁发的威廉·本特应用数学奖;颁奖词记载8:-
多贝西在广泛的科学与数学学科中做出了杰出贡献,她在推动移动智能手机革命方面的工作堪称这一时代的象征。作为数学应用于艺术品伪造检测的先驱,多贝西积极支持在发展中国家和女性中鼓励对数学产生更大兴趣。
她关于小波的研究使她在2018年获得了复旦-中植科学奖:-
该奖项授予她,以表彰她对小波的卓越贡献,特别是正交多贝西小波和双正交CDF(Cohen-多贝西-Feauveau)小波,她在发展小波理论和现代时频分析方面的领导作用,这些从根本上改变了图像和信号处理,她对数据分析和科学计算许多领域的深远影响,以及她对图像压缩、模数转换和基于阈值化的反问题算法的贡献。
次年,她当选为德国利奥波第那科学院成员,并于2020年获得阿斯图里亚斯公主技术与科学研究奖。评审团表示[10]:-
多贝西引领了现代小波数学理论的发展,小波就像数学的心跳,使我们能够接近梵高并发现他的风格,或聆听宇宙看似噪声中蕴含的音乐,以及各种其他应用。简而言之,它们使我们能够可视化我们看不见的东西,聆听我们听不到的东西。
在2020年6月23日获奖时,她说:-
我非常荣幸能被选中获得这个奖项,既因为奖项本身的声望,正如历届获奖者的名声所证明的,也因为我的共同获奖者的身份,他们每个人都是我个人英雄画廊中的巨人。
她于2023年获得沃尔夫数学奖,成为首位获得该奖项的女性。引文写道[11]:-
多贝西因在小波理论和现代时频分析的创建与发展方面的工作而获得沃尔夫奖。她发现光滑、紧支撑小波,以及双正交小波的发展,改变了图像和信号处理与滤波。
她的工作在图像压缩、医学成像、遥感和数字摄影中具有极其重要的意义。多贝西还在发展调和分析的实际应用方面做出了无与伦比的贡献,将复杂的图像处理技术引入从艺术到进化生物学等领域及其他领域。
多贝西最重要的贡献是她在1988年引入了光滑紧支撑正交小波基。这些基彻底改变了信号处理,导致了数字化、存储、压缩和分析数据(如音频和视频信号、计算机断层扫描和磁共振成像)的高效方法。这些小波的紧支撑使得信号数字化时间与信号长度线性相关成为可能。这是信号处理研究人员和工程师能够快速将信号分解为各种尺度贡献的叠加的关键因素。
在随后与A Cohen和J C Feauveau的合作中,多贝西引入了对称双正交小波基。这些小波基放弃正交性以换取对称性。这种基更适合处理有限长度信号边界处出现的不连续性并提高图像质量。她的双正交小波成为JPEG 2000图像压缩和编码系统的基础。
Ingrid Daubechies' parents were Marcel and Simone Daubechies. Both are now retired but Marcel was a civil mining engineer while Simone was criminologist with a great interest in history which she went back to university to study as an undergraduate after she retired.
Ingrid's father encouraged her to take an interest in science while she was studying at school. However, even as a child an interest in mathematics was evident. She has written [4]:-
I was always interested in how things worked and how to make things. For instance, I really like weaving and pottery, and I have liked this kind of craft pursuit since my childhood. But I also was interested in seeing how machinery worked, or in why certain mathematical things were true (like the fact that a number is divisible by nine if, when you add all its digits together, you get another number divisible by 9 - try it with 73512 and 8577, both multiples of 9; there is no rule that is quite as simple for divisibility by 7, say).
In [4] Daubechies recalls other memories of her childhood:-
When I was eight or nine, the thing I liked best when playing with my dolls was to sew clothes for them. I liked trying to make patterns that would fit them well - it was fascinating to me that by putting together flat pieces of fabric one could make something that was not flat at all, but followed curved surfaces. Around the same time, when I couldn't fall asleep at night, I would compute the powers of 2 in my head: 1, 2, 4, 8, 16, ... (multiplying by 2 every time). The numbers became very large very quickly but I would keep going quite a while. It was fascinating, again, to see how fast these numbers grew.
After her school education in Belgium, Daubechies entered the Free University Brussels to read for a degree in physics. Many of the mathematicians in this archive have started out as mathematicians and have later come to apply their mathematical skills in other scientific disciplines. Daubechies, however, is one of the few who progressed in the opposite direction, starting off by training in physics. She obtained her Bachelor's degree in Physics in 1975 and continued to undertake research at the Free University Brussels for a doctorate in physics.
From the time that she was awarded a Bachelor's degree, Daubechies became a Research Assistant in the Department for Theoretical Physics at the Free University Brussels. She held this post from 1975 until 1984 and, midway through this period, in 1980 she was awarded her Ph.D. in physics for a thesis entitled Representation of quantum mechanical operators by kernels on Hilbert spaces of analytic functions. By this time she had already published quite a number of papers, having written around ten articles. In 1978 An application of hyperdifferential operators to holomorphic quantization appeared, then a number of papers written jointly with Dirk Aerts: A characterization of subsystems in physics; Physical justification for using the tensor product to describe two quantum systems as one joint system; A mathematical condition for a sublattice of a propositional system to represent a physical subsystem, with a physical interpretation; and A connection between propositional systems in Hilbert spaces and von Neumann algebras. In a review of the third of these four papers E G Beltrametti explained the context of both this work and the second of these joint papers:-
One of the basic rules of Hilbert space quantum mechanics is that when two physical systems, say and S, are viewed as the pieces of a compound system S, then the Hilbert space to be associated to S is the tensor product of the Hilbert spaces and H associated to and S. This rule found little justification in the traditional logico-algebraic approach to quantum mechanics. The authors make a significant contribution to its justification.
In 1981 Daubechies went to the United States, although she continued to hold her Research Assistant position at the Free University Brussels, spending two years there undertaking postdoctoral work. Returning to Belgium she was then appointed to the tenured position of Assistant Professor in the Department for Theoretical Physics at the Free University Brussels in 1984. In fact she received her first major prize in 1984 when she was awarded the Louis Empain Prize for Physics. This prestigious prize is awarded once every five years to a Belgium scientist on the basis of work done before age 29.
In 1987 Daubechies returned to the United States to take up the post of Technical Staff Member at the Mathematics Research Center of AT&T Bell Laboratories, Murray Hill, New Jersey. In the same year she married A Robert Calderbank, who is also a mathematician.
In addition to the events we have mentioned, the year 1987 was an important one for Daubechies from a mathematical point of view. Let us set the scene and put her work in context. In the early 1980s J Morlet had discovered new way to represent geophysical signals. Rather than using Fourier transform methods to analyse signals he had the intuitive idea of using wavelets and later, in collaboration with Alex Grossmann, he put his intuition on a firm mathematical basis by introducing the continuous wavelet transform. In about 1985 Daubechies, in collaboration with Yves Meyer and Alex Grossmann, introduced a discrete approach which enabled functions to be reconstructed from a discrete set of values. It was a breakthrough by Daubechies in 1987, when she constructed compactly supported continuous wavelets, which led to many important applications. Daubechies believes that she is considered a mathematician because these applications are outside physics:-
... even as a physicist, my work was very theoretical, very mathematical. I became interested in applications of mathematics outside physics (especially in engineering), and that is how I am now considered a mathematician.
Although she still remained on the staff at the Bell Laboratories, Daubechies took leave to take university posts. In 1990 she spent six months at the University of Michigan, and in the following year she was appointed Professor in the Mathematics Department at Rutgers University, spending two years there. While at Rutgers, she published Ten lectures on wavelets in 1992. This important book led to her being awarded the Steele Prize for mathematical exposition by the American Mathematical Society in 1994. The citation for the award reads:-
The concept of wavelets has its origins in many fields, and part of the accomplishment of Daubechies is finding those places where the concept arose and showing how all the approaches relate to one another. The use of wavelets as an analytical tool is like Fourier analysis - simple and yet very powerful. In fact, wavelets are an extension of Fourier analysis to the case of localization in both frequency and space. And like Fourier analysis, it has both a theoretical side and practical importance. ...
Ian Stewart has described a number of applications of wavelet analysis. Here are two such descriptions:-
Perhaps the best-known application of wavelet analysis to date derived from the U.S. FBI's decision in 1993 to use a wavelet transform for encoding digitised fingerprint records. A wavelet transform occupies less computer memory than conventional methods for image storage, and its use was predicted to reduce the amount of computer memory needed for fingerprint records by 93 %. ...
In the past two decades, medical centres had come to employ various kinds of scanner-based imaging systems, such as computed tomography and magnetic resonance imaging, that use computers to assemble the digitised data collected by the scanner into two- or three-dimensional pictures of the body's internal structures. ... a poor digitised image can be smoothed and cleaned up by taking a wavelet transform of it, removing unwanted components, and "detransforming" the wavelet representation to yield an image again. The method reduced the time of the patient's exposure to the radiation involved in the scanning process and thus made the imaging technique cheaper, quicker, and safer.
In January 1992 Daubechies gave a lecture Wavelets making waves in mathematics and engineering to a joint AMS-MAA meeting in Baltimore, Maryland. The lecture was published on a videocassette by the American Mathematical Society. G Walter's review of the video tells us much about Daubechies' style as a lecturer:-
Most videos of mathematicians giving lectures are not very successful. They are trying to convey ideas but the camera focuses our attention on the appearance of the speaker and his or her idiosyncrasies, for example a bright tie or a habit of pinching the nose. In this video of Ingrid Daubechies' lecture on wavelets, the reverse is true. Her manner of illustrating dilations and translations by stretching her arms contributes to an understanding of the principal concepts of wavelet theory. Furthermore, her obvious enthusiasm for the subject helps focus our attention on it and not on her, and contributes to a very successful video.
Daubechies was a fellow of the John D and Catherine T MacArthur Foundation from 1992 to 1997 and was elected to the American Academy of Arts and Science in 1993. In the following year Daubechies left Bell Laboratories and took up her present position of Professor in the Mathematics Department and the Program in Applied and Computational Mathematics at Princeton University. Between 1997 to 2001 she was Director of the Program in Applied and Computational Mathematics. The American Mathematical Society further honoured Daubechies in 1997 with the award of the Ruth Lyttle Satter Prize in Mathematics for:-
... her deep and beautiful analysis of wavelets and their applications.
The National Academy of Sciences elected Daubechies a Member in 1998, then made their four yearly Award in Mathematics to her in 2000:-
... for fundamental discoveries on wavelets and wavelet expansions and for her role in making wavelets methods a practical basic tool of applied mathematics.
She has received many honours for her remarkable achievements other than those we have mentioned above. For example she was elected a Fellow of The Institute of Electrical and Electronics Engineers in 1998 and, in the same year, was awarded their Information Theory Society Golden Jubilee Award for Technological Innovation. In 1999 she was elected to the Royal Netherlands Academy of Arts and Sciences, and in 2000 she received numerous honours such as an honorary degree from the Université Libre de Bruxelles and the Eduard Rhein Foundation 2000 Basic Research Award.
Daubechies was awarded an honorary degree by the Université Polytechnique Fédérale, Lausanne, Switzerland (2001), the Université Pierre et Marie Curie, Paris, France (2005), the Universita degli Studi di Genova, Genoa, Italy (2006), the Universiteit Hasselt, Belgium (2008) and Oxford University (2013). She was awarded the Gold Medal of the Flemish Royal Academy of Arts and Sciences, Belgium (2005), and the ICIAM Pioneer Prize (2008). In 2004 she was named William R Kenan Jr Professor at Princeton University.
Daubechies was elected a London Mathematical Society Honorary Member in 2007. In 2015 she was elected a member of the National Academy of Engineering for her [7]:-
... contributions to the mathematics and applications of wavelets.
She was awarded the William Benter Prize in Applied Mathematics from the City University of Hong Kong in 2018; the citation records that [8]:-
Professor Daubechies has made exceptional contributions to a wide spectrum of scientific and mathematical subjects, and her work in enabling the mobile smartphone revolution is truly symbolic of the era. A pioneer in mathematical applications for the detection of art forgery, Professor Daubechies is a prominent supporter of encouraging greater interest in mathematics in developing countries and among women.
Her work on wavelets led to her receiving the Fudan-Zhongzhi Science Award in 2018:-
The award is presented for her remarkable contributions to wavelets, especially the orthogonal Daubechies wavelet and the biorthogonal CDF (Cohen-Daubechies-Feauveau) wavelet, her leadership in developing wavelet theory and modern time-frequency analysis which fundamentally changed image and signal processing, her deep influence on many areas of data analysis and scientific computing, and her contribution to image compression, analogue-to-digital conversion and thresholding-based algorithms for inverse problems.
In the following year she was elected to membership of the German Academy of Sciences Leopoldina and in 2020 she received the Princess of Asturias Award for Technical and Scientific Research. The jury stated [10]:-
Ingrid Daubechies has led the development of the modern mathematical theory of wavelets, which are like mathematical heartbeats that enable us to approach Van Gogh and discover his style or to listen to the music enclosed in the apparent noise of the Universe, among many other applications of all kinds. In short, they enable us to visualise what we cannot see and listen to what we cannot hear.
On receiving the award on 23 June 2020, she said:-
I am very honoured to have been selected for this award, both by the prestige of the Award itself, as witnessed by the renown of past awardees, and by the identity of my co-awardees, each of them giants in my own personal gallery of heroes.
She was awarded the Wolf Prize in Mathematics in 2023, becoming the first woman to receive this award. The citation reads [11]:-
Ingrid Daubechies is awarded the Wolf Prize for her work in the creation and development of wavelet theory and modern time-frequency analysis. Her discovery of smooth, compactly supported wavelets, and the development of biorthogonal wavelets transformed image and signal processing and filtering.
Her work is of tremendous importance in image compression, medical imaging, remote sensing, and digital photography. Daubechies has also made unparalleled contributions to developing real-world applications of harmonic analysis, introducing sophisticated image-processing techniques to fields ranging from art to evolutionary biology and beyond.
Daubechies's most important contribution is her introduction in 1988 of smooth compactly supported orthonormal wavelet bases. These bases revolutionised signal processing, leading to highly efficient methods for digitising, storing, compressing, and analysing data, such as audio and video signals, computed tomography, and magnetic resonance imaging. The compact support of these wavelets made it possible to digitise a signal in time linearly dependent on the length of the signal. This was a critical ingredient for researchers and engineers in signal processing to be able to rapidly decompose a signal as a superposition of contributions at various scales.
In subsequent joint work with A Cohen and J C Feauveau, Daubechies introduced symmetrical biorthogonal wavelet bases. These wavelet bases give up orthonormality in favour of symmetry. Such bases are much more suitable for treating the discontinuities arising at the boundaries of finite-length signals and improving image quality. Her biorthogonal wavelets became the basis for the JPEG 2000 image compression and coding system.
正文里的方括号编号指向这里,悬停即可直接看到条目。书目保留原文——译了书名反而查不到文献。
原站列出的延伸阅读与外部数据库,照原样保留,目标多为英文页面。
原站的交叉引用。指向本站已镜像专题的留在站内,其余仍指回原站。