Personal Website of Yujie Xu
armech
  • Home
  • ARCH
    • Selected Designs >
      • Weihai Tower
      • Nutshell • Universe
      • Transfusion Pipe
      • Despina
      • Rock Climbing Wall
    • Other Design Projects
    • Simulation and Analysis
    • Other Academic Works
  • MECH
    • Work Experience
    • Other Academic Works
  • Art/Sport
    • Piano
    • Paintings
    • EXTRACURRICULAR ACTIVITIES
    • 2D
    • Photo Blog >
      • What we learn in Set Theory
  • CV
  • Essay
  • Courses
  • Projects
    • Energy Mapping >
      • Documentation
      • New Development of Visualization
      • Developing History
    • Senior Center (Archive) >
      • Event
      • Thoughts and Drafts
      • Draft Design (Round 1)
      • Technique and Products Searching
      • Pre-design
      • Integration of Elderly and Children
      • Related Reading and Notes
      • Case Studies >
        • Green Building Technology
      • Senior Housing FAQ
      • Related Newspaper Articles
      • Design Platform Prototype
    • Zero Energy House Case Study >
      • Code Compliance
      • Material
    • Howl's Moving Castle (Archive)
    • Geometry (3D Puzzle Refined Design) (Archive)
  • Blog
  • Climate Map View
Picture
Yujie Xu
PhD Student of Building Performance and Diagnostics, SOA
Carnegie Mellon University
yujiex at andrew dot cmu dot edu

Getting to Know Me
  • History
  • Interests
  1. Energy Mapping and Geographical Information System (GIS)
  2. Computational Geometry (CG)
  3. Building Related Computation Applications
  • Summary

History >>
My passion for Art and Science together shaped my history. The former took the form of an instinctive craving and gift for complicated forms, which lead me to pursue an  undergraduate degree in Architecture, at Tsinghua University, China. The latter triggered my passion for Mechanical Engineering in the final year of my undergraduate study, which further led to a two-year work experience in Mechanical Engineering.  Before that it also resulted in my success in science competitions in high school as well as my addiction to sci-fi.

2013 fall I began to pursue a graduate degree in Building Performance and Diagnostics, a technical sub-branch of Architecture in fulfillment of my passion for Art and Science, at Carnegie Mellon University. This was also the very moment when I encountered my destiny, which is the field of Computer Science. I have developed a deep love for Computer Science, because it is an area that better suits my mixed passion.

The origin of this passion was the course Principles of Computing (15-110) instructed by Prof. Dilsun Kaynar and Prof. Roger Dannerberg in 2013 fall. Through a great variety of topics, the course gave me a broad view of Computer Science and the assignments helped me discover an obsession and talent in coding. I was motivated  to implement additional functionality beyond the minimum requirements. I also found myself enjoy thinking about deeper questions which were sometimes beyond the scope of the course, such as the more detailed assumptions of ensuring constant time costs for hashing, NP-complete reductions and the method of cost bound analysis. Based on my interest in these more detailed ideas, Prof. Kaynar suggested a list of more advanced courses for me to look into that cover these topics and more in more detail and depth.

From that point on, I started a journey of exploration through course work and projects. 

In the Spring of 2014, I underwent extensive training and built-up my basic knowledge and programming skills. I took three computer science undergraduate courses (15-122, 15-150, 15-210) and one math course (21-127), together with two architecture courses and a piano lesson. Though my schedule became almost impossible to manage, I felt a great joy being exposure to such a bountiful feast of knowledge. My piano instructor, Prof. Carla Larocca said people would not believe that I had only been playing piano for a year and could already play difficult pieces by Mozart, Haydn and Schubert (Please click here to listen). The same thing happened in Parallel and Sequential Data Structures (15-210), instructed by Prof. Guy Blelloch and Prof. Margaret Reid-miller. Prof. Reid-miller said it was hard to believe I had only been programming for under a year. I originally planned to take it the next semester, but after the first lecture I was so obsessed that I found myself incapable of dropping it. The higher order and functional thinking was difficult for me to grab at first, but I finally eventually understood it, owing to the discipline I acquired from my CS and Musical training. In this semester, I had a great learning experience where I took a giant leap forward in my programming skills and dramatically strengthened my knowledge base. Currently my passion lies in three major areas:  
  • Computational Geometry (CG), 
  • Artificial Intelligence (AI),
  • Building Related Computation Applications.
<< Back to top
<< Skip to end

Computational Geometry (CG) >>
My enthusiasm for CG is a combined result of my past and present project experiences and my increased absorption of the Mathematical aspects of Computer Science. My most recent CG related project, 3D Puzzle, was done during the summer of 2014, where I was mentored by Prof. Daniel Sleator. I learned not only from the project itself, but also from the process of working with a great computer scientist.

I also broadened my exploration of computer systems during the summer by taking the course Introduction to Computer Systems (15-213). 

Although my formal CS journey  began at Carnegie Mellon, my passion and project experiences in CG date back to my undergraduate architecture studies. In the Weihai Tower Project, I applied a 3D Voronoi algorithm to generate a partition of  space from a set of input points. Back then, I first encountered the power of computer science to enhance creative activities, and now, I have begun to appreciate the underlying mathematical structures more. 

Experiences in Computer Science also provided me with new angles of viewing the field of Architecture, which I am currently in. When I took Advanced CAD BIM and 3D Visualization (48-568, please click here to see the project), I realized that AutoCAD is just like any other programming language. its alphabet is the set of geometry primitives such as points, line segments and curves and its reduction rules are transformation, rotation scaling and their combination. The interface shares many features with VIM and Bash even though they appear to be so different at first glance. And now after  my hands on experiences with programming languages such as C,  assembly,  Python,  SML, and  AutoCAD, I am  looking forward to the course Foundations of Programming Languages (15-312) which I plan to take next semester. By taking this course, I hope to develop a deeper understanding of the mathematical foundations of computer languages.

Some intermediate topics between CG and Parametric Design Practice I consider beneficial  for me to look at in further detail include:
  • The Grammar and Logic of 3D form Generation and Selection with Embedded Manufacturing or Assembly Constraints
3D form generation has a dramatically larger set of physical constraints than in 2D space. Neglecting these additional constraints in the design or generating stage will lead to a huge waste of work in generating non-buildable structures which are destined to be filtered out in later design stages. Under this knowledge, developing the methodology or logic to impose the physical constraints in the early stage might be helpful in improving the efficiency in form generating routines.
  •  The Machine Learning Approach or Human Computation Approach in New Form Generating
I learned about the game "Eterna'' in the last lecture of the Artificial Intelligence course that I attended this semester. Eterna  illustrates the idea that humans are much better at generating physically and structurally sound designs. This leads me to consider the possibility of integrating human computation into the processes for automatic form generation. 
<< Back to top
<< Skip to end

Artificial Intelligence (AI) >>
To balance the ratio of exploration and exploitation (terminology stolen from the Reinforcement Learning section), I attended the Artificial Intelligence (15-381) lecture this semester, extending my scope to the robotics sub-branch, from which I discovered my second area of interest. 

First, AI resonates with my Mechanical Engineering experience on developing precision positioning stages. The stages are agents that allow one control to follow a trajectory or move to a specific location with desired orientation in ultra-precision. Back then my work was mainly focused on the structural aspects, including the design, simulation and modeling of machine bodies and motors, and testing  sensors. Based on the knowledge of manufacturing and sensor error, I was curious about how this limited determinism, including the imperfect information about the world and the randomness in action execution, is combated to ensure the ultra-precision movement. The Motion Planning, MDP and Reinforcement Learning section settled this curiousness and extended my knowledge of Mechanical Engineering from structural aspect into the control aspect.

More importantly, I discovered a second field that attracts me a lot, namely  Game AI Programming and Machine Learning, which were covered a little in previous courses I have taken and were extended with more details on Searching, Game Theory and Learning in the AI course, with a supplementary guest lecture on Imperfect Information Game AI Programming of Poker, providing up-to-date insight into the current development of these algorithms. I'm also looking forward to some exposure to the area of Machine Learning next semester.

Another topic in AI Game Computing I will consider  diving  deeper into is:
  • Faster Model Adjusting Online Algorithms or Randomness in Game Computing to Cope with "Adversary'' players
In the guest lecture on poker computing, one of the obstacles brought about in Game Computing is that human players are often better at recognizing the strategies of computers and thus can adversely trick the computer into choosing the wrong move or classifying their opponent into a wrong category. However, the computers cannot use the same techniques to trick people, since they'll mainly act according to some previously learned model. Re-computing or adjusting the model in an on-line fashion is difficult at present  and is a meaningful topic for further exploration and research.
<< Back to top
<< Skip to end

Building Related Computation Applications >>
My third area of interest, Building Related Applications rose naturally from my architectural experiences. The built environment has crucial influence on all aspects of people's life, including productivity (my previous study on Acoustic Environment and Productivity: report, which is part of the BIDS project) and health (my study on the Urban Sprawl and Public Health: report, data processing log). Architecture is sometimes misunderstood as as being purely concerned with appearance, in much the same way as Computer Science is often misinterpreted as a field dealing with the corralling and wrangling of computational machines. Indeed, what you can see, namely the "appearance" is meant to achieve the quality of what you can't see, "the space". In a more romantic way of expressing this, it creates the opportunities of destined encountering of people and events and is blended into the history of individuals and communities. 

Based on the reasoning above, public participation throughout the whole building delivery process is crucial to ensure the quality of a building and the quality of life of its direct and indirect users. Although software such as Revit, ArchiCAD enhance the interactive working process amongst the whole design team (architects, structural engineer, mechanical engineer, interior designer, etc.), there is a need for tools that facilitate public participation in the design process, especially  regarding the building and near-building-environment. Based on limited literature searching, GIS is considered to be such a suitable platform for large scale urban planning (a list of related literature can be found here), but there is no suitable instance of such a platform for smaller scale building level design. 

In 2015 fall, I attended the lecture of Computational Geometry instructed by Prof. Gary Miller. The course took me through a journey through the aspects of CG in both graphics and machine learning applications and above all the method of backward analysis in solving complicated cost bounds and the method of considering problem in a higher dimension and project it down to the original dimension.
<< Back to top
<< Skip to end

Summary >>
Looking back, one of the most important  ideas I have learned dating back to the very beginning of my CS related history from Principles of Computing is the idea of a mathematical reduction between computer science problems. Specifically, this idea involves the underlying similarities between  problem structures. My  appreciation of the art of reduction is  one of the major reasons I'm captivated by Computer Science, and it is leading me toward learning about computational theory. This semester I took the Algorithm Design and Analysis (15-451) course ,taught by Prof. Daniel Sleator and Prof. Anupam Gupta, and also attended the lectures of Great Theoretical Ideas in Computer Science (15-251), instructed by Prof. Victor Adamchik and Prof. Venkatesan Guruswami, for fun. Both of these courses further strengthened my theoretical background and reassured me that my interest in  Computer Science  is not a passing  crush, but a life-long love. 

The goal of my future studies will be on building up a solid theoretical foundation, learning to do high quality research, exploring more areas and converging my interests, and, as always, polishing the "Computational'' thinking process.

(Updated on Dec. 28, 2014)
(Many Thanks to Bryce Summers who helped me revise the content)
<< Back to top

my path of arch and mech

图片

ARMECH

This site is a looking back of my zigzaging way of Architecture and Mechanical Engineering and then CS maybe.

Arch

ARCH is short for architecture. This section includes my architecture works: selected designs and other academic or extracurricular projects.

Selected Design
This section includes several of my selected self-satisfactory architecture design works.
    Weihai Tower        A parametric design applying the 3D voronoi algorythm in generating the form and space.                  
    Nutshell • Universe        An interior design for the disabled, with detail and device designs.           
    Transfusion Pipe        A group project with Chenchen Lv focusing on recovering of the polluted ecological environment with architectural approach.
    Despina        A meseum design with a special focus on the form searching with the mud draft sculpturing method.
    Rock Climbing Wall        A reform design in Tsinghua University, Beijing, China, carried out on 2010.
Other Design Projects
This section includes my other architecture design projects during my undergraduate study.
Other Academic Project
    Survey and drawing of ancient buildings        I'm responsible for the drawing of the Site Plan and the roof plan.
    Digital Fabrication_Huarun Project        A Laser cut and CNC machine modeling experience, considering manufacturing.
    Lunch on the lawn_Sitting Device Design        Sustainable design reusing old clothes, mowed grass and dried straw to build sitting devices       
图片
Enter

MECh

MECH is short for mechanical design. This section includes my mechanical projects during work and my undergraduate study
Work Experience
    Mechanical Design of Ultra-Precision Motion Stages
Other MECH Projects
    Machine Sketching Practice        Disassembling, measuring and drawing of a reduction gearbox
    Jumping Robot        Mechanical Design Competition
    
图片
Enter

OTHER

This page demonstrated what I did in my spare time:
    Painting
    Sports
图片
Enter
- Back to Top -
Powered by Create your own unique website with customizable templates.