   Python tricks     Widgets      Progress bar in Python     bash pip install tqdm      Then in the script      python from tqdm import tqdm  for i in tqdm range 100 , desc  Loading...          ...      If the description needed to be updated at each iteration      python from tqdm import tqdm  for i in  pdar    tqdm range 100        pdar.set description  ...        ...          Matplotlib colors    List of named colors   Matplotlib 3.9.2 documentation       img   My preferred ones     goldenrod   teal   olive   lightcoral     Environment  See also     Conda frequent commands      Export conda  environment.yml     Exporting the environment.yml file   conda 25.3.2.dev62 documentation     bash conda env export   environment.yml         ...

Python tricks

   Deep learning environment setup     Install Ubuntu 22.04 LTS    Install Ubuntu Desktop   Ubuntu   Ubuntu Releases  Select the right version of Ubuntu is important. Softwares for deep learning like CUDA and NCCL provide specific compiles for different Ubuntu versions  not all Ubuntu versions are available. In 2025, Ubuntu 22.04 and Ubuntu 20.04 are the safe choices.  After install the OS, full upgrade of kernel packages must be done. Otherwise, newly added packages could clash.     bash sudo apt update sudo apt full upgrade   yes sudo apt autoremove   yes sudo apt autoclean   yes reboot         Fish shell      Install    How to Install and Set Up the Fish Shell   by Saad Jamil   Medium     bash sudo apt add repository ppa fish shell release 3 sudo apt get update sudo apt get install fish         ...

Deep learning environment setup

   Good reads of 2024  Finally achieved the goal of reading 12 books in 2024    Having finished these books, I don t feel particularly different. Every day in life is still the same as always. But perhaps my perception of life has become sharper than before. If the length of life cannot easily be extended through personal effort, then perhaps maintaining sensitivity in perception is equivalent to broadening the width of life.   img  Following are my favorite books   , movies   , podcasts    , and articles    of this year. I hope that in 2025, I can also strive to remain spiritually free.     Favorite Book of 2024     Hayek s  The Road to Serfdom  was the most challenging book I read this year, so much so that I still haven t finished organizing my notes its sentences feel as if they are engraved on stone tablets, impossible to rewrite or condense, only to be quoted. Looking back, the book speaks of nothing more than common sense, but holding on to common se...

Good reads of 2024

   Git frequent operations     Basic configurations  User name       git config   global user.name   Your Name        Email       git config   global user.email   email example.com        Main branch name       git config   global init.defaultBranch  name        P.S. Change main branch name of an existed repository        git branch  m  old name   new name        If it has been pushed to remote        git push  remote    delete  old name  git push  u  remote   new name          Local repository operations      Initialize the repository      git init          Current branch operations       Stage changes      git add  file        P.S. Use   f  to force stage a file even if it s set to be ignored in  .gitignore .     Unstage the file changes       git restore   staged  file        P.S. Unstage all files  cahnges        git reset        Discard the changes in the working directory       git restore .          git restore  file        P.S. The difference is  git restore   staged  file   af...

Git frequent operations

   Travel time analysis with Apple Shortcut     Motivation  Travel during public holidays in China is a headache inducing activity, especially during national day holidays, one of the longest public holidays. People tend to pour out to visit their family friends or travel to long expecting tour sites. Traffic could be overwhelmingly heavy during this period.  To deal with this situation, some schedule their travel time carefully. Long before the off duty time of 31 Sep, the day before the holiday begins, people check their map apps anxiously again and again to get a glimpse of the development of traffic jams. Experiences of the best time for departure spread on social media, though they tend to convey conflicting speculations. I got on the road at 7 30 on 1 Oct, hoping an earlier departure could make my journey easier, yet I got stuck on the road for around 10 hours, which was around 3x the ordinary travel time.   P.S. I later realized that I almost chose the worst time to go...    P.S...

Travel time analysis with Apple Shortcut

   Cal Newport s time management system     Cal Newport    s time management system is summarized in The Time Block Planner   Cal Newport. Combining the techniques discussed in this tiny pamphlet and his other books podcasts, I compiled this blog post.     Core value of time management     Cal Newport advocates his time management system in his books  and youtube channel. It s both time efficient, lightweight, and deliver the most important values of any time management system     Time management  frees your mental stress  on remembering all the todos   dues bouncing around your head.   Time management has you  articulately  think about how can you complete your tasks in time. By doing this, you will soon realize that the task is more complicated than your expectation and you don t have so many allocable time, which boost motivation.    e.g. Think about  PhD application  ...

Cal Newport's time management system

   latexdiff  mark changes of LaTeX documents  When submitting academic manuscript, it s usually mandatory to submit the marked changes for revisions resubmissions. Word is particularly handy in this regard. But if your manuscript is consists of lots math formula, e.g. in the field of deep learning,   LaTeX  is a more common choice. In this case,  latexdiff  is the go to choice for generating the document with marked changes.     Install  latexdiff   Follow the instruction on the official documentation     GitHub   ftilmann latexdiff  Compares two latex files and marks up significant differences between them. Releases on www.ctan.org and mirrors     Prepare files  Copy the whole folder of the two versions of   LaTeX  documents that are being compared, say one is  draft 0  while the other is  draft 2 . Make sure that each of them can be successfully compiled.      About the figures and other resources  If you stored figures in the other folder, y...

latexdiff: mark changes of LaTeX documents

   How to set the padding of convolution to make the stride acting as a scale factor   img    Cover image source  A guide to convolution arithmetic for deep learning   Vincent Dumoulin, Francesco Visin   ArXiv  When doing convolution, the kernel is shift around the input image. Let s consider the case when the kernel moves left to right. The up to down scenario are just the same.     Constrains  Let s denote the padding size as  p , the image width as  w , the kernel size as  k , and the stride as  s . Then the total length of image width plus padding should be     2p   w    At the begging, the kernel is placed on the left of this total length, and is then moved towards right side with step length of  s , until no more space is available. Denotes the number of the placable potions as  n , then the output width of the convolution is also  n . It should satisfys     2p   w   s   k    n 1  s  leq 2p   w    a   When will the ...

How to set the padding of convolution to make the stride acting as a scale factor

   Why transposed convolution can be used to reconstruct activating features from the input images    Cover image source  A guide to convolution arithmetic for deep learning   Vincent Dumoulin, Francesco Visin   ArXiv  In ZF Net, the transposed convolution is used to approximate the inverse of convolution, leading to the reconstruction of the activating features from the input image that activates a particular layer. However, strictly speaking, the transposed convolution is not really the inverse of convolution. Then why can it be used to do the reconstruction      What is transposed convolution  An ordinary convolution can be written in the form of   Y     C     X  . Then the corresponding transposed convolution is defined as      X      C  T    Y        What yields the largest response  The convolution   Y     C     X   and be transformed to the vectorized form     y i    c  i T  x    where   x   is...

Why transposed convolution can be used to reconstruct activating features from the input images

   A complete guide to setup WSL  Windows Subsystem for Linux   WSL  Windows Subsystem for Linux  has various great advantages against dual boot  installing Linux and Windows on the same computer , among them the most significant is that you can literally running both systems at the same time without the need to stop every software that s running and reboot to another system. Procedure to set up a WSL is roughly the same as setting up a real Linux OS, therefore this guide can also be used as a reference to Linux setup. The specifications of WSL setup procedure are mainly in  Install WSL ,  CUDA and PyTorch ,  VTK with PyVista, and  Remote development with SSH  as described below.     Install WSL    Install WSL   Microsoft Learn  In PowerShell, run       wsl   install      After restart the computer, the terminal will prompt you to set the user name and password for the WSL. Then you can then enter the linux subsystem by enter  ws...

A complete guide to setup WSL (Windows Subsystem for Linux)

   Setup Windows machine as SSH host  In the previous blog post, I set up the development environment on a Windows machine. I further set it up as a  ssh  server that can be access by my laptop over the internet. Following is the summarization of some critical steps and relevant materials.     Setup SSH host      Install OpenSSH    Get started with OpenSSH for Windows   Microsoft Learn  Check installation status of OpenSSH       Get WindowsCapability  Online   Where Object Name  like  OpenSSH        Install the OpenSSH Client       Add WindowsCapability  Online  Name OpenSSH.Client    0.0.1.0      Install the OpenSSH Server       Add WindowsCapability  Online  Name OpenSSH.Server    0.0.1.0      Set PowerShell as the default shell    the shell you enter when ssh into this host        New ItemProperty  Path  HKLM  SOFTWARE OpenSSH   Name DefaultShell  Value  ...

Setup Windows machine as SSH host

   Setup Windows development environment  Recently, I tried to set up a Windows machine for development, focusing on Python, machine learning  PyTorch with CUDA , and web development  Node.js with NPM . Following is the summarization of some critical steps and relevant materials.  Later, I further set it up as a  ssh  server that can be accessed by my laptop over the internet. If you are interested, please check this post.     PowerShell  In Windows, PowerShell is more powerful than the legacy  cmd  shell in terms of system administration, configuration management, and perform administrative tasks.       Terminal execution policy  Firstly, enter PowerShell in administrator mode and modify the execution policy to enable shell scripting       Set ExecutionPolicy  Scope CurrentUser  ExecutionPolicy Unrestricted          Powershell autocomplete    Fish like Autosuggestion in Powershell   DEV Community...

Setup Windows development environment

   LaTeX frequent snippets     Math  Matrix      latex  begin bmatrix      x  11    x  12    x  13     dots   x  1n         x  21    x  22    x  23     dots   x  2n         x  d1    x  d2    x  d3     dots   x  dn      end bmatrix       Matrix bolding        mathbf      Transpose         top        hat x,  tilde x,  overline x         hat x,  tilde x,  overline x      Additional symbols     List of LaTeX mathematical symbols   OeisWiki     Listing     latex  begin itemize       item   end itemize         latex  begin enumerate       item   end enumerate         latex  begin lstlisting  language Python    end lstlisting        P.S. The frame need to be set as   fragile   to use  lstlisting        latex  begin frame  fragile     begin lstlisting  language Python    end lstlisting   end frame          Table     latex  begin table  h    centering   caption       label tab       vskip5pt   begin tabular  lcc     toprule            ...

LaTeX frequent snippets

   Conda frequent commands  When working on a Python project, it s considered good practice to create a virtual environment. In this case, every project is running in a separate environment, with their own dependencies. This can save you tons of headache   if you install all your projects  dependencies in one place, things may quickly crash with each other.    I personally learned this lesson the hard way after spending an entire night trying to fix version crashes. Eventually, I had no choice but to delete everything and start from scratch.  You certainly do not want to experience the same frustration as me   and you don t have to. With  conda  virtual environment, all your projects have their own territory to play and there is zero chance for crashing with each other.  To gain a proper understanding of  conda , the following resources will be helpful. Additionally, some commonly used commands will be extracted and listed in this blog post for easy reference.      Managing conda   Con...

Conda frequent commands

   Sphinx   ReadTheDocs  My optimized documentation workflow  Sphinx is an excellent tool for documenting Python projects. It can generate comprehensive documentation from docstrings in the code, which means that the comments you write for your code also serve as its documentation.  If you are not familiar with writing docstrings or Sphinx documentation, I highly recommend checking out this tutorial     A  How to  Guide for Sphinx   ReadTheDocs   sglvladi  My workflow was adapted from this tutorial with some personalization. Here is an example    example     Folder structure and setup  Under the root folder of the project, there are  docs   for storing documentation,   projectna...

Sphinx + ReadTheDocs: My optimized documentation workflow

   Embed GIF and MP4 in LaTeX Beamer  As an research postgraduate student, I frequently make  academic looking  slides for meetings, which comprise a lot of references, tables, charts, and sometimes math equations. LaTeX is excellent for typesetting these elements, so using LaTeX to make slides sounds like a good idea.  With tutorials like Beamer   Overleaf, I got started with this approach and feel more and more comfortable. I even made a series of templates for me and my friends  for more details  Beamer LaTeX Themes . Here are some example pages    PolyU  math  However, LaTeX only generates PDF files, which are usually static. Excluding videos and GIFs completely from your presentation  How pitiful. I searched the internet for a while a...

Embed GIF and MP4 in LaTeX Beamer

 LaTeX is a great tool for academic writing. However, not all collaborators may be familiar with it, and at times, you may need to convert your document into a Word file for reviewing and editing purposes. In my opinion, the best tool for this task is Pandoc   an open source command line tool for free.  Following the tutorial below, I installed  pandoc  and successfully converted  paper.tex  to  paper.docx      How to Convert from Latex to MS Word with  Pandoc    Zhelin Chen  In this blog, I will share a simple command snippet that converts LaTeX to Word file. Additionally, I will also discuss some aspects that were not fully covered in the aforementioned tutorial.     Conversion command  After runing this command, a Word file will be generated in the  . word   folder       pandoc paper.tex  o word paper.docx   reference doc word template.docx   bibliography bibliogra...

Convert LaTeX to Word

   CVPR 2025 oral notes   P.S. The paper descriptions are based on my personal understanding. Some text were extracted from the abstracts.      Domain      Vision language models  VLMs     Dataset    Spatial reasoning     PixMo MattDeitke2025CVPR provides a dataset of highly detailed image captions for pre training, a free form image Q A dataset for fine tuning, and an innovative 2D pointing dataset. It s used to train an fully open sourced 72B VLM from scratch, outperforming some proprietary models          VSI Bench JihanYang2025CVPR presents a video based visual spatial intelligence benchmark of over 5,000 question answer pairs. It shows that spatial reasoning capabilities remain the primary ...

CVPR 2025 oral notes

   ICLR 2025 oral notes   P.S. The paper descriptions are based on my personal understanding. Some text were extracted from the abstracts and reviews.      Domain      LLM    Transfusion ChuntingZhou2025ICLR combines  next token prediction  for text and  diffusion based learning  for images within a single transformer architecture, bridging the modality gap without quantizing images into discrete tokens        Embedding    AlexIacob2025ICLR introduces a pre training framework that decouples embedding layers from the transformer body, enabling robust training on heterogeneous data   avoiding the curse of multi linguality  , improving generalization, and reducing memory footprint    ZiyueLi2025ICLR  investigates the limitation of using decoder only models for embedding and finds that a good embedding can be acquired from the MoE layer, by combining rou...

ICLR 2025 oral notes

   How Computers Really Work   Matthew Justic   P.S. Part names are added by me. Some chapters are merged and renamed.      Part I  Basics      Electrical circuits    Term   Explanation   Measured in   Water analogy                               Electric charge    Causes matter to experience a force   Coulombs   Water     Electric current   The flow of charge   Amps   Flow of water through a pipe     Voltage   The difference in electric potential between two points   Volts   Water pressure     Resistance   A measure of the difficulty for current to flow through a material   Ohms   The width of a pipe      Ohm s Law     I    frac V  R      Kirchhoff s voltage law     The sum of the voltages around a circuit is zero.      Binary    Analogy  v.s.  Digital   Decimal  v.s.  Binary  v.s.  Hexadecimal   Binary data   Text, image, etc.    8 Bits  b    1 Bytes  B     1 KB    1024  Bytes    2  10   Byte    1 MB    1024  2   Bytes    2  20   Bytes    1 GB    1024  3   Bytes    2  30   Bytes    1 ...

How Computers Really Work | Matthew Justic

   Straight line Leadership   Dusan Djukich  Sam Matla described this book as the one that changed his life in this youtube video. After reading it through, I had to say that this book does have a transformative power in it. The  distinctions  delivered in this book forces me to face the reality and, definitively, helped me jump out of the  vicious bubble  I created for myself to keep procrastinate.  Indeed, usually it s ourselves who stumble our productivity and refuse to see the straight line to our goal, because facing the  straight   real solution  always expose ourselves to risks and uncertainty, which is very uncomfortable. We may even  invent  things to do to insulate us from it     Waiting for the best time to do something    Endless information collection without taking any action    Trying to find the best productivity note taking app that solve all of your problem once and for all    ...  You nam...

Straight-line Leadership | Dusan Djukich

   NeurIPS 2024 oral notes   P.S. The paper descriptions are based on my personal understanding. Some text were extracted from the abstracts and reviews.      Domain      LLM    ZhenghaoLin2024NeurIPS scores training tokens using a reference model and then training the language model with a focused loss on tokens with higher scores selectively, proving  not all tokens are what you need    Quantization    HaokunLin2024NeurIPS utilizes rotation and permutation transformations to more effectively mitigate both massive and normal outliers when quantizing LLMs    VladimirMalinovskii2024NeurIPS proposes a quantization aware strategies for fine tuning the LLMs after quantization, improving its performance especially in extreme compression   Alignment    JiamingJi2024NeurIPS proposes Aligner, a small model that l...

NeurIPS 2024 oral notes

   The Road to Serfdom   Friedrich Hayek    Nor am I arguing that these developments are inevitable. If they were, there would be no point in writing this.    Friedrich Hayek    The ideas of economists and political philosophers, both when they are right and when they are wrong, are more powerful than commonly understood... Madmen in authority, who hear voices in the air, are distilling their frenzy from some academic scribbler of a few years back.    John Maynard Keynes   P.S. Part name was added by me.      Introduction    The supreme tragedy is still not seen that in Germany it was largely people of good will, men who were admired and held up as models in the democratic countries, who prepared the way for, if they did not actually create, the forces which now stand for everything they detest.    Mere hatred of everything German instead of the particular ideas which now dominate the Germans is, moreover, very dangerous, because it blinds those who indulge in it against a real threat....

The Road to Serfdom | Friedrich Hayek

   Mastering Your PhD   Patricia Gosling   Bart Noordam   P.S. Part name was added by me. Some chapters are reordered.      Part I  Getting Started      Introduction  As a PhD student, it s important to understand the larger scene of the academic system and look from your supervisor s perspective. The author first introduce 3 typical stages of professorship, corresponding to 3 types of research teams     The start up group  Assistant Professor    Your supervisor is in his her junior years, whose career depends on your outcome. You will have frequent   intense interactions will your supervisor.   The up and running group  Associate Professor    Your supervisor is in his her mid career, who has a lot of responsibilities and less time for you. Your fellow graduate students are your main support.   The mature group  Full Professor        The small but established group       The large and established group  the empire       Getting Started  At the beginning of your PhD     Become familiar ...

Mastering Your PhD | Patricia Gosling & Bart Noordam

   Computing and Processing Correspondences with Functional Maps   Maks Ovsjanikov et. al    Computing and processing correspondences with functional maps   ACM SIGGRAPH 2017 Courses      Part I  Functional Maps Basics   P.S. Part name was added by me.       What are Functional Maps     Shape correspondence can be represented as the transportation  T F  of  functions  between shapes. If such functions are represented with  bases      phi i    on different shapes, the formulation becomes   T F f    T F  sum i a i  phi i M     sum i a i T F  phi i M       Here the core is  T F  phi i M     sum j c  ji   phi j N    T F f     sum j sum i a i c  ji   phi j N        Noted that here the matrix  C  or    c  ji     can fully encode the shape correspondence, in the form of transportation of basis functions between shapes.       If only consider the spectral embeddings    a i      T F a    Ca        If  C  is correct, the point to point map can be...

Computing and Processing Correspondences with Functional Maps | Maks Ovsjanikov et. al

   Spectral Mesh Processing   Bruno L vy   Hao Zhang    Spectral mesh processing   ACM SIGGRAPH 2010 Courses     A gentle introduction    Setting the stage     Laplacian, Laplacian smoothing and its relationship with midpoint smoothing   Spectral transform     Use eigenvectors of the laplacian as the basis to represent coordinates functions  x i , y i  .       From linear algebra, we know that since L is symmetric, it has real eigenvalues and a set of real and orthogonal set of eigenvectors which form a basis. Any vector of size n can be expressed as a linear sum of these basis vectors.       Signal reconstruction and compression with this basis         As well as filtering and smoothing       Connection with discrete Fourier transform           The connection we seek, between DFT and spectral analysis with respect to the Laplace operator, is that the DFT basis functions, the  g k  s, form a set of eigenvectors of the 1D discrete Laplac...

Spectral Mesh Processing | Bruno Lévy & Hao Zhang

   Deep Learning Tuning Playbook   Google Research    GitHub   google research tuning  playbook  A playbook for systematically maximizing the performance of deep learning models.     Guide for starting a new project    Guides on choosing the model architecture, the optimizer, and the batch size.   The goal of tuning batch size is to saturate the GPUs, which can be monitored by         training throughput      examples processed per second          time per step    batch size     training throughput          P.S. When batch size is tuned, most hyper parameters need to be tuned. Among them, the learning rate and the regularization term are the most important.      A scientific approach to improving model performance    Design the experiment       Scientific hyper parameters         The experiment is aimed at explore the effect of the scientific hyper parameters.       Nuisance hyper parameters         To compare different approach, nu...

Deep Learning Tuning Playbook | Google Research

   Deep Work   Cal Newport     Part I  The Idea      Deep Work Is Valuable    3 types who thrived in the new economy  high skilled workers super stars owners.   How to thrive in the new economy  quickly master hard things   produce at elite level.   Deep work   deliberate practice  intense concentration   feedbacks towards productivity.     Reasoning  myelin   attention residue   law of productivity.   Exception  Jack Dorsey s case.      Deep Work Is Rare  Why hard works is rare     Deep work is hard and shallow work is easier.   The  metric black hole   hard to measure different knowledge work behavior s actual impact on the bottom line  outcome     The Internet  technopoly culture that believes whatever the IT revolution provides are revolutionarily good.      Deep Work Is Meaningful    Neurologically, as Winifred Gallagher discovered, our brain construct our worldview based on what we pay attention to. Deep work requires concentration and thus it both focuses your  worldview  to you...

Deep Work | Cal Newport

   Clean Code Handbook  LeetCode 50 Common Interview Questions   LeetCode     Data structure      Array string     Transform   simplify combinatorial iterations         Representation             13 Longest Palindromic Substring             Instead of representing the substring with starting and ending indices, represent it as center index and length is much more efficient, improving runtime from  O n 3   to  O n 2  .        Proxy items             2 Two Sums II   Input Array Is Sorted             If the array is sorted, then  a i     a j   leq a i    a j   leq a i    a j   . Thus  a i    a j   can be used as a proxy of a bunch of other items in the iterations, improving the runtime to  O n  .        Iteration folding             1 Two Sums                 is commutative, thus hal...

Clean Code Handbook: LeetCode 50 Common Interview Questions | LeetCode

   Computer Vision  Algorithms and Applications   Richard Szelisky    Free official electronic version  Computer Vision  Algorithms and Applications, 2nd ed.   P.S. Part name was added by me.      Introduction    Sample Editor Command     Very difficult due to imperfect information.   A brief history of computer vision.     Part I  The Basics      Image formation    Sample Editor Command     Homogeneous coordinates and how to use it to describe 2D 3D transformation compactly.   Photometric image formation     Models of lighting, reflectance, shading, and optics.   Digital camera       Sampling and aliasing when digitizing image.       Demosaicing color filter arrays and describe colors in different color space.       Gamma correction and white balance.       Image compression technique used by JPEG and other image formats.                  Image processing    Operations       Point level operators       Convolutional operators  linear filtering        Neig...

Computer Vision: Algorithms and Applications | Richard Szelisky

   So Good They Can t Ignore You   Cal Newport    So Good They Can t Ignore You   Cal Newport     Rule  1 Don t Follow Your Passion     x  The  Passion  of Steve Jobs    x  Passion Is Rare    x  Passion Is Dangerous     Rule  2 Be So Good They Can t Ignore You     x  The Clarity of the Craftsman    x  The Power of Career Capital    x  The Career Capitalists    x  Becoming a Craftsman  5 steps to start becoming a craftsman who accumulates  career capital  with  deliberate practice       Decide what capital market you re in . I m in academia. To win good positions, the only capital is  good publications . Therefore, this is a  winner take all  marker.    Identify your capital type .  Good publication  is the major capital in the academic market. To gain this kind of capital, critical skills to master include grand vision of the field  e.g. keeping tracking of up to date papers , ability to hunt down important impactful manageable problems,...

So Good They Can't Ignore You | Cal Newport

   How to Become A Straight A Student   Cal Newport    How to Become a Straight A Student   Cal Newport     Part I  Study Basics    Step 1  Manage Your Time in Five Minutes a Day  See  Cal Newport s time management system   Two column todo list.    Step 2  Declare War on Procrastination   Step 3  Choose When, Where, and How Long     Part II  Quizzes and Exams    Step 1  Take Smart Notes   Step 2  Demote Your Assignments   Step 3  Marshal Your Resources   Step 4  Conquer the Material   Step 5  Invest in  Academic Disaster Insurance   Step 6  Provide  A   Answers   The Plan in Action     Part III  Essays and Papers  In part 3, Cal devises a detail scheme to approach thesis essay assignments. I got 3 takeaways that not only could help my academic writing but may also be applicable to research works      Split the mentally demanding tasks into steps.  Intertwining thesis ...

How to Become A Straight-A Student | Cal Newport

   The Time Block Planner   Cal Newport     The Time Block Planner   Cal Newport  Cal Newport s time management system is summarized in this book. Combining the techniques discussed in this tiny pamphlet and his other books podcasts, I compiled this blog post  Cal Newport s time management system.

The Time-Block Planner | Cal Newport

Finding dense correspondences between 3D shapes is a fundamental yet unresolved challenge, especially in real-world environments. These environments present severe challenges, including the lack of time and sufficient samples for training, the prevalence of uncurated extreme-high resolution data with topological distortions, and the need to handle diverse 3D representations. In this paper, we present ATM, a zero-shot framework that requires no correspondence-specific training and robustly addresses these issues at once through an articulate-then-match paradigm. Rather than relying on intrinsic geometric properties, we leverage powerful pretrained vision foundation models and parametric shape priors to estimate parametric shape models from multi-view renderings, and systematically ground these estimations via multi-view geometric consistency. By mapping diverse inputs into a shared canonical parametric space, we inherently establish robust coarse correspondences that bypass topological noise, which are then refined into precise dense mappings via spectral refinement. Operating purely on test-time optimized parametric reconstructions, ATM requires no correspondence training data, is naturally immune to connectivity artifacts, and seamlessly handles diverse 3D modalities, including meshes, point clouds, and 3D Gaussians. Extensive experiments demonstrate that our method achieves strong results on non-isometric benchmarks (average geodesic errors of 2.4-TOPKIDS, 3.8-SMAL), reducing errors by 73% and 37% respectively compared to the baseline URSSM. Furthermore, it exhibits unprecedented robustness on in-the-wild raw scans of up to 200k vertices per shape while maintaining near-constant computation time and consistent superior accuracy.

Articulating then Matching: Zero-Shot Shape Matching for Uncurated Data

Establishing dense correspondence across 3D shapes is crucial for fundamental downstream tasks, including texture transfer, shape interpolation, and robotic manipulation. However, learning these mappings without manual supervision remains a formidable challenge, particularly under severe non-isometric deformations and in inter-class settings where geometric cues are ambiguous. Conventional functional map methods, while elegant, typically struggle in these regimes due to their reliance on isometry. To address this, we present GLASS, a framework that bridges the gap by integrating geometric spectral analysis with rich semantic priors from vision-language foundation models. GLASS introduces three key innovations: (i) a view-consistent strategy that enables robust multi-view visual feature extraction from powerful vision foundation models; (ii) the injection of language embeddings into vertex descriptors via zero-shot 3D segmentation, capturing high-level part semantics; and (iii) a graph-assisted contrastive loss that enforces structural consistency between regions (e.g., source's head'' ↔ target's head'') by leveraging geodesic and topological relationships between regions. This design allows GLASS to learn globally coherent and semantically consistent maps without ground-truth supervision. Extensive experiments demonstrate that GLASS achieves state-of-the-art performance across all regimes, maintaining high accuracy on standard near-isometric tasks while significantly advancing performance in challenging settings. Specifically, it achieves average geodesic errors of 0.21, 4.5, and 5.6 on the inter-class benchmark SNIS and non-isometric benchmarks SMAL and TOPKIDS, reducing errors from URSSM baselines of 0.49, 6.0, and 8.9 by 57%, 25%, and 37%, respectively.

GLASS: Graph and Vision-Language Assisted Semantic Shape Correspondence

Diabetes mellitus is a worldwide epidemic that leads to significant changes in foot shape, deformities, and ulcers. Precise classification of diabetic foot not only helps identify foot abnormalities but also facilitates personalized treatment and preventive measures through the engineering design of foot orthoses. In this study, we propose a novel deep learning method based on DiffusionNet which incorporates a self-attention mechanism and external features to classify the foot types of diabetic patients into six categories by using simple 3D foot images directly. Our approach achieves a high accuracy of 82.9% surpassing existing machine and deep learning methods. The proposed model offers a cost-effective way to analyse foot shapes and facilitate the customization process for both the footwear industry and medical applications.

A novel deep learning approach to classify 3D foot types of diabetic patients

Assessing plantar pressure is crucial for fabricating diabetic insoles and preventing diabetic foot ulcers (DFUs), which are caused by increased plantar pressure. However, the commonly used methods for assessing plantar pressure distribution involve professional sensor-based equipment and expertise, which are costly and time-consuming. Given the qualitative association between ink footprint images and plantar pressure, this study proposes using the footprint images to predict the quantitative values of dynamic plantar pressure in barefoot and 4 different insole conditions (including Nora Lunalastik EVA, Nora Lunalight A fresh, Pe-Lite, and PORON Medical 4708) based on a multilayer perceptron (MLP) neural network model. To provide more precise insole material recommendations for specific foot regions for better plantar pressure distribution, the plantar of the foot is divided into 5 regions: the toes, metatarsal heads, medial midfoot, lateral midfoot, and heel. Patch-based MLP with localization embedding is introduced to learn the correspondence between ink density and plantar pressure information. Ground-truth data collected from 52 diabetes patients is constructed as a dataset named diabetes-footprint-to-pressure and used to train and validate the model. The mean absolute error (MAE) of the models for the barefoot and 4 insole conditions is 5.51% (33.06 kPa), 3.99% (23.94 kPa), 4.85% (29.10 kPa), 4.25% (25.50 kPa), and 3.57% (21.42 kPa) of the sensing range, respectively. Compared to traditional methods for plantar pressure assessment, this approach streamlines the process of acquiring the overall and regional dynamic plantar pressure with barefoot and 4 different insole materials. Clinicians can quickly provide recommendations on the type of insole material for individual patients.

Prediction of Dynamic Plantar Pressure from Insole Intervention for Diabetic Patients Based on Patch-based Multilayer Perceptron with Localization Embedding

This study aims to investigate the biomechanical behaviour and the stiffness impact of the breast internal components during running. To achieve this, a novel nonlinear multi-component dynamic finite element method (FEM) has been established, which uses experimental data obtained via 4D scanning technology and a motion capture system. The data are used to construct a geometric model that comprises the rigid body, layers of soft tissues, skin, pectoralis major muscle, fat, ligaments and glandular tissues. The traditional point-to-point method has a relative mean absolute error of less than 7.92% while the latest surface-to-surface method has an average Euclidean distance (d) of 7.05 mm, validating the simulated results. After simulating the motion of the different components of the breasts, the displacement analysis confirms that when the motion reaches the moment of largest displacement, the displacement of the breast components is proportional to their distance from the chest wall. A biomechanical analysis indicates that the stress sustained by the breast components in ascending order is the glandular tissues, pectoralis major muscle, adipose tissues, and ligaments. The ligaments provide the primary support during motion, followed by the pectoralis major muscle. In addition, specific stress points of the breast components are identified. The stiffness impact experiment indicates that compared with ligaments, the change of glandular tissue stiffness had a slightly more obvious effect on the breast surface. The findings serve as a valuable reference for the medical field and sports bra industry to enhance breast protection during motion.

Construction of multi-component finite element model to predict biomechanical behaviour of breasts during running and quantification of the stiffness impact of internal structure

Understanding the dynamic deformation pattern and biomechanical properties of breasts is crucial in various fields, including designing ergonomic bras and customized prostheses, as well as in clinical practice. Previous studies have recorded and analyzed the dynamic behaviors of the breast surface using 4D scanning, which provides a sequence of 3D meshes during movement with high spatial and temporal resolutions. However, these studies are limited by the lack of robust and automated data processing methods which result in limited data coverage or error-prone analysis results. To address this issue, we identify revealing inter-frame dense correspondence as the core challenge towards conducting reliable and consistent analysis of the 4D scanning data. We proposed a fully-automatic approach named Ulta-dense Motion Capture (UdMC) using Thin-plate Spline (TPS) to augment the sparse landmarks recorded via motion capture (MoCap) as initial dense correspondence and then rectified it with a sophisticated post-alignment scheme. Two downstream tasks are demonstrated to validate its applicability: virtual landmark tracking and deformation intensity analysis. For evaluation, a dynamic 4D human breast anthropometric dataset DynaBreastLite was constructed. The results show that our approach can robustly capture the dynamic deformation characteristics of the breast surfaces, significantly outperforms baselines adapted from previous works in terms of accuracy, consistency, and efficiency. For 10 fps dataset, average error of 0.25 cm on control-landmarks and 0.33 cm on non-control (arbitrary) landmarks were achieved, with 17-70 times faster computation time. Evaluation was also carried out on 60 fps and 120 fps datasets, with consistent and large performance gaining being observed. The proposed method may contribute to advancing research in breast anthropometry, biomechanics, and ergonomics by enabling more accurate tracking of the breast surface deformation patterns and dynamic characteristics.

Ultra-dense Motion Capture: An exploratory full-automatic approach for dense tracking of breast motion in 4D

This thesis presents a comprehensive approach to facilitate breast biomechanics research and ergonomic sports bra design. The study involves three main components: dynamic 4D scanning of the female subject during running, dense tracking of breast deformation, and finite element modeling with material properties fine-tuning using 4D scanning data. The innovative use of 4D scanning technology captures whole-surface information of the human body during dynamic activities, providing high-temporal and spatial resolutions mesh data for analysis. Based on the anthropometric landmarks labelled from the 4D scanning sequence, the overall trajectories and the accumulated regional displacement of the breast soft tissues, as well as the distribution of deformation intensity can be precisely analyzed. Results indicated that the accumulated trajectory lengths of different landmarks range from 50 cm to 80 cm. Of which, the vertical and lateral swinging are the primary movement trends of the breasts during running. The large trajectory differences (48.6%) amongst the landmarks also confirm the highly nonlinear deformation patterns of the breasts during dynamic motion. A robust dense tracking method, the Ultra-dense Motion Capture (UdMC) algorithm, is proposed to capture the dense whole-surface deformation proﬁle of the breasts, advancing the traditional motion capture technology from the sparse landmark level to the dense surface level. Comprehensive evaluation shown that our approach significantly outperforms previous works in accuracy, consistency, and efficiency. With reference to the complete 120 fps dataset, the average errors are found as 0.43cm for the control-landmarks and 0.78cm for the non-control (arbitrary) points. As compared to the traditional approach, the calculation speed of the proposed UdMC algorithm is 40-200 times faster. Lastly, a subject-specific finite element (FE) model is constructed and fine-tuned with the dense deformation proﬁle captured by UdMC, making it capable to align with the realistic breast behavior more reliably. To facilitate efficient determination of the subject-specific Mooney-Rivlin material parameters, the principle parameters inflation scheme was proposed to transform the optimization problem from the 5 dimensional space search to the 2 dimensional space search. This FE breasts model has successfully simulated and predicted the characteristics and response of the breasts when wearing different sports bra with varying design factors, which has significant application value in breasts soft tissue biomechanics research as well as validating and optimizing sports bras prototype designs.

Official guidelines state that suitable physical activity is recommended for patients with diabetes mellitus. However, since walking at a rapid pace could be associated with increased plantar pressure and potential foot pain, the footwear condition is particularly important for optimal foot protection in order to reduce the risk of tissue injury and ulceration of diabetic patients. This study aims to analyze foot deformation and plantar pressure distribution at three different walking speeds (slow, normal, and fast walking) in dynamic situations. The dynamic foot shape of 19 female diabetic patients at three walking speeds is obtained by using a novel 4D foot scanning system. Their plantar pressure distributions at the three walking speeds are also measured by using the Pedar in-shoe system. The pressure changes in the toes, metatarsal heads, medial and lateral midfoot, and heel areas are systematically investigated. Although a faster walking speed shows slightly larger foot measurements than the two other walking speeds, the difference is insignificant. The foot measurement changes at the forefoot and heel areas, such as the toe angles and heel width, are found to increase more readily than the measurements at the midfoot. The mean peak plantar pressure shows a significant increase at a faster walking speed with the exception of the midfoot, especially at the forefoot and heel areas. However, the pressure time integral decreases for all of the foot regions with an increase in walking speed. Suitable offloading devices are essential for diabetic patients, particularly during brisk walking. Design features such as medial arch support, wide toe box, and suitable insole material for specific area of the foot (such as polyurethane for forefoot area and ethylene-vinyl acetate for heel area) are essential for diabetic insole/footwear to provide optimal fit and offloading. The findings contribute to enhancing the understanding of foot shape deformation and plantar pressure changes during dynamic situations, thus facilitating the design of footwear/insoles with optimal fit, wear comfort, and foot protection for diabetic patients.

Analysis of Diabetic Foot Deformation and Plantar Pressure Distribution of Women at Different Walking Speeds

Background: Alterations in the lower limb kinematics and kinetics of diabetic patients have been reported in previous studies. Inappropriate choices of orthopedic insole materials, however, fail to prevent diabetic foot ulcers and modify abnormal gait. The aim of this study was to quantitatively compare the effects of contoured insoles with different materials on the kinematics of and kinetics changes in the diabetic elderly during gait. Methods: There were 21 diabetic patients who participated in this study. Three-dimensional (3D) experimental contoured insoles constructed of soft (i.e., Nora Lunalastik EVA and PORON® Medical 4708) and rigid (i.e., Nora Lunalight A fresh and Pe-Lite) materials with Langer Biomechanics longitudinal PPT® arch pads were adopted. An eight-camera motion capture system (VICON), two force plates, and an insole measurement system—Pedar® with 99 sensors—were utilized to obtain the kinematics and kinetics data. The plug-in lower body gait model landmarks were used for dynamic data acquisition during gait. The corresponding data from five gait cycles were selected and calculated. Results: The range of motions (ROMs) of the ankle joint (p = 0.001) and knee joint (p = 0.044) were significantly influenced when the contoured insoles were worn in comparison to the barefoot condition. The joint moments of the lower limbs with maximum ankle plantarflexion during the loading response and maximum knee and hip flexions were significantly influenced by the use of contoured insoles with different materials in the diabetic elderly. The peak plantar pressure (PPP) of the forefoot (p < 0.001), midfoot (p = 0.009), and rearfoot (p < 0.001) was significantly offloaded by the contoured insoles during the stance phase, whilst the PPP of the rearfoot (p < 0.001) was significantly offloaded during the swing phase. Conclusions: The contoured insoles, especially those constructed with soft materials, significantly offloaded the PPP during gait—hence accommodating certain abnormal gait patterns more effectively compared to going barefoot.

Influence of contoured insoles with different materials on kinematics and kinetics changes in diabetic elderly during Gait

Background: Sports bras are engineer designed to enhance sports performance, which means that they need to provide an excellent fit, and offer adequate support and protection of the breasts to optimize their functionality. To effectively reduce breast motion during different intensity levels of exercise, the materials of sports bras are generally rigid which exert compressive forces onto the soft tissues of the breasts. However, these materials may still restrict air flow and inhibit body heat loss, while the pressure from the bra exerted onto the skin may also increases physiological strain and wear discomfort. This excessively high exerted pressure is known to produce an inhibitory effect on the sweating rate and associated with a significant rise in the axillary and core temperatures. This preliminary study therefore investigates the influence of bra pressure on the upper body temperature and thermal comfort following a short duration of treadmill running. Objective: The purpose of this study is to investigate the effect of increased bra pressure on thermal response following exercise. The findings provide bra designers with insight into bra pressure and related bra design features necessary for optimal wear comfort during physical activities. Methods: A total of 12 young women have participated in this study to don a changeable sports bra that allows adjustment of tension or replacement of the bra components. The skin and body core temperatures as well as heart rate for four bra conditions during treadmill running for 15 minutes at 8 km/h are recorded by using temperature and heart rate sensors. The subjectively perceived thermal and pressure comfort are evaluated by using a visual analog scale with ratings of 1 to 10.Results: Following exercise, there is no change in core temperature for all of the bra conditions studied. Even though the body core temperature may increase due to the higher rate of heat production with muscular work done during treadmill running, the increase in heat dissipation tends to balance the increase in rate of metabolic heat production to maintain a stable core temperature. After a short duration of treadmill running, the change in skin temperature ranges from 0.22oC to 3.56oC amongst the 4 bra conditions. The shoulder strap area shows a slight change in skin temperature during exercise, and the participants are particularly sensitive to the increased pressure in this area, thus adversely affecting their ratings of the thermal and pressure comfort. Conclusion: In this study, the increased bra pressure does not show significant change in core temperature and heart rate during short duration exercise. Even though the results are not statistically significant, the shoulder strap pressure is found to be related to the changes in skin temperature and subjective ratings of thermal and pressure comfort.

Sports bra pressure: effect on core body temperature and comfort sensation

In view of the problems existing in biomedical engineering education, such as the lack of pertinence, disconnection from the needs of the industry and lack of a perfect teaching experimental platform and data system, this paper puts forward a biomedical engineering experimental teaching system for the demand of the industry. The whole process with fine refinement and high efficiency for the biomedical engineering experimental teaching is provided by constructing the experimental teaching platform with the “Three-in-one” integration of the experimental equipment, experimental teaching materials and video resources. The system has the characteristics of modularization, and standardization, and combines closely with the leading-edge concepts of the industry, teaching and research practice, achieving good teaching practice effect and positive feedback.

Construction of experimental teaching system of biomedical engineering for demand of industry

A reference management plugin built entirely on Obsidian, with contextual, flexible, Markdown-powered literature notes

BibTeX Scholar

Scrape BibTeX & abstracts from top AI conferences

Top AI conferences scraping

A simple CV template with customized functionalities

 This is a simple CV template with customized functionalities.   img     How to use this template      Import the template package  First add the package  myCV  to the preamble        usepackage myCV           Meta information  Then add the following meta information, e.g.        name A Cool Cat   email coolcat mail.com   profile profile.jpg   setProfileWidth 3.5cm       You can add links bar under your name by adding        addlink      https   scholar.google.com citations user N2 7ArsAAAAJ hl en       Google Scholar  includegraphics height 0.3cm  icon google scholar.png    addlink      https   orcid.org 0000 0001 7843 1925       ORCID  includegraphics height 0.3cm  icon orcid.png    addlink      https   github.com liu qilong       GitHub  includegraphics height 0.3cm  icon github.png    addlink      https   qilong liu.vercel.app       Homepage  includegraphics height 0.3cm  icon blog.png     ...      Noted that the...

CV template

My customized Obsidian Dataview templates, including TagLens, DailyLens, and PaperThread

   My Obsidian setup  I use Obsidian as my major note taking tool. This repository stores my customized Dataview templates including     PaperThread  Visualize papers  relations as threads   TagLens  Show related notes in different types   DailyLens  Benchmark my time spend, task completions, and other acitvities     PaperThread  Visualize papers  relations as threads    img  Generate  bibtex  of all papers for   LaTeX     img  Retrieve a paper s meta data via  doi     img     TagLens  Show related notes in different types    img   P.S. For a conta...

obsidian-setup

Distribute FE simulation tasks across multiple computers via SSH

   FEcluster  Distribute FE simulation tasks across multiple computers via SSH.   img

FEcluster

Data analysis framework for pedar plantar pressure sensor

 Documentation Status  Data analysis framework for pedar plantar pressure sensor.   heatmap   Fig. An example averaged foot peak pressure distribution heatmap.      Documentation  The documentation web pages can be found in  docs build html  . Please open  index.html  to enter the documentation which provides comprehensive descriptions and working examples for each class and function we provided.  Other than the documentation,  examples   folder provides some example analysis code for quick start.      Note  The documentation is generated with Sphinx. If you are not familiar with it, I would recommend two tutorials for quick start      A  How to  Guide for Sphinx   ReadTheDocs   sglvladi  https   sphinx rtd tutori...

pedarProbe

A collection of the self-customized Mobjects for generating math animation with Manim

 A collection of the self customised Mobjects for generating math animation with Manim.      Example Gallery    Transformer     20221001 multi head attention.py      Transformer   Convolutional Neural Network     20211016 cnn.py      CNN   Convolution operation     20211016 cnn.py      Convolution   Single fully connected layer      20211016 cnn.py  https   github.com liu q...

mobjTOB

Toolkit for 4D (3D + T) data visualisation, operation, and dynamic estimation

    mesh4d      Introduction  This package is developed for the data processing of the 3dMD 4D scanning system. Compared with traditional motion capture systems, such as Vicon     Vicon motion capture system can provide robust   accurate key points tracking based on physical marker points attached to the human body. But it suffers from the lack of continuous surface deformation information.    3dMD 4D scanning system can record continuous surface deformation information. But it doesn t provide key point tracking functionality and it s challenging to track the key points via the Computer Vision approach, even with the state of the art methods in academia  Min Z 2021 .    Min Z 2021   Min, Z., Liu, J., Liu, L.,   Meng, M. Q. H.  2021 . Generalized coherent point drift with multi variate gaussian distribution and Watson distribution. IEEE Robotics and Automation Letters, 6 4 , 6749 6756. https   doi.org 10.1109 lra.2021.3093011  To facilitate dynamic shape analysis research, we deem it an...

mesh4d

Customized beamer templates for colleges/institutes

Hello 👋