- Introduction to Python
- Getting started with Python and the IPython notebook
- Functions are first class objects
- Data science is OSEMN
- Working with text
- Preprocessing text data
- Working with structured data
- Using SQLite3
- Using HDF5
- Using numpy
- Using Pandas
- Computational problems in statistics
- Computer numbers and mathematics
- Algorithmic complexity
- Linear Algebra and Linear Systems
- Linear Algebra and Matrix Decompositions
- Change of Basis
- Optimization and Non-linear Methods
- Practical Optimizatio Routines
- Finding roots
- Optimization Primer
- Using scipy.optimize
- Gradient deescent
- Newton’s method and variants
- Constrained optimization
- Curve fitting
- Finding paraemeters for ODE models
- Optimization of graph node placement
- Optimization of standard statistical models
- Fitting ODEs with the Levenberg–Marquardt algorithm
- 1D example
- 2D example
- Algorithms for Optimization and Root Finding for Multivariate Problems
- Expectation Maximizatio (EM) Algorithm
- Monte Carlo Methods
- Resampling methods
- Resampling
- Simulations
- Setting the random seed
- Sampling with and without replacement
- Calculation of Cook’s distance
- Permutation resampling
- Design of simulation experiments
- Example: Simulations to estimate power
- Check with R
- Estimating the CDF
- Estimating the PDF
- Kernel density estimation
- Multivariate kerndel density estimation
- Markov Chain Monte Carlo (MCMC)
- Using PyMC2
- Using PyMC3
- Using PyStan
- C Crash Course
- Code Optimization
- Using C code in Python
- Using functions from various compiled languages in Python
- Julia and Python
- Converting Python Code to C for speed
- Optimization bake-off
- Writing Parallel Code
- Massively parallel programming with GPUs
- Writing CUDA in C
- Distributed computing for Big Data
- Hadoop MapReduce on AWS EMR with mrjob
- Spark on a local mahcine using 4 nodes
- Modules and Packaging
- Tour of the Jupyter (IPython3) notebook
- Polyglot programming
- What you should know and learn more about
- Wrapping R libraries with Rpy
文章来源于网络收集而来，版权归原创者所有，如有侵权请及时联系！
Why and when does distributed computing matter?
- data growing exponentially
- Whole genome sequencing 100-200 GB per BAM file
- NYSE 1 TB of trade data per day
- Large Hadron Collider 15 PB per year (1 PB = 1,000 TB)
- storage \(\gg\) access speeds
- how long does it take to read or write a 1 TB disk?
- parallel reads can result in large speed ups
- how long does it take to read or write a 1 TB disk?
- most relevant for big data
- 1 billion rows
- \(\gg\) 128 MB (default block size for Hadoop)
- other solutions may be more suitable
- shared memory parallel system
- Relational databases (seek time is bottleneck)
- Grid computing for compute-intensive jobs where netwrok bandwidth is not bottleneck (HPC, MPI)
- Volunteer computing (compute time \(\gg\) data transfer time)
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论