Lecture #11: Numpy for Numerical Applications and Data Model/Interface Design

Lecture Objectives

create and manipulate numpy arrays for scientific/mathematical applicatinos
use a variety of containers to design the data structures for a software project

Chapter 9, pages 201-227

What is numpy?
- mostly about arrays that can be used mathematically
- a lot of resemblance to Matlab
Importing numpy
- common convention: import numpy as np
Common ndarray methods
- initialization: arange, zeros, ones, empty, linspace, logspace
- attributes (see table 9-1), esp:
  - ndim, shape, size
  - dtype
- slicing
- arithmetic & broadcasting
  - mathematical operations are elementwise by default
  - compatible arrays & broadcasting
  - experiment - formulate hypotheses and test
Structured Arrays
- tables with named columns and varying data types
- define the structure of a row: names and types
- access slices by rows and column names
Designing data models and interfaces
1. Many reason to break project into distinct pieces with well-defined interfaces
  - division of labor
  - separation of concerns
  - reviewability
2. all components must share the same data model
  - how will data be shared among components
  - balance between flexibility and simplicity
  - depends heavily on features of language
    - object-oriented
    - advanced containers
3. define interfaces
  - input, output, behavior