Computation

Computation#

Computers essentially do 2 things:

  • Store information

  • Operate on this information

Computers are very good at repeating the same task over and over again.

Our goal is to learn how to write programs to automate operations on files (via the Unix shell) and manage data / do numerical calculations (via C++).

Limitations#

Computers have a finite amount of memory and we want our answers in a reasonable amount of time. This means that we will need to make approximations in a lot of our work—a great example is storing real numbers. You cannot store the infinite real numbers in [0, 1] with a finite amount of memory. So computers will approximate how real numbers are stored—we will learn about this format shortly.

Languages#

In physics and astronomy, a wide range of programming languages are commonly used. The job of a programming language is to bridge how humans think and how computers think. We write a program in a human-friendly way (plain text with English keywords expressing logic) and then a tool called a compiler translates this into instructions that the computer understands.

High-level languages have extensive libraries that implement commonly-used features that you can rely on when writing your code, saving you time.

There are many ways we can categorize programming languages (and there are exceptions to most of what I write here). Some examples:

  • Compiled vs. interpreted

    • A compiled language is passed through a program called a compiler that translates the human-readable source code into instructions that are native to the CPU in your computer.

      Examples include: C, C++, and Fortran

    • An interpreted language is processed a runtime, by running the program through an interpreter.

      Examples include: python, ruby, mathlab

Generally speaking, compiled languages generate faster code, but interpreted language can be easier to program in.

  • Dynamic vs. static typing

    When we work with data, we are going to store it in variables with a descriptive name, and then refer to that name throughout the program to access the data.

    • Dynamically typed languages do not require us to specify up front what type of data we are going to store in a variable. It is inferred at runtime.

      This can give a lot of flexibility to codes.

      python is an example.

    • Statically typed languages require you to specify up front what type of data will be stored in a variable.

      This can allow the compiler to optimize the code.

Note

You can implement any algorithm in any of these languages, but some make it easier than others.

Why C++?#

This course aims to teach a compiled language, and traditionally, Fortran and/or C++ are taught.

We will focus only on C++.

C++ is:

  • a modern compiled language

  • updated with new features frequently (currently a 3-year cycle)

  • used extensively in physics (in particular in high-energy physics and in high-performance computing in astrophysics)

  • versatile

  • widely used outside of academic research environments—this means that you can take your C++ skills to industry jobs if desired

Important

We will focus on the C++20 standard.

Contents