Lecture2 Note

Programs Implement Functions

  Definition: Function is a type of procedure or routine.

Example of function:

Inputs and Outputs of Programs

Clear what are the inputs of the program and what are the outputs of the program.

Program Parameters

  • Parameters are the special inputs, which are the necessary context for the program to run. (But not every program needs prameters)
  • Parameters can be chose by youself or set the program to choose a random parameter or optimize the parameter based on some other functions.

E.g., number of clusters in clustering:

Treating Programs as "Black Boxes"

  • Don’t have to understand complex programming in order to use software
  • This is why we often refer to sofrware as a “black box”
  • Only need to understand inputs and outputs and the program’s function

Compilers

Translate Advanced programming language into machine language.

Turing-complete

  1. Turing machines – mathematical framework to understand computing machines

    • They can be used to study how an algorithm works

  2. A programming language is turing-complete if it is possible to implement a turning machine in that languate

    • They must include special instructions such as loops, iterations and conditional statements.

  3. All programminglanguages are Turing-complete and there they are computationally equivalent

    • The differences among them are in the ease of use for implementing specific functions.

Algorithms

An algorithm is as mechanical procedure that describes how to carry out a computation on some data (the logic)

Workflow & Computational Workflows

A Workflow is a Composition of Functions

  1. Workflow is represented as a graph of connected nodes
    • Nodes represent programs and data
    • Links represent how data flows from program to program
  2. Computatinal workflows are compositions of programs
    • No user interaction during execution
    • No cycles/loops or iterations allowed

Workflow Example:

Document Similarity

Repeatability & Reproducibility & Replication

  1. Repeatability : the same lab can re-run an analysis and get the same results

    • Not so obvious due to new versions of software.
    • Can collect new data, which is called In-lab Replication.

  2. Reproducibility : another lab can run analysis with the same/original data and get the same results

    • Original lab must share their data directly or make it available for download.

  3. Replication: another lab can re-run study and get the same results

    • Original lab has to document in detail what was done.
    • Different lab uses the same methods and new dataset to try to “replicate” the finding.
OG User New User
OG Data Repeatability Reproducibility
New Data In-lab Replication Replication