sandc23102015.fsx

/// LinearStudent: a first tutorial on programming the Simplex method in F#
/// F# is a functional-first programming language developed in the Cambridge Microsoft research lab
/// under the direction of a group of well-qualified researchers headed by Simon Peyton-Jones

let A2FM = "good class!"

// glorified calculator
2.*asin(1.0)
let pi = 3.141592654
sin(pi/2.0)

[|sin(3.*pi/2.0); cos(2.*pi/3.0); sin(2.*pi/3.); sqrt(3.)/2.|]

(fun y -> 2*y) (3+2)

// Create a matrix of floating point numbers
let matrix = [| [| 0.; 1.; 2.; 3. |]; [|4.0; 5.0; 6.0; 7.0|]; [|8.0; 9.0; 10.0; 11.0|]|]

// float converts an integer to a float
// and Array.map float r converts all the entries of the array r to floats

matrix.[0].[1]
// this takes the 2nd element of the 1st row of matrix.
// in F# arrays are indexed from 0 rather than 1.  Cope!

// pivotDivideMax: ensures the maximum entry in an array is 1
let inline pivotDivideMax l = Array.map (fun x -> x/(Array.max l)) l

// Array.map f builds a new array whose elements have all had function f applied to them
Array.map (fun x -> 2.*x ) matrix.[0]

matrix.[0]
// The old array is still there. This is functional (first) programming

// A better but more generic version without requiring the "inline" directive
// pivotDivideMaxHigher : div:('a -> 'a -> 'b) -> l:'a [] -> 'b []
let pivotDivideMaxHigher div l = Array.map (fun x -> div x (Array.max l)) l

// Test better version
pivotDivideMaxHigher (fun x y -> x/y) matrix.[1]

// F# uses a special syntax x <- v to change the value of mutable variables
matrix.[0] <- pivotDivideMaxHigher (fun x y -> x/y) matrix.[0]

// The entire first row of matrix has now been normalised
matrix

// we could -- with a bit of jiggery pokery -- define a generic division function ...
let inline myDiv (x:^a) (y:^a) = x/y
// test myDiv
myDiv 3 4
myDiv 3.0 4.0

// .... and use it to abbreviate pivot division actions
pivotDivideMaxHigher myDiv matrix.[1]
let inline pivotDivideMax2 l = pivotDivideMaxHigher myDiv l

matrix.[1] <- pivotDivideMax2 matrix.[1]
// The second row of matrix has been normalised as well
matrix

// rowOp applies a generic row operation to 2 rows
// rowOp : ('a -> 'b -> 'c) -> 'a [] -> 'b [] -> 'c []
let rowOp op r1 r2 = Array.map2 op r1 r2

// Compute r1 - 2 r2 as a new row
rowOp (fun x y -> x - 2.0*y) matrix.[0] matrix.[1]

// Note that rowOp works on any array, so columns work as well

// let badDiv x y = x/y // this does not have a general type
// badDiv 3 4 // this is ok
// badDiv 3.0 4.0 // but this is not ok

/// How to locate the _position_ of an element in an Array
(fun x -> x = 1.0) 2.0 // false
(fun x -> x = 1.0) 1.0 // true

let unit = Array.findIndex (fun x -> x = 1.0) matrix.[0] // find the (first) element equal to 1.0
matrix.[0].[unit]

/// Exercise: now find the position of the _maximum_ element in the 3rd row of matrix