geo-nd 0.6.4

Traits and types particularly for 2D and 3D geometry with implementations for [float] and optionally SIMD
Documentation
//a Imports
use crate::vector_op as vector;
use crate::{Float, Num};

//a Constructors
//fp zero
/// Crate a new matrix which is all zeros
pub fn zero<V: Num, const RC: usize>() -> [V; RC] {
    [V::zero(); RC]
}

//mp set_zero
/// Set the matrix to have all elements of zero
pub fn set_zero<V: Num>(m: &mut [V]) {
    vector::set_zero(m)
}

//fp is_zero
/// Return true if the matrix is all zeros
pub fn is_zero<V: Num>(m: &[V]) -> bool {
    vector::is_zero(m)
}

//cp scale
/// Consume the vector and return a new vector that is the original
/// scaled in each coordinate a single scaling factor
pub fn scale<V: Num, const RC: usize>(m: [V; RC], s: V) -> [V; RC] {
    vector::scale(m, s)
}

//cp reduce
/// Consume the vector and return a new vector that is the original
/// reduces in scale in each coordinate by a single scaling factor
pub fn reduce<V: Num, const RC: usize>(m: [V; RC], s: V) -> [V; RC] {
    vector::reduce(m, s)
}

//cp add
/// Consume the vector, and return a new vector that is the sum of
/// this and a borrowed other vector scaled
pub fn add<V: Num, const RC: usize>(m: [V; RC], other: &[V; RC], scale: V) -> [V; RC] {
    vector::add(m, other, scale)
}

//cp sub
/// Consume the vector, and return a new vector that is the difference of
/// this and a borrowed other vector scaled
pub fn sub<V: Num, const RC: usize>(m: [V; RC], other: &[V; RC], scale: V) -> [V; RC] {
    vector::sub(m, other, scale)
}

//cp absmax
/// Consume the vector, and return a new vector that is the sum of
/// this and a borrowed other vector scaled
pub fn absmax<V: Float>(m: &[V]) -> V {
    m.iter().fold(V::zero(), |acc, c| V::max(acc, V::abs(*c)))
}

//cp normalize
/// Update the new matrix with every element e scaled so that max(abs(e)) is 1.
pub fn normalize<V: Float, const RC: usize, const C: usize>(mut m: [V; RC]) -> [V; RC] {
    let l = absmax::<V>(&m);
    if l < V::epsilon() {
        set_zero::<V>(&mut m);
        m
    } else {
        reduce::<V, RC>(m, l)
    }
}

//cp transpose
/// Transpose the matrix
///
/// The matrices are row-major, so successive entries of m are adjacent columns
///
/// The output matrix has adjacent entries that are therefore adjacent rows in m
pub fn transpose<V: Num, const RC: usize, const R: usize, const C: usize>(m: [V; RC]) -> [V; RC] {
    assert_eq!(RC, R * C);
    let mut v = zero::<V, RC>();
    for r in 0..R {
        for c in 0..C {
            v[r + c * R] = m[c + r * C];
        }
    }
    v
}

//mp multiply
/// Multiply two matrices
///
/// The first matrix is R by X, the second ix X by C, so the result is
/// R by C
#[track_caller]
pub fn multiply<
    V: Num,
    const RX: usize, // Total elements in first matrix
    const XC: usize, // Total elements in second matrix
    const RC: usize, // Total elements in result
    const R: usize,  // Number of rows in first matrix
    const X: usize,  // Number of cols in first matrix, rows in second matrix
    const C: usize,  // Number of cols in second matrix
>(
    a: &[V; RX],
    b: &[V; XC],
) -> [V; RC] {
    assert_eq!(RX, R * X, "RX must equal R*X");
    assert_eq!(RC, R * C, "RC must equal R*C");
    assert_eq!(XC, X * C, "XC must equal X*C");
    let mut m = [V::zero(); RC];
    for r in 0..R {
        for c in 0..C {
            let mut v = V::zero();
            for x in 0..X {
                v = v + a[r * X + x] * b[x * C + c];
            }
            m[r * C + c] = v;
        }
    }
    m
}

//mp tranpose_dyn
/// Transpose a matrix
pub fn transpose_dyn<V: Copy>(
    r: usize,
    c: usize,
    m: &[V],       // r by c
    m_t: &mut [V], // c by r
) {
    assert!(m.len() == r * c);
    assert!(m_t.len() == r * c);
    for ri in 0..r {
        for ci in 0..c {
            m_t[ci * r + ri] = m[ri * c + ci];
        }
    }
}

//mp multiply_dyn
/// Multiply two matrices, dynamically sized
///
/// Until generic const expressions work cleanly this is a workaround
pub fn multiply_dyn<V: Num>(
    r: usize,
    x: usize,
    c: usize,
    a: &[V],     // r by x
    b: &[V],     // x by c
    m: &mut [V], // r by c
) {
    assert!(
        a.len() >= r * x,
        "Expected a (length {}) to be r*x = {} * {}",
        a.len(),
        r,
        x
    );
    assert!(
        b.len() >= x * c,
        "Expected a (length {}) to be r*x = {} * {}",
        a.len(),
        r,
        x
    );
    assert!(
        m.len() >= r * c,
        "Expected a (length {}) to be r*x = {} * {}",
        a.len(),
        r,
        x
    );
    for ri in 0..r {
        for ci in 0..c {
            let mut v = V::zero();
            for xi in 0..x {
                v = v + a[ri * x + xi] * b[xi * c + ci];
            }
            m[ri * c + ci] = v;
        }
    }
}

//mp transform_vec
/// Transform a vector
pub fn transform_vec<V: Num, const RD: usize, const R: usize, const D: usize>(
    m: &[V; RD],
    v: &[V; D],
) -> [V; R] {
    multiply::<V, RD, D, R, R, D, 1>(m, v)
}

//a Formatting
//mp fmt - format a `Matrix` for display
/// Format the matrix for display
///
/// This can be used by the Display or Debug traits of a type which
/// contains an array. It is not possible to use this method directly
/// without a `Formatter` instance.
///
/// To display an array `m` as a matrix use `&MatrixType::new(&m)` (see [MatrixType])
///
/// # Examples
///
/// ```
/// use geo_nd::matrix;
/// struct Mat { c : [f32;6] };
/// impl std::fmt::Display for Mat {
///   fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result { matrix::fmt::<f32,2>(f, &self.c) }
/// }
/// assert_eq!( format!("{}", &Mat{c:[0., 1., 2., 3., 4., 5.]} ), "[0,1 2,3 4,5]" );
/// ```
pub fn fmt<V: Num, const C: usize>(f: &mut std::fmt::Formatter, v: &[V]) -> std::fmt::Result {
    let mut c = 0;
    for (i, value) in v.iter().enumerate() {
        if i == 0 {
            write!(f, "[{value}")?;
        } else if c == 0 {
            write!(f, " {value}")?;
        } else {
            write!(f, ",{value}")?;
        }
        c += 1;
        if c == C {
            c = 0;
        }
    }
    write!(f, "]")
}

//a MatrixType, used in formatting
/// A structure that supports the Debug and Display traits, borrowing
/// a matrix; this permits a relatively simple format of a matrix
/// through using the `[new]` constructor of the type
///
/// # Example
///
/// ```
/// use geo_nd::matrix::{identity2, MatrixType};
/// assert_eq!( format!("{}", MatrixType::<f32,4,2>::new(&identity2())), "[1,0 0,1]" );
/// ```
///
#[derive(Debug)]
pub struct MatrixType<'a, V: Num, const RC: usize, const C: usize> {
    /// Contents of the matrix
    m: &'a [V; RC],
}

//ip MatrixType
impl<'a, V: Num, const RC: usize, const C: usize> MatrixType<'a, V, RC, C> {
    //fp new
    /// Create a new MatrixType by borrowing an array of Num; this may then be formatted using its Display trait
    pub fn new(m: &'a [V; RC]) -> Self {
        Self { m }
    }
}

//ip Display for MatrixType
impl<'a, V: Num, const RC: usize, const C: usize> std::fmt::Display for MatrixType<'a, V, RC, C> {
    //
    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
        fmt::<V, C>(f, self.m)
    }
}