In this article we will see how to implement the multiplication of quaternions in Rust and NodeJS. I encourage you to try solve this problem before seeing solutions. Below I attaching link to this exercise:

Coding Games and Programming Challenges to Code Better

CodinGame is a challenge-based training platform for programmers where you can play with the hottest programming topics. Solve games, code AI bots, learn from your peers, have fun.

CodinGame

The quaternions belong to a number system that extends the complex numbers. A quaternion is defined by the sum of scalar multiples of the constants i,j,k and 1.
More information is available at:

Quaternion -- from Wolfram MathWorld

The quaternions are members of a noncommutative division algebra first invented by William Rowan Hamilton. The idea for quaternions occurred to him while he was walking along the Royal Canal on his way to a meeting of the Irish Academy, and Hamilton was so pleased with his discovery that he scratche…

from Wolfram MathWorld

Consider the following properties:
jk = i
ki = j
ij = k
i² = j² = k² = -1

These properties also imply that:
kj = -i
ik = -j
ji = -k

The order of multiplication is important.

Your program must output the result of the product of a number of bracketed simplified quaternions.

Pay attention to the formatting
The coefficient is appended to the left of the constant.
If a coefficient is 1 or -1, don't include the 1 symbol.
If a coefficient or scalar term is 0, don't include it.
The terms must be displayed in order: ai + bj + ck + d.

Example Multiplication
(2i+2j)(j+1) = (2ij+2i+2j² +2j) = (2k+2i-2+2j) = (2i+2j+2k-2)

Input:

Line 1: The expression expr to evaluate. This will always be the product of simplified bracketed expressions.

Output: A single line containing the simplified result of the product expression. No brackets are required.

Constraints: All coefficients in any part of evaluation will be less than 10^9
The input contains no more than 10 simplified bracketed expressions

Example

Input

(i+j)(k)

Output

i-j

Solution

I decided to present only most important parts here. Full solution can be found in repository:

GitHub - gustawdaniel/codingame-quaternion-multiplication

Contribute to gustawdaniel/codingame-quaternion-multiplication development by creating an account on GitHub.

GitHubgustawdaniel

We can divide our problem to three steps:

• parsing input to Quaternion structure
• multiplication of Quaternions
• formatting Quaternion again to string

These high-level operations can be implemented in NodeJS

import {Quaternion} from "./lib";

process.stdin.on('data', (buff) => {
    const line = buff.toString();
    const qs = Quaternion.parse(line);
    process.stdout.write(qs.reduce((p, n) => p.multiply(n)).format());
})

and in Rust

fn main() {
    let mut input_line = String::new();
    io::stdin().read_line(&mut input_line).unwrap();
    let expr = input_line.trim_matches('\n').to_string();

    let qs = Quaternion::parse(&expr);
    let out = qs.into_iter().reduce(|p, n| p.multiply(n)).unwrap();
    println!("{}", out);
}

You can see that this code is really similar but in both cases we have to implement Struct/Class named Quaternion. Now we will go through three steps mentioned before using TDD. Tests are natively supported in rust, but in NodeJS I decided to use jest as testing framework.

Parsing input to Quaternion structure

Our input

(i+j)(k)

should be treated as array of quaternions - separated by brackets. In any brackets we have array of coefficients. So we can divide our parsing to 4 parts:

• spliting by brackets
• splitings any bracket to coefficients
• creating Quaternions from arrays of coefficients
• extracing number from coefficient

In NodeJS we can start from two tests. First for simple cases:

    it('simple parse', () => {
        const qs = Quaternion.parse('(i+j)');
        expect(qs.length).toEqual(1);
        expect(qs[0].r).toEqual(0);
        expect(qs[0].i).toEqual(1);
        expect(qs[0].j).toEqual(1);
        expect(qs[0].k).toEqual(0);
    });

Second for more advanced coefficients:

    it('complex parse', () => {
        const qs = Quaternion.parse('(9+i-j)(k-8.4j)');
        expect(qs.length).toEqual(2);
        expect(qs[0].r).toEqual(9);
        expect(qs[0].i).toEqual(1);
        expect(qs[0].j).toEqual(-1);
        expect(qs[0].k).toEqual(0);

        expect(qs[1].r).toEqual(0);
        expect(qs[1].i).toEqual(0);
        expect(qs[1].j).toEqual(-8.4);
        expect(qs[1].k).toEqual(1);
    });

The same tests in rust can be written as

#[cfg(test)]
mod tests {
    use crate::{Quaternion};


    #[test]
    fn simple_parse() {
        let qs = Quaternion::parse("(i+j)");
        assert_eq!(qs.len(), 1);
        assert_eq!(qs[0].r, 0f64);
        assert_eq!(qs[0].i, 1f64);
        assert_eq!(qs[0].j, 1f64);
        assert_eq!(qs[0].k, 0f64);
    }

    #[test]
    fn complex_parse() {
        let qs = Quaternion::parse("(9+i-j)(k-8.4j)");
        assert_eq!(qs.len(), 2);
        assert_eq!(qs[0].r, 9f64);
        assert_eq!(qs[0].i, 1f64);
        assert_eq!(qs[0].j, -1f64);
        assert_eq!(qs[0].k, 0f64);

        assert_eq!(qs[1].r, 0f64);
        assert_eq!(qs[1].i, 0f64);
        assert_eq!(qs[1].j, -8.4f64);
        assert_eq!(qs[1].k, 1f64);
    }
}

Our base Quaternion class will have 4 properties. In NodeJS:

export class Quaternion {
    r: number = 0;
    i: number = 0;
    j: number = 0;
    k: number = 0;
}

where r means real part that inherits arithmetic from real numbers. In rust we are using struct keyword instead of class

#[derive(Debug)]
struct Quaternion {
    r: f64,
    i: f64,
    j: f64,
    k: f64,
}

Splitting string using regular expressions

To split input and provide arrays of coefficients to Quaternion constructors we can write methods in NodeJS:

    static parse(input: string): Quaternion[] {
        const qs = (input.match(/\(.*?\)/g) ?? []).map(
            (e: string) => (e
                    .replace('(', '')
                    .replace(')', '')
                    .match(/[-+]?[\d.]*[ijk]?/g) ?? []
            ).filter(v => v).map(
                v => v.replace(/^\+/, '')
            )
        );
        return qs.map((q) => new Quaternion(q));
    }

and Rust

impl Quaternion {
    fn parse(input: &str) -> Vec<Quaternion> {
        let re = Regex::new(r"\((.*?)\)").expect("can't create regex");

        let qs = re.captures_iter(input).filter_map(|cap| Some(cap.get(1)?.as_str()))
            .map(|m| m.to_string()).collect::<Vec<_>>();

        let re = Regex::new(r"\+?(-?[\d.]*[ijk]?)").expect("can't create regex");

        let res = qs.iter().map(|q| {
            let args = re.captures_iter(&q).filter_map(|cap| Some(cap.get(1)?.as_str()))
                .map(|m| m.to_string()).collect::<Vec<_>>();
            Quaternion::new(args)
        });

        res.collect::<Vec<_>>()
    }
}

Generally there are the same regex expressions but rust requires here external library called regex. Additionally rust checks correctness of regex expressions and ensure handling errors in them, that can be skipped by default in node js code. Generally I feel that node js approach to regex is more clean and readable.

Now we have the following problem. Our coefficients can contains numbers, numbers with names of component like: i, j or k, or even lonely letters like i what means 1i. There are also possible signs like -k.

We need code that will extract numbers from them. Lets name i, j or k as type and full coefficient string as input. Then extracting number can be considered as:

• removing type from input
• if rest not end with digit then add 1 on the end ( example is - )
• finally parse it as float

Implementation in node js

    static getCoefficient(type: string, input: string): number {
        const coefficient = input.replace(type, '');
        return Number.parseFloat(/\d$/.test(coefficient) ? coefficient : coefficient + '1')
    }

and analogical in rust

    fn get_coefficient(t: &str, input: String) -> f64 {
        let c = input.replace(t, "");
        if Regex::new(r"\d$").expect("ff").is_match(&c) {
            c.parse::<f64>().unwrap()
        } else {
            (c + "1").parse::<f64>().unwrap()
        }
    }

In rust generally handling conversions between types require more characters but is more reliable. In this case writing unwrap we are enforced to think about possible ways of handle problems with parsing.

Now we can present constructors. In them we will pass array of strings with coefficients like 8, -9k, or i. In node js:

    constructor(args: Array<string>) {
        for (let arg of args) {
            if (arg.includes('i')) {
                this.i = Quaternion.getCoefficient('i', arg);
            } else if (arg.includes('j')) {
                this.j = Quaternion.getCoefficient('j', arg);
            } else if (arg.includes('k')) {
                this.k = Quaternion.getCoefficient('k', arg);
            } else {
                this.r = Number.parseFloat(arg);
            }
        }
    }

or in rust:

    fn new(args: Vec<String>) -> Quaternion {
        let mut q = Quaternion {
            i: 0f64,
            j: 0f64,
            k: 0f64,
            r: 0f64,
        };

        for arg in args {
            if arg.contains("i") {
                q.i = Quaternion::get_coefficient("i", arg)
            } else if arg.contains("j") {
                q.j = Quaternion::get_coefficient("j", arg)
            } else if arg.contains("k") {
                q.k = Quaternion::get_coefficient("k", arg)
            } else {
                q.r = arg.parse::<f64>().unwrap()
            }
        }

        q
    }

Multiplication of Quaternions

Multiplication of two quaternions is similar to multiplication of polynomials. Firstly we muliply any components pairs each other. Then we can group them by component type and finally add coefficients. Difference is that in polynomial multiplication we always adding powers, but in Quaternions we using noncommutative division algebra presented on table:

So basically we can divide our problem to:

• multiplication of base elements
• multiplication of linear combinations of base elements

Multiplication of base elements

I rewrote all possible cases in test file in NodeJS

    it('multiply base', () => {
        expect(Quaternion.multiplyBase('r', 'r')).toEqual({c: 1, d: 'r'});
        expect(Quaternion.multiplyBase('r', 'i')).toEqual({c: 1, d: 'i'});
        expect(Quaternion.multiplyBase('r', 'j')).toEqual({c: 1, d: 'j'});
        expect(Quaternion.multiplyBase('r', 'k')).toEqual({c: 1, d: 'k'});

        expect(Quaternion.multiplyBase('i', 'r')).toEqual({c: 1, d: 'i'});
        expect(Quaternion.multiplyBase('i', 'i')).toEqual({c: -1, d: 'r'});
        expect(Quaternion.multiplyBase('i', 'j')).toEqual({c: 1, d: 'k'});
        expect(Quaternion.multiplyBase('i', 'k')).toEqual({c: -1, d: 'j'});

        expect(Quaternion.multiplyBase('j', 'r')).toEqual({c: 1, d: 'j'});
        expect(Quaternion.multiplyBase('j', 'i')).toEqual({c: -1, d: 'k'});
        expect(Quaternion.multiplyBase('j', 'j')).toEqual({c: -1, d: 'r'});
        expect(Quaternion.multiplyBase('j', 'k')).toEqual({c: 1, d: 'i'});

        expect(Quaternion.multiplyBase('k', 'r')).toEqual({c: 1, d: 'k'});
        expect(Quaternion.multiplyBase('k', 'i')).toEqual({c: 1, d: 'j'});
        expect(Quaternion.multiplyBase('k', 'j')).toEqual({c: -1, d: 'i'});
        expect(Quaternion.multiplyBase('k', 'k')).toEqual({c: -1, d: 'r'});
    })

and Rust

    #[test]
    fn multiply_base() {
        assert_eq!(Quaternion::multiply_base('r', 'r'), SignedCoefficient { c: 1f64, d: 'r' });
        assert_eq!(Quaternion::multiply_base('r', 'i'), SignedCoefficient { c: 1f64, d: 'i' });
        assert_eq!(Quaternion::multiply_base('r', 'j'), SignedCoefficient { c: 1f64, d: 'j' });
        assert_eq!(Quaternion::multiply_base('r', 'k'), SignedCoefficient { c: 1f64, d: 'k' });

        assert_eq!(Quaternion::multiply_base('i', 'r'), SignedCoefficient { c: 1f64, d: 'i' });
        assert_eq!(Quaternion::multiply_base('i', 'i'), SignedCoefficient { c: -1f64, d: 'r' });
        assert_eq!(Quaternion::multiply_base('i', 'j'), SignedCoefficient { c: 1f64, d: 'k' });
        assert_eq!(Quaternion::multiply_base('i', 'k'), SignedCoefficient { c: -1f64, d: 'j' });

        assert_eq!(Quaternion::multiply_base('j', 'r'), SignedCoefficient { c: 1f64, d: 'j' });
        assert_eq!(Quaternion::multiply_base('j', 'i'), SignedCoefficient { c: -1f64, d: 'k' });
        assert_eq!(Quaternion::multiply_base('j', 'j'), SignedCoefficient { c: -1f64, d: 'r' });
        assert_eq!(Quaternion::multiply_base('j', 'k'), SignedCoefficient { c: 1f64, d: 'i' });

        assert_eq!(Quaternion::multiply_base('k', 'r'), SignedCoefficient { c: 1f64, d: 'k' });
        assert_eq!(Quaternion::multiply_base('k', 'i'), SignedCoefficient { c: 1f64, d: 'j' });
        assert_eq!(Quaternion::multiply_base('k', 'j'), SignedCoefficient { c: -1f64, d: 'i' });
        assert_eq!(Quaternion::multiply_base('k', 'k'), SignedCoefficient { c: -1f64, d: 'r' });
    }

In Rust I have to define SignedCoefficient that was simple anonymous objects in node

#[derive(Debug)]
struct SignedCoefficient {
    c: f64,
    d: char,
}

additionally I have to implement equation relation on them to use assert_eq.

impl PartialEq<SignedCoefficient> for SignedCoefficient {
    fn eq(&self, other: &SignedCoefficient) -> bool {
        self.c == other.c && self.d == other.d
    }
}

Function multiplyBase is super simple and to build it we have to see that:

• multiplication by 1 is always the other element. a * 1 = a and 1 * a = a
• excluding 1 we always have a * a = -1
• excluding 1 and diagonal we always receive coefficient different that these used to multiply, sign can be determined using %2 operation and direction of multiplication.

Using these observations we can define multiplication in node as

    static multiplyBase(a: Base, b: Base): { c: -1 | 1, d: Base } {
        if (a === 'r') return {c: 1, d: b};
        if (b === 'r') return {c: 1, d: a};
        if (a === b) return {c: -1, d: 'r'};
        const diff = a.charCodeAt(0) - b.charCodeAt(0);
        return {
            c: (diff > 0 ? -1 : 1) * ((diff + 2) % 2 === 0 ? -1 : 1) as -1 | 1,
            d: ['i', 'j', 'k'].find((e) => e !== a && e !== b) as Base
        }
    }

and in rust

    fn multiply_base(a: char, b: char) -> SignedCoefficient {
        if a == 'r' { return SignedCoefficient { c: 1f64, d: b }; }
        if b == 'r' { return SignedCoefficient { c: 1f64, d: a }; }
        if a == b { return SignedCoefficient { c: -1f64, d: 'r' }; }
        let diff = u32::from(a) as i32 - u32::from(b) as i32;

        SignedCoefficient {
            c: (if diff > 0 { -1f64 } else { 1f64 }) * (if (diff + 2i32) % 2 == 0 { -1f64 } else { 1f64 }),
            d: vec!['i', 'j', 'k'].iter().find(|&&e| e != a && e != b).unwrap().to_owned(),
        }
    }

Multiplication of linear combinations

I divided multiplication tests to simple and complex cases

    it('simple multiply', () => {
        const res = (new Quaternion(['1']))
            .multiply(new Quaternion(['1']))

        expect(res.r).toEqual(1);
        expect(res.i).toEqual(0);
        expect(res.j).toEqual(0);
        expect(res.k).toEqual(0);
    })

    it('complex multiply', () => {
        const res = (new Quaternion(['2i', '2j']))
            .multiply(new Quaternion(['j', '1']))

        expect(res.r).toEqual(-2);
        expect(res.i).toEqual(2);
        expect(res.j).toEqual(2);
        expect(res.k).toEqual(2);
    })

and

    #[test]
    fn simple_multiply() {
        let res = Quaternion::new(vec![String::from("1")])
            .multiply(Quaternion::new(vec![String::from("1")]));

        assert_eq!(res, Quaternion {
            r: 1f64,
            i: 0f64,
            j: 0f64,
            k: 0f64,
        })
    }

    #[test]
    fn complex_multiply() {
        let res = Quaternion::new(vec![String::from("2i"), String::from("2j")])
            .multiply(Quaternion::new(vec![String::from("j"), String::from("1")]));

        assert_eq!(res, Quaternion {
            r: -2f64,
            i: 2f64,
            j: 2f64,
            k: 2f64,
        })
    }

here to compare Quaternions we have to implement PartialEq

impl PartialEq<Quaternion> for Quaternion {
    fn eq(&self, other: &Quaternion) -> bool {
        self.r == other.r && self.i == other.i && self.j == other.j && self.k == other.k
    }
}

In NodeJS is can be reduced to nested loop like this

    multiply(a: Quaternion): Quaternion {
        const res = new Quaternion([]);
        for (let p of ['r', 'i', 'j', 'k'] as Array<Base>) {
            for (let n of ['r', 'i', 'j', 'k'] as Array<Base>) {
                const {c, d} = Quaternion.multiplyBase(p, n);
                res[d] += c * this[p] * a[n];
            }
        }
        return res;
    }

c is sign, d is name of coefficient.

In Rust we cant have access to dynamic properties that are chars so we have to add two auxiliary methods to get and set values using chars

    fn get(&self, key: char) -> f64 {
        match key {
            'r' => self.r,
            'i' => self.i,
            'j' => self.j,
            'k' => self.k,
            _ => 0f64
        }
    }

    fn set(&mut self, key: char, value: f64) -> &Quaternion {
        match key {
            'r' => self.r = value,
            'i' => self.i = value,
            'j' => self.j = value,
            'k' => self.k = value,
            _ => ()
        }

        self
    }

    fn multiply(&self, a: Quaternion) -> Quaternion {
        let mut res = Quaternion::new(vec![]);
        for p in vec!['r', 'i', 'j', 'k'] {
            for n in vec!['r', 'i', 'j', 'k'] {
                let SignedCoefficient { c, d } = Quaternion::multiply_base(p, n);
                res.set(d, res.get(d) + c * self.get(p) * a.get(n));
            }
        }
        res
    }

but generally idea is the same.

Now we have program that can read input, convert it to array of Quaternions and multiply them.

Last lacking element is formatting result as string.

Formatting Quaternion to strings

Formatting results can be considered as:

• formatting any single coefficient using special treatments for 1
• building ordered array of coefficients that is joined as sting

These operations are inversion of parsing presented in first part. Lest start from tests in node js

    it('format coefficient', () => {
       expect(Quaternion.formatCoefficient('i', 20)).toEqual('20i');
       expect(Quaternion.formatCoefficient('i', 1)).toEqual('i');
       expect(Quaternion.formatCoefficient('', 0)).toEqual('0');
    });

    it('format', () => {
        expect((new Quaternion([]).format())).toEqual('0');
        expect((new Quaternion(['1']).format())).toEqual('1');
        expect((new Quaternion(['i', '1']).format())).toEqual('i+1');
        expect((new Quaternion(['i', '-3.4j', '1']).format())).toEqual('i-3.4j+1');
        expect((new Quaternion(['j', 'k']).format())).toEqual('j+k');

    })

Analogical tests in rust

    #[test]
    fn format_coefficient() {
        assert_eq!(Quaternion::format_coefficient('i', 20f64), String::from("20i"));
        assert_eq!(Quaternion::format_coefficient('i', 1f64), String::from("i"));
        assert_eq!(Quaternion::format_coefficient(' ', 0f64), String::from("0"));
    }

    #[test]
    fn format() {
        assert_eq!(format!("{}", Quaternion::new(vec![])), String::from("0"));
        assert_eq!(format!("{}", Quaternion::new(vec![String::from("1")])), String::from("1"));
        assert_eq!(format!("{}", Quaternion::new(vec![String::from("i"), String::from("1")])), String::from("i+1"));
        assert_eq!(format!("{}", Quaternion::new(vec![String::from("i"), String::from("-3.4j"), String::from("1")])), String::from("i-3.4j+1"));
        assert_eq!(format!("{}", Quaternion::new(vec![String::from("j"), String::from("k")])), String::from("j+k"));
    }

In function formatCoefficient we carry about cases like 1, - and deciding if component name like i, j or k should be added to result.

    static formatCoefficient(type: Base | '', value: number) {
        const out = `${Math.abs(value) === 1 ? (
            Math.sign(value) === 1 ? '' : '-'
        ) : value}${type}`;
        return /[\dijk]$/.test(out) ? out : `${out}1`;
    }

and

    fn format_coefficient(t: char, value: f64) -> String {
        let out = if f64::abs(value) == 1f64 {
            if f64::signum(value) == 1f64 {
                String::from("") + &t.to_string()[..].trim()
            } else {
                String::from("-") + &t.to_string()[..].trim()
            }
        } else {
            format!("{}", value) + &t.to_string()[..].trim()
        };

        match Regex::new(r"[\dijk]$").unwrap().captures(&out[..]) {
            Some(_) => out,
            None => out + "1"
        }
    }
}

In function format we collecting these components and decide about joining. We can't join by + because of some elements starts from -. But we have to handle case of 0. Finally in NodeJS we have:

    format(): string {
        let out = [];
        if (this.i) {
            out.push(Quaternion.formatCoefficient('i', this.i));
        }
        if (this.j) {
            out.push(Quaternion.formatCoefficient('j', this.j));
        }
        if (this.k) {
            out.push(Quaternion.formatCoefficient('k', this.k));
        }
        if (this.r) {
            out.push(Quaternion.formatCoefficient('', this.r));
        }

        if (!out.length) return '0';

        return out.reduce((p, n) => p + (
            p.length && Quaternion.getCoefficient('',n.replace(/[kij]/, '')) > 0 ? `+${n}` : `${n}`), ''
        );
    }

while in rust implementation of formatting can be done by fmt function

impl fmt::Display for Quaternion {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        let mut out: Vec<String> = vec![];
        if self.i != 0f64 {
            out.push(Quaternion::format_coefficient('i', self.i))
        }
        if self.j != 0f64 {
            out.push(Quaternion::format_coefficient('j', self.j))
        }
        if self.k != 0f64 {
            out.push(Quaternion::format_coefficient('k', self.k))
        }
        if self.r != 0f64 {
            out.push(Quaternion::format_coefficient(' ', self.r))
        }

        let out = out.into_iter().reduce(
            |p, n| format!(
                "{}{}",
                p.clone(),
                if p.len() > 0 && Quaternion::get_coefficient(&"", n.replace("i", "").replace("j", "").replace("k", "")) > 0f64 {
                    format!("{}{}", "+", n)
                } else {
                    n
                }
            )
        );

        write!(f, "{}", out.unwrap_or(String::from("0")))
    }
}

Final e2e tests

To check if all parts of programs match to each other we can prepare some e2e test cases in node

    it('e2e', () => {
        const cases = [
            {
                in: '(i+j)(k)',
                out: 'i-j'
            },
            {
                in: '(i+j+20)(j-9)',
                out: '-9i+11j+k-181'
            },
            {
                in: '(10i)(10j-k+1)(-99i+j-10k+7)(4)',
                out: '-520i-38920j+6800k+7920'
            },
            {
                in: '(i+j+k+1)(i+2j+4k+8)(i+3j+9k+27)(i+j+8k+8)(i-j+k-10)(99i-j+k-1)(k)(j)(i)(3)',
                out: '11415288i-8751432j-5206896k+9766704'
            }
        ]
        for (const c of cases) {
            const qs = Quaternion.parse(c.in);
            const out = qs.reduce((p, n) => p.multiply(n)).format();
            expect(out).toEqual(c.out);
        }
    })

and analogically in rust

#[test]
    fn e2e() {
        struct Case {
            input: String,
            output: String,
        }

        let cases: Vec<Case> = vec![
            Case {
                input: String::from("(i+j)(k)"),
                output: String::from("i-j"),
            },
            Case {
                input: String::from("(i+j+20)(j-9)"),
                output: String::from("-9i+11j+k-181"),
            },
            Case {
                input: String::from("(10i)(10j-k+1)(-99i+j-10k+7)(4)"),
                output: String::from("-520i-38920j+6800k+7920"),
            },
            Case {
                input: String::from("(i+j+k+1)(i+2j+4k+8)(i+3j+9k+27)(i+j+8k+8)(i-j+k-10)(99i-j+k-1)(k)(j)(i)(3)"),
                output: String::from("11415288i-8751432j-5206896k+9766704"),
            },
        ];
        for c in cases {
            let qs = Quaternion::parse(&c.input[..]);
            let out = qs.into_iter().reduce(|p, n| p.multiply(n)).unwrap();
            assert_eq!(format!("{}", out), c.output);
        }
    }

It is the end of this exercise. If you are interested in learning more about Quaternions multiplication and how it is connected with geometry I recommend you video:

As you can see rust and typescript have a lot of similar elements. All descriptions and logic elements are identical and only differences can be seen on level of syntax, that is more focused on elimination of undefined behaviors in rust. On the other hand in typescript code can be written in a little more concise way that can improve readability.