Curve intersection, scale width based on ribbon orientation

Percentage Accurate: 97.5% → 99.4%

Time: 29.0s

Alternatives: 10

Speedup: 60.1×

Specification

\[\left(\left(\left(0 \leq normAngle \land normAngle \leq \frac{\pi}{2}\right) \land \left(-1 \leq n0_i \land n0_i \leq 1\right)\right) \land \left(-1 \leq n1_i \land n1_i \leq 1\right)\right) \land \left(2.328306437 \cdot 10^{-10} \leq u \land u \leq 1\right)\]

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \frac{1}{\sin normAngle}\\ \left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot t_0\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot t_0\right) \cdot n1_i \end{array} \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i)
 :precision binary32
 (let* ((t_0 (/ 1.0 (sin normAngle))))
   (+
    (* (* (sin (* (- 1.0 u) normAngle)) t_0) n0_i)
    (* (* (sin (* u normAngle)) t_0) n1_i))))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	float t_0 = 1.0f / sinf(normAngle);
	return ((sinf(((1.0f - u) * normAngle)) * t_0) * n0_i) + ((sinf((u * normAngle)) * t_0) * n1_i);
}

Fortran

real(4) function code(normangle, u, n0_i, n1_i)
    real(4), intent (in) :: normangle
    real(4), intent (in) :: u
    real(4), intent (in) :: n0_i
    real(4), intent (in) :: n1_i
    real(4) :: t_0
    t_0 = 1.0e0 / sin(normangle)
    code = ((sin(((1.0e0 - u) * normangle)) * t_0) * n0_i) + ((sin((u * normangle)) * t_0) * n1_i)
end function

Julia

function code(normAngle, u, n0_i, n1_i)
	t_0 = Float32(Float32(1.0) / sin(normAngle))
	return Float32(Float32(Float32(sin(Float32(Float32(Float32(1.0) - u) * normAngle)) * t_0) * n0_i) + Float32(Float32(sin(Float32(u * normAngle)) * t_0) * n1_i))
end

MATLAB

function tmp = code(normAngle, u, n0_i, n1_i)
	t_0 = single(1.0) / sin(normAngle);
	tmp = ((sin(((single(1.0) - u) * normAngle)) * t_0) * n0_i) + ((sin((u * normAngle)) * t_0) * n1_i);
end

Initial Program: 97.5% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \frac{1}{\sin normAngle}\\ \left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot t_0\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot t_0\right) \cdot n1_i \end{array} \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i)
 :precision binary32
 (let* ((t_0 (/ 1.0 (sin normAngle))))
   (+
    (* (* (sin (* (- 1.0 u) normAngle)) t_0) n0_i)
    (* (* (sin (* u normAngle)) t_0) n1_i))))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	float t_0 = 1.0f / sinf(normAngle);
	return ((sinf(((1.0f - u) * normAngle)) * t_0) * n0_i) + ((sinf((u * normAngle)) * t_0) * n1_i);
}

Fortran

real(4) function code(normangle, u, n0_i, n1_i)
    real(4), intent (in) :: normangle
    real(4), intent (in) :: u
    real(4), intent (in) :: n0_i
    real(4), intent (in) :: n1_i
    real(4) :: t_0
    t_0 = 1.0e0 / sin(normangle)
    code = ((sin(((1.0e0 - u) * normangle)) * t_0) * n0_i) + ((sin((u * normangle)) * t_0) * n1_i)
end function

Julia

function code(normAngle, u, n0_i, n1_i)
	t_0 = Float32(Float32(1.0) / sin(normAngle))
	return Float32(Float32(Float32(sin(Float32(Float32(Float32(1.0) - u) * normAngle)) * t_0) * n0_i) + Float32(Float32(sin(Float32(u * normAngle)) * t_0) * n1_i))
end

MATLAB

function tmp = code(normAngle, u, n0_i, n1_i)
	t_0 = single(1.0) / sin(normAngle);
	tmp = ((sin(((single(1.0) - u) * normAngle)) * t_0) * n0_i) + ((sin((u * normAngle)) * t_0) * n1_i);
end

Alternative 1: 99.4% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}, n0_i\right) \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i)
 :precision binary32
 (fma
  u
  (-
   (/ n1_i (/ (sin normAngle) normAngle))
   (/ n0_i (/ (sin normAngle) (* normAngle (cos normAngle)))))
  n0_i))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	return fmaf(u, ((n1_i / (sinf(normAngle) / normAngle)) - (n0_i / (sinf(normAngle) / (normAngle * cosf(normAngle))))), n0_i);
}

Julia

function code(normAngle, u, n0_i, n1_i)
	return fma(u, Float32(Float32(n1_i / Float32(sin(normAngle) / normAngle)) - Float32(n0_i / Float32(sin(normAngle) / Float32(normAngle * cos(normAngle))))), n0_i)
end

Derivation

1. Initial program 97.6%

   \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
2. Add Preprocessing

3. Taylor expanded in u around 0 90.7%

   \[\leadsto \color{blue}{n0_i + u \cdot \left(-1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}\right)} \]
4. Step-by-step derivation

   1. +-commutative 90.7%

      \[\leadsto \color{blue}{u \cdot \left(-1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}\right) + n0_i} \]

   2. fma-def 90.8%

      \[\leadsto \color{blue}{\mathsf{fma}\left(u, -1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}, n0_i\right)} \]

   3. +-commutative 90.8%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i \cdot normAngle}{\sin normAngle} + -1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}}, n0_i\right) \]

   4. mul-1-neg 90.8%

      \[\leadsto \mathsf{fma}\left(u, \frac{n1_i \cdot normAngle}{\sin normAngle} + \color{blue}{\left(-\frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}\right)}, n0_i\right) \]

   5. unsub-neg 90.8%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i \cdot normAngle}{\sin normAngle} - \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}}, n0_i\right) \]

   6. associate-/l* 96.6%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i}{\frac{\sin normAngle}{normAngle}}} - \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}, n0_i\right) \]

   7. associate-/l* 99.6%

      \[\leadsto \mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \color{blue}{\frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}}, n0_i\right) \]

5. Simplified 99.6%

   \[\leadsto \color{blue}{\mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}, n0_i\right)} \]

6. Final simplification 99.6%

   \[\leadsto \mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}, n0_i\right) \]
7. Add Preprocessing

Alternative 2: 98.8% accurate, 3.6× speedup

Math

\[\begin{array}{l} \\ n0_i + u \cdot \left(\left(n1_i + {normAngle}^{2} \cdot \left(n1_i \cdot 0.16666666666666666 + n0_i \cdot 0.3333333333333333\right)\right) - n0_i\right) \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i)
 :precision binary32
 (+
  n0_i
  (*
   u
   (-
    (+
     n1_i
     (*
      (pow normAngle 2.0)
      (+ (* n1_i 0.16666666666666666) (* n0_i 0.3333333333333333))))
    n0_i))))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	return n0_i + (u * ((n1_i + (powf(normAngle, 2.0f) * ((n1_i * 0.16666666666666666f) + (n0_i * 0.3333333333333333f)))) - n0_i));
}

Fortran

real(4) function code(normangle, u, n0_i, n1_i)
    real(4), intent (in) :: normangle
    real(4), intent (in) :: u
    real(4), intent (in) :: n0_i
    real(4), intent (in) :: n1_i
    code = n0_i + (u * ((n1_i + ((normangle ** 2.0e0) * ((n1_i * 0.16666666666666666e0) + (n0_i * 0.3333333333333333e0)))) - n0_i))
end function

Julia

function code(normAngle, u, n0_i, n1_i)
	return Float32(n0_i + Float32(u * Float32(Float32(n1_i + Float32((normAngle ^ Float32(2.0)) * Float32(Float32(n1_i * Float32(0.16666666666666666)) + Float32(n0_i * Float32(0.3333333333333333))))) - n0_i)))
end

MATLAB

function tmp = code(normAngle, u, n0_i, n1_i)
	tmp = n0_i + (u * ((n1_i + ((normAngle ^ single(2.0)) * ((n1_i * single(0.16666666666666666)) + (n0_i * single(0.3333333333333333))))) - n0_i));
end

Derivation

1. Initial program 97.6%

   \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
2. Add Preprocessing

3. Taylor expanded in u around 0 90.7%

   \[\leadsto \color{blue}{n0_i + u \cdot \left(-1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}\right)} \]
4. Step-by-step derivation

   1. +-commutative 90.7%

      \[\leadsto \color{blue}{u \cdot \left(-1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}\right) + n0_i} \]

   2. fma-def 90.8%

      \[\leadsto \color{blue}{\mathsf{fma}\left(u, -1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}, n0_i\right)} \]

   3. +-commutative 90.8%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i \cdot normAngle}{\sin normAngle} + -1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}}, n0_i\right) \]

   4. mul-1-neg 90.8%

      \[\leadsto \mathsf{fma}\left(u, \frac{n1_i \cdot normAngle}{\sin normAngle} + \color{blue}{\left(-\frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}\right)}, n0_i\right) \]

   5. unsub-neg 90.8%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i \cdot normAngle}{\sin normAngle} - \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}}, n0_i\right) \]

   6. associate-/l* 96.6%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i}{\frac{\sin normAngle}{normAngle}}} - \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}, n0_i\right) \]

   7. associate-/l* 99.6%

      \[\leadsto \mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \color{blue}{\frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}}, n0_i\right) \]

5. Simplified 99.6%

   \[\leadsto \color{blue}{\mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}, n0_i\right)} \]

6. Taylor expanded in normAngle around 0 99.4%

   \[\leadsto \color{blue}{n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \left(u \cdot \left(-0.16666666666666666 \cdot n0_i - \left(-0.5 \cdot n0_i + -0.16666666666666666 \cdot n1_i\right)\right)\right)\right)} \]
7. Step-by-step derivation

   1. associate--r+ 99.4%

      \[\leadsto n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \left(u \cdot \color{blue}{\left(\left(-0.16666666666666666 \cdot n0_i - -0.5 \cdot n0_i\right) - -0.16666666666666666 \cdot n1_i\right)}\right)\right) \]

   2. cancel-sign-sub-inv 99.4%

      \[\leadsto n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \left(u \cdot \color{blue}{\left(\left(-0.16666666666666666 \cdot n0_i - -0.5 \cdot n0_i\right) + \left(--0.16666666666666666\right) \cdot n1_i\right)}\right)\right) \]

   3. distribute-rgt-out-- 99.4%

      \[\leadsto n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \left(u \cdot \left(\color{blue}{n0_i \cdot \left(-0.16666666666666666 - -0.5\right)} + \left(--0.16666666666666666\right) \cdot n1_i\right)\right)\right) \]

   4. metadata-eval 99.4%

      \[\leadsto n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \left(u \cdot \left(n0_i \cdot \color{blue}{0.3333333333333333} + \left(--0.16666666666666666\right) \cdot n1_i\right)\right)\right) \]

   5. metadata-eval 99.4%

      \[\leadsto n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \left(u \cdot \left(n0_i \cdot 0.3333333333333333 + \color{blue}{0.16666666666666666} \cdot n1_i\right)\right)\right) \]

8. Applied egg-rr 99.4%

   \[\leadsto n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \left(u \cdot \color{blue}{\left(n0_i \cdot 0.3333333333333333 + 0.16666666666666666 \cdot n1_i\right)}\right)\right) \]

9. Taylor expanded in u around 0 99.4%

   \[\leadsto \color{blue}{n0_i + u \cdot \left(\left(n1_i + {normAngle}^{2} \cdot \left(0.16666666666666666 \cdot n1_i + 0.3333333333333333 \cdot n0_i\right)\right) - n0_i\right)} \]

10. Final simplification 99.4%

    \[\leadsto n0_i + u \cdot \left(\left(n1_i + {normAngle}^{2} \cdot \left(n1_i \cdot 0.16666666666666666 + n0_i \cdot 0.3333333333333333\right)\right) - n0_i\right) \]
11. Add Preprocessing

Alternative 3: 98.7% accurate, 3.6× speedup

Math

\[\begin{array}{l} \\ n0_i + \left({normAngle}^{2} \cdot \left(u \cdot \left(n1_i \cdot 0.16666666666666666\right)\right) + u \cdot \left(n1_i - n0_i\right)\right) \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i)
 :precision binary32
 (+
  n0_i
  (+
   (* (pow normAngle 2.0) (* u (* n1_i 0.16666666666666666)))
   (* u (- n1_i n0_i)))))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	return n0_i + ((powf(normAngle, 2.0f) * (u * (n1_i * 0.16666666666666666f))) + (u * (n1_i - n0_i)));
}

Fortran

real(4) function code(normangle, u, n0_i, n1_i)
    real(4), intent (in) :: normangle
    real(4), intent (in) :: u
    real(4), intent (in) :: n0_i
    real(4), intent (in) :: n1_i
    code = n0_i + (((normangle ** 2.0e0) * (u * (n1_i * 0.16666666666666666e0))) + (u * (n1_i - n0_i)))
end function

Julia

function code(normAngle, u, n0_i, n1_i)
	return Float32(n0_i + Float32(Float32((normAngle ^ Float32(2.0)) * Float32(u * Float32(n1_i * Float32(0.16666666666666666)))) + Float32(u * Float32(n1_i - n0_i))))
end

MATLAB

function tmp = code(normAngle, u, n0_i, n1_i)
	tmp = n0_i + (((normAngle ^ single(2.0)) * (u * (n1_i * single(0.16666666666666666)))) + (u * (n1_i - n0_i)));
end

Derivation

1. Initial program 97.6%

   \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
2. Add Preprocessing

3. Taylor expanded in u around 0 90.7%

   \[\leadsto \color{blue}{n0_i + u \cdot \left(-1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}\right)} \]
4. Step-by-step derivation

   1. +-commutative 90.7%

      \[\leadsto \color{blue}{u \cdot \left(-1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}\right) + n0_i} \]

   2. fma-def 90.8%

      \[\leadsto \color{blue}{\mathsf{fma}\left(u, -1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}, n0_i\right)} \]

   3. +-commutative 90.8%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i \cdot normAngle}{\sin normAngle} + -1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}}, n0_i\right) \]

   4. mul-1-neg 90.8%

      \[\leadsto \mathsf{fma}\left(u, \frac{n1_i \cdot normAngle}{\sin normAngle} + \color{blue}{\left(-\frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}\right)}, n0_i\right) \]

   5. unsub-neg 90.8%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i \cdot normAngle}{\sin normAngle} - \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}}, n0_i\right) \]

   6. associate-/l* 96.6%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i}{\frac{\sin normAngle}{normAngle}}} - \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}, n0_i\right) \]

   7. associate-/l* 99.6%

      \[\leadsto \mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \color{blue}{\frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}}, n0_i\right) \]

5. Simplified 99.6%

   \[\leadsto \color{blue}{\mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}, n0_i\right)} \]

6. Taylor expanded in normAngle around 0 99.4%

   \[\leadsto \color{blue}{n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \left(u \cdot \left(-0.16666666666666666 \cdot n0_i - \left(-0.5 \cdot n0_i + -0.16666666666666666 \cdot n1_i\right)\right)\right)\right)} \]

7. Taylor expanded in n0_i around 0 99.3%

   \[\leadsto n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \color{blue}{\left(0.16666666666666666 \cdot \left(n1_i \cdot u\right)\right)}\right) \]
8. Step-by-step derivation

   1. associate-*r* 99.3%

      \[\leadsto n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \color{blue}{\left(\left(0.16666666666666666 \cdot n1_i\right) \cdot u\right)}\right) \]

9. Simplified 99.3%

   \[\leadsto n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \color{blue}{\left(\left(0.16666666666666666 \cdot n1_i\right) \cdot u\right)}\right) \]

10. Final simplification 99.3%

    \[\leadsto n0_i + \left({normAngle}^{2} \cdot \left(u \cdot \left(n1_i \cdot 0.16666666666666666\right)\right) + u \cdot \left(n1_i - n0_i\right)\right) \]
11. Add Preprocessing

Alternative 4: 98.2% accurate, 3.7× speedup

Math

\[\begin{array}{l} \\ n0_i + u \cdot \left(\left(n1_i - n0_i\right) - n0_i \cdot \left({normAngle}^{2} \cdot -0.3333333333333333\right)\right) \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i)
 :precision binary32
 (+
  n0_i
  (*
   u
   (- (- n1_i n0_i) (* n0_i (* (pow normAngle 2.0) -0.3333333333333333))))))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	return n0_i + (u * ((n1_i - n0_i) - (n0_i * (powf(normAngle, 2.0f) * -0.3333333333333333f))));
}

Fortran

real(4) function code(normangle, u, n0_i, n1_i)
    real(4), intent (in) :: normangle
    real(4), intent (in) :: u
    real(4), intent (in) :: n0_i
    real(4), intent (in) :: n1_i
    code = n0_i + (u * ((n1_i - n0_i) - (n0_i * ((normangle ** 2.0e0) * (-0.3333333333333333e0)))))
end function

Julia

function code(normAngle, u, n0_i, n1_i)
	return Float32(n0_i + Float32(u * Float32(Float32(n1_i - n0_i) - Float32(n0_i * Float32((normAngle ^ Float32(2.0)) * Float32(-0.3333333333333333))))))
end

MATLAB

function tmp = code(normAngle, u, n0_i, n1_i)
	tmp = n0_i + (u * ((n1_i - n0_i) - (n0_i * ((normAngle ^ single(2.0)) * single(-0.3333333333333333)))));
end

Derivation

1. Initial program 97.6%

   \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
2. Add Preprocessing

3. Taylor expanded in u around 0 90.7%

   \[\leadsto \color{blue}{n0_i + u \cdot \left(-1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}\right)} \]
4. Step-by-step derivation

   1. +-commutative 90.7%

      \[\leadsto \color{blue}{u \cdot \left(-1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}\right) + n0_i} \]

   2. fma-def 90.8%

      \[\leadsto \color{blue}{\mathsf{fma}\left(u, -1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}, n0_i\right)} \]

   3. +-commutative 90.8%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i \cdot normAngle}{\sin normAngle} + -1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}}, n0_i\right) \]

   4. mul-1-neg 90.8%

      \[\leadsto \mathsf{fma}\left(u, \frac{n1_i \cdot normAngle}{\sin normAngle} + \color{blue}{\left(-\frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}\right)}, n0_i\right) \]

   5. unsub-neg 90.8%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i \cdot normAngle}{\sin normAngle} - \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}}, n0_i\right) \]

   6. associate-/l* 96.6%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i}{\frac{\sin normAngle}{normAngle}}} - \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}, n0_i\right) \]

   7. associate-/l* 99.6%

      \[\leadsto \mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \color{blue}{\frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}}, n0_i\right) \]

5. Simplified 99.6%

   \[\leadsto \color{blue}{\mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}, n0_i\right)} \]

6. Taylor expanded in normAngle around 0 99.4%

   \[\leadsto \color{blue}{n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \left(u \cdot \left(-0.16666666666666666 \cdot n0_i - \left(-0.5 \cdot n0_i + -0.16666666666666666 \cdot n1_i\right)\right)\right)\right)} \]

7. Taylor expanded in n0_i around inf 99.0%

   \[\leadsto n0_i + \left(u \cdot \left(n1_i - n0_i\right) + {normAngle}^{2} \cdot \left(u \cdot \color{blue}{\left(0.3333333333333333 \cdot n0_i\right)}\right)\right) \]

8. Taylor expanded in u around -inf 99.0%

   \[\leadsto \color{blue}{n0_i + -1 \cdot \left(u \cdot \left(-1 \cdot \left(n1_i - n0_i\right) + -0.3333333333333333 \cdot \left(n0_i \cdot {normAngle}^{2}\right)\right)\right)} \]
9. Step-by-step derivation

   1. mul-1-neg 99.0%

      \[\leadsto n0_i + \color{blue}{\left(-u \cdot \left(-1 \cdot \left(n1_i - n0_i\right) + -0.3333333333333333 \cdot \left(n0_i \cdot {normAngle}^{2}\right)\right)\right)} \]

   2. unsub-neg 99.0%

      \[\leadsto \color{blue}{n0_i - u \cdot \left(-1 \cdot \left(n1_i - n0_i\right) + -0.3333333333333333 \cdot \left(n0_i \cdot {normAngle}^{2}\right)\right)} \]

   3. +-commutative 99.0%

      \[\leadsto n0_i - u \cdot \color{blue}{\left(-0.3333333333333333 \cdot \left(n0_i \cdot {normAngle}^{2}\right) + -1 \cdot \left(n1_i - n0_i\right)\right)} \]

   4. mul-1-neg 99.0%

      \[\leadsto n0_i - u \cdot \left(-0.3333333333333333 \cdot \left(n0_i \cdot {normAngle}^{2}\right) + \color{blue}{\left(-\left(n1_i - n0_i\right)\right)}\right) \]

   5. unsub-neg 99.0%

      \[\leadsto n0_i - u \cdot \color{blue}{\left(-0.3333333333333333 \cdot \left(n0_i \cdot {normAngle}^{2}\right) - \left(n1_i - n0_i\right)\right)} \]

   6. *-commutative 99.0%

      \[\leadsto n0_i - u \cdot \left(\color{blue}{\left(n0_i \cdot {normAngle}^{2}\right) \cdot -0.3333333333333333} - \left(n1_i - n0_i\right)\right) \]

   7. associate-*l* 99.0%

      \[\leadsto n0_i - u \cdot \left(\color{blue}{n0_i \cdot \left({normAngle}^{2} \cdot -0.3333333333333333\right)} - \left(n1_i - n0_i\right)\right) \]

10. Simplified 99.0%

    \[\leadsto \color{blue}{n0_i - u \cdot \left(n0_i \cdot \left({normAngle}^{2} \cdot -0.3333333333333333\right) - \left(n1_i - n0_i\right)\right)} \]

11. Final simplification 99.0%

    \[\leadsto n0_i + u \cdot \left(\left(n1_i - n0_i\right) - n0_i \cdot \left({normAngle}^{2} \cdot -0.3333333333333333\right)\right) \]
12. Add Preprocessing

Alternative 5: 98.2% accurate, 4.0× speedup

Math

\[\begin{array}{l} \\ \mathsf{fma}\left(u, n1_i - n0_i, n0_i\right) \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i) :precision binary32 (fma u (- n1_i n0_i) n0_i))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	return fmaf(u, (n1_i - n0_i), n0_i);
}

Julia

function code(normAngle, u, n0_i, n1_i)
	return fma(u, Float32(n1_i - n0_i), n0_i)
end

Derivation

1. Initial program 97.6%

   \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
2. Add Preprocessing

3. Taylor expanded in u around 0 90.7%

   \[\leadsto \color{blue}{n0_i + u \cdot \left(-1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}\right)} \]
4. Step-by-step derivation

   1. +-commutative 90.7%

      \[\leadsto \color{blue}{u \cdot \left(-1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}\right) + n0_i} \]

   2. fma-def 90.8%

      \[\leadsto \color{blue}{\mathsf{fma}\left(u, -1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle} + \frac{n1_i \cdot normAngle}{\sin normAngle}, n0_i\right)} \]

   3. +-commutative 90.8%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i \cdot normAngle}{\sin normAngle} + -1 \cdot \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}}, n0_i\right) \]

   4. mul-1-neg 90.8%

      \[\leadsto \mathsf{fma}\left(u, \frac{n1_i \cdot normAngle}{\sin normAngle} + \color{blue}{\left(-\frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}\right)}, n0_i\right) \]

   5. unsub-neg 90.8%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i \cdot normAngle}{\sin normAngle} - \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}}, n0_i\right) \]

   6. associate-/l* 96.6%

      \[\leadsto \mathsf{fma}\left(u, \color{blue}{\frac{n1_i}{\frac{\sin normAngle}{normAngle}}} - \frac{n0_i \cdot \left(normAngle \cdot \cos normAngle\right)}{\sin normAngle}, n0_i\right) \]

   7. associate-/l* 99.6%

      \[\leadsto \mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \color{blue}{\frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}}, n0_i\right) \]

5. Simplified 99.6%

   \[\leadsto \color{blue}{\mathsf{fma}\left(u, \frac{n1_i}{\frac{\sin normAngle}{normAngle}} - \frac{n0_i}{\frac{\sin normAngle}{normAngle \cdot \cos normAngle}}, n0_i\right)} \]

6. Taylor expanded in normAngle around 0 98.9%

   \[\leadsto \color{blue}{n0_i + u \cdot \left(n1_i - n0_i\right)} \]
7. Step-by-step derivation

   1. +-commutative 98.9%

      \[\leadsto \color{blue}{u \cdot \left(n1_i - n0_i\right) + n0_i} \]

   2. fma-def 99.0%

      \[\leadsto \color{blue}{\mathsf{fma}\left(u, n1_i - n0_i, n0_i\right)} \]

8. Simplified 99.0%

   \[\leadsto \color{blue}{\mathsf{fma}\left(u, n1_i - n0_i, n0_i\right)} \]

9. Final simplification 99.0%

   \[\leadsto \mathsf{fma}\left(u, n1_i - n0_i, n0_i\right) \]
10. Add Preprocessing

Alternative 6: 71.6% accurate, 27.9× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;n1_i \leq -1.99999996490334 \cdot 10^{-14} \lor \neg \left(n1_i \leq 2.20000002915631 \cdot 10^{-14}\right):\\ \;\;\;\;u \cdot n1_i\\ \mathbf{else}:\\ \;\;\;\;n0_i \cdot \left(1 - u\right)\\ \end{array} \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i)
 :precision binary32
 (if (or (<= n1_i -1.99999996490334e-14) (not (<= n1_i 2.20000002915631e-14)))
   (* u n1_i)
   (* n0_i (- 1.0 u))))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	float tmp;
	if ((n1_i <= -1.99999996490334e-14f) || !(n1_i <= 2.20000002915631e-14f)) {
		tmp = u * n1_i;
	} else {
		tmp = n0_i * (1.0f - u);
	}
	return tmp;
}

Fortran

real(4) function code(normangle, u, n0_i, n1_i)
    real(4), intent (in) :: normangle
    real(4), intent (in) :: u
    real(4), intent (in) :: n0_i
    real(4), intent (in) :: n1_i
    real(4) :: tmp
    if ((n1_i <= (-1.99999996490334e-14)) .or. (.not. (n1_i <= 2.20000002915631e-14))) then
        tmp = u * n1_i
    else
        tmp = n0_i * (1.0e0 - u)
    end if
    code = tmp
end function

Julia

function code(normAngle, u, n0_i, n1_i)
	tmp = Float32(0.0)
	if ((n1_i <= Float32(-1.99999996490334e-14)) || !(n1_i <= Float32(2.20000002915631e-14)))
		tmp = Float32(u * n1_i);
	else
		tmp = Float32(n0_i * Float32(Float32(1.0) - u));
	end
	return tmp
end

MATLAB

function tmp_2 = code(normAngle, u, n0_i, n1_i)
	tmp = single(0.0);
	if ((n1_i <= single(-1.99999996490334e-14)) || ~((n1_i <= single(2.20000002915631e-14))))
		tmp = u * n1_i;
	else
		tmp = n0_i * (single(1.0) - u);
	end
	tmp_2 = tmp;
end

Derivation

1. Split input into 2 regimes

2. if n1_i < -1.99999996e-14 or 2.20000003e-14 < n1_i

   1. Initial program 97.0%

      \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
   2. Add Preprocessing

   3. Taylor expanded in normAngle around 0 98.8%

      \[\leadsto \color{blue}{n0_i \cdot \left(1 - u\right) + n1_i \cdot u} \]

   4. Taylor expanded in n0_i around 0 69.6%

      \[\leadsto \color{blue}{n1_i \cdot u} \]
   5. Step-by-step derivation

      1. *-commutative 69.6%

         \[\leadsto \color{blue}{u \cdot n1_i} \]

   6. Simplified 69.6%

      \[\leadsto \color{blue}{u \cdot n1_i} \]

   if -1.99999996e-14 < n1_i < 2.20000003e-14

   1. Initial program 98.1%

      \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
   2. Add Preprocessing

   3. Taylor expanded in normAngle around 0 98.6%

      \[\leadsto \color{blue}{n0_i \cdot \left(1 - u\right) + n1_i \cdot u} \]

   4. Taylor expanded in n0_i around inf 78.9%

      \[\leadsto \color{blue}{n0_i \cdot \left(1 - u\right)} \]
3. Recombined 2 regimes into one program.

4. Final simplification 75.1%

   \[\leadsto \begin{array}{l} \mathbf{if}\;n1_i \leq -1.99999996490334 \cdot 10^{-14} \lor \neg \left(n1_i \leq 2.20000002915631 \cdot 10^{-14}\right):\\ \;\;\;\;u \cdot n1_i\\ \mathbf{else}:\\ \;\;\;\;n0_i \cdot \left(1 - u\right)\\ \end{array} \]
5. Add Preprocessing

Alternative 7: 85.9% accurate, 27.9× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;n1_i \leq -9.999999887266023 \cdot 10^{-27} \lor \neg \left(n1_i \leq 1.999999936531045 \cdot 10^{-20}\right):\\ \;\;\;\;n0_i + u \cdot n1_i\\ \mathbf{else}:\\ \;\;\;\;n0_i \cdot \left(1 - u\right)\\ \end{array} \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i)
 :precision binary32
 (if (or (<= n1_i -9.999999887266023e-27)
         (not (<= n1_i 1.999999936531045e-20)))
   (+ n0_i (* u n1_i))
   (* n0_i (- 1.0 u))))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	float tmp;
	if ((n1_i <= -9.999999887266023e-27f) || !(n1_i <= 1.999999936531045e-20f)) {
		tmp = n0_i + (u * n1_i);
	} else {
		tmp = n0_i * (1.0f - u);
	}
	return tmp;
}

Fortran

real(4) function code(normangle, u, n0_i, n1_i)
    real(4), intent (in) :: normangle
    real(4), intent (in) :: u
    real(4), intent (in) :: n0_i
    real(4), intent (in) :: n1_i
    real(4) :: tmp
    if ((n1_i <= (-9.999999887266023e-27)) .or. (.not. (n1_i <= 1.999999936531045e-20))) then
        tmp = n0_i + (u * n1_i)
    else
        tmp = n0_i * (1.0e0 - u)
    end if
    code = tmp
end function

Julia

function code(normAngle, u, n0_i, n1_i)
	tmp = Float32(0.0)
	if ((n1_i <= Float32(-9.999999887266023e-27)) || !(n1_i <= Float32(1.999999936531045e-20)))
		tmp = Float32(n0_i + Float32(u * n1_i));
	else
		tmp = Float32(n0_i * Float32(Float32(1.0) - u));
	end
	return tmp
end

MATLAB

function tmp_2 = code(normAngle, u, n0_i, n1_i)
	tmp = single(0.0);
	if ((n1_i <= single(-9.999999887266023e-27)) || ~((n1_i <= single(1.999999936531045e-20))))
		tmp = n0_i + (u * n1_i);
	else
		tmp = n0_i * (single(1.0) - u);
	end
	tmp_2 = tmp;
end

Derivation

1. Split input into 2 regimes

2. if n1_i < -9.99999989e-27 or 1.99999994e-20 < n1_i

   1. Initial program 97.1%

      \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
   2. Add Preprocessing

   3. Taylor expanded in u around 0 87.9%

      \[\leadsto \color{blue}{1} \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]

   4. Taylor expanded in normAngle around 0 89.4%

      \[\leadsto \color{blue}{n0_i + n1_i \cdot u} \]
   5. Step-by-step derivation

      1. *-commutative 89.4%

         \[\leadsto n0_i + \color{blue}{u \cdot n1_i} \]

   6. Simplified 89.4%

      \[\leadsto \color{blue}{n0_i + u \cdot n1_i} \]

   if -9.99999989e-27 < n1_i < 1.99999994e-20

   1. Initial program 98.4%

      \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
   2. Add Preprocessing

   3. Taylor expanded in normAngle around 0 99.0%

      \[\leadsto \color{blue}{n0_i \cdot \left(1 - u\right) + n1_i \cdot u} \]

   4. Taylor expanded in n0_i around inf 89.0%

      \[\leadsto \color{blue}{n0_i \cdot \left(1 - u\right)} \]
3. Recombined 2 regimes into one program.

4. Final simplification 89.2%

   \[\leadsto \begin{array}{l} \mathbf{if}\;n1_i \leq -9.999999887266023 \cdot 10^{-27} \lor \neg \left(n1_i \leq 1.999999936531045 \cdot 10^{-20}\right):\\ \;\;\;\;n0_i + u \cdot n1_i\\ \mathbf{else}:\\ \;\;\;\;n0_i \cdot \left(1 - u\right)\\ \end{array} \]
5. Add Preprocessing

Alternative 8: 60.9% accurate, 32.2× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;n1_i \leq -1.99999996490334 \cdot 10^{-14} \lor \neg \left(n1_i \leq 2.20000002915631 \cdot 10^{-14}\right):\\ \;\;\;\;u \cdot n1_i\\ \mathbf{else}:\\ \;\;\;\;n0_i\\ \end{array} \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i)
 :precision binary32
 (if (or (<= n1_i -1.99999996490334e-14) (not (<= n1_i 2.20000002915631e-14)))
   (* u n1_i)
   n0_i))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	float tmp;
	if ((n1_i <= -1.99999996490334e-14f) || !(n1_i <= 2.20000002915631e-14f)) {
		tmp = u * n1_i;
	} else {
		tmp = n0_i;
	}
	return tmp;
}

Fortran

real(4) function code(normangle, u, n0_i, n1_i)
    real(4), intent (in) :: normangle
    real(4), intent (in) :: u
    real(4), intent (in) :: n0_i
    real(4), intent (in) :: n1_i
    real(4) :: tmp
    if ((n1_i <= (-1.99999996490334e-14)) .or. (.not. (n1_i <= 2.20000002915631e-14))) then
        tmp = u * n1_i
    else
        tmp = n0_i
    end if
    code = tmp
end function

Julia

function code(normAngle, u, n0_i, n1_i)
	tmp = Float32(0.0)
	if ((n1_i <= Float32(-1.99999996490334e-14)) || !(n1_i <= Float32(2.20000002915631e-14)))
		tmp = Float32(u * n1_i);
	else
		tmp = n0_i;
	end
	return tmp
end

MATLAB

function tmp_2 = code(normAngle, u, n0_i, n1_i)
	tmp = single(0.0);
	if ((n1_i <= single(-1.99999996490334e-14)) || ~((n1_i <= single(2.20000002915631e-14))))
		tmp = u * n1_i;
	else
		tmp = n0_i;
	end
	tmp_2 = tmp;
end

Derivation

1. Split input into 2 regimes

2. if n1_i < -1.99999996e-14 or 2.20000003e-14 < n1_i

   1. Initial program 97.0%

      \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
   2. Add Preprocessing

   3. Taylor expanded in normAngle around 0 98.8%

      \[\leadsto \color{blue}{n0_i \cdot \left(1 - u\right) + n1_i \cdot u} \]

   4. Taylor expanded in n0_i around 0 69.6%

      \[\leadsto \color{blue}{n1_i \cdot u} \]
   5. Step-by-step derivation

      1. *-commutative 69.6%

         \[\leadsto \color{blue}{u \cdot n1_i} \]

   6. Simplified 69.6%

      \[\leadsto \color{blue}{u \cdot n1_i} \]

   if -1.99999996e-14 < n1_i < 2.20000003e-14

   1. Initial program 98.1%

      \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
   2. Add Preprocessing

   3. Taylor expanded in u around 0 61.6%

      \[\leadsto \color{blue}{n0_i} \]
3. Recombined 2 regimes into one program.

4. Final simplification 64.9%

   \[\leadsto \begin{array}{l} \mathbf{if}\;n1_i \leq -1.99999996490334 \cdot 10^{-14} \lor \neg \left(n1_i \leq 2.20000002915631 \cdot 10^{-14}\right):\\ \;\;\;\;u \cdot n1_i\\ \mathbf{else}:\\ \;\;\;\;n0_i\\ \end{array} \]
5. Add Preprocessing

Alternative 9: 98.1% accurate, 60.1× speedup

Math

\[\begin{array}{l} \\ n0_i + u \cdot \left(n1_i - n0_i\right) \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i)
 :precision binary32
 (+ n0_i (* u (- n1_i n0_i))))

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	return n0_i + (u * (n1_i - n0_i));
}

Fortran

real(4) function code(normangle, u, n0_i, n1_i)
    real(4), intent (in) :: normangle
    real(4), intent (in) :: u
    real(4), intent (in) :: n0_i
    real(4), intent (in) :: n1_i
    code = n0_i + (u * (n1_i - n0_i))
end function

Julia

function code(normAngle, u, n0_i, n1_i)
	return Float32(n0_i + Float32(u * Float32(n1_i - n0_i)))
end

MATLAB

function tmp = code(normAngle, u, n0_i, n1_i)
	tmp = n0_i + (u * (n1_i - n0_i));
end

Derivation

1. Initial program 97.6%

   \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
2. Add Preprocessing

3. Taylor expanded in normAngle around 0 98.7%

   \[\leadsto \color{blue}{n0_i \cdot \left(1 - u\right) + n1_i \cdot u} \]

4. Taylor expanded in u around -inf 98.9%

   \[\leadsto \color{blue}{n0_i + -1 \cdot \left(u \cdot \left(n0_i + -1 \cdot n1_i\right)\right)} \]
5. Step-by-step derivation

   1. mul-1-neg 98.9%

      \[\leadsto n0_i + \color{blue}{\left(-u \cdot \left(n0_i + -1 \cdot n1_i\right)\right)} \]

   2. unsub-neg 98.9%

      \[\leadsto \color{blue}{n0_i - u \cdot \left(n0_i + -1 \cdot n1_i\right)} \]

   3. neg-mul-1 98.9%

      \[\leadsto n0_i - u \cdot \left(n0_i + \color{blue}{\left(-n1_i\right)}\right) \]

   4. unsub-neg 98.9%

      \[\leadsto n0_i - u \cdot \color{blue}{\left(n0_i - n1_i\right)} \]

6. Simplified 98.9%

   \[\leadsto \color{blue}{n0_i - u \cdot \left(n0_i - n1_i\right)} \]

7. Final simplification 98.9%

   \[\leadsto n0_i + u \cdot \left(n1_i - n0_i\right) \]
8. Add Preprocessing

Alternative 10: 47.8% accurate, 421.0× speedup

Math

\[\begin{array}{l} \\ n0_i \end{array} \]

FPCore

(FPCore (normAngle u n0_i n1_i) :precision binary32 n0_i)

C

float code(float normAngle, float u, float n0_i, float n1_i) {
	return n0_i;
}

Fortran

real(4) function code(normangle, u, n0_i, n1_i)
    real(4), intent (in) :: normangle
    real(4), intent (in) :: u
    real(4), intent (in) :: n0_i
    real(4), intent (in) :: n1_i
    code = n0_i
end function

Julia

function code(normAngle, u, n0_i, n1_i)
	return n0_i
end

MATLAB

function tmp = code(normAngle, u, n0_i, n1_i)
	tmp = n0_i;
end

Derivation

1. Initial program 97.6%

   \[\left(\sin \left(\left(1 - u\right) \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n0_i + \left(\sin \left(u \cdot normAngle\right) \cdot \frac{1}{\sin normAngle}\right) \cdot n1_i \]
2. Add Preprocessing

3. Taylor expanded in u around 0 45.5%

   \[\leadsto \color{blue}{n0_i} \]

4. Final simplification 45.5%

   \[\leadsto n0_i \]
5. Add Preprocessing

Reproduce

herbie shell --seed 2024017 
(FPCore (normAngle u n0_i n1_i)
  :name "Curve intersection, scale width based on ribbon orientation"
  :precision binary32
  :pre (and (and (and (and (<= 0.0 normAngle) (<= normAngle (/ PI 2.0))) (and (<= -1.0 n0_i) (<= n0_i 1.0))) (and (<= -1.0 n1_i) (<= n1_i 1.0))) (and (<= 2.328306437e-10 u) (<= u 1.0)))
  (+ (* (* (sin (* (- 1.0 u) normAngle)) (/ 1.0 (sin normAngle))) n0_i) (* (* (sin (* u normAngle)) (/ 1.0 (sin normAngle))) n1_i)))