\[e^{a \cdot x} - 1\]
Test:
NMSE section 3.5
Bits:
128 bits
Bits error versus a
Bits error versus x
Time: 7.5 s
Input Error: 14.8
Output Error: 0.1
Log: ⚲
Profile: 🕒
\(\begin{cases} {\left(\sqrt[3]{\log \left(e^{e^{a \cdot x} - 1}\right)}\right)}^3 & \text{when } a \cdot x \le -0.0025438569f0 \\ \left(\frac{1}{2} \cdot \left(x \cdot a\right)\right) \cdot \left(x \cdot a\right) + x \cdot a & \text{otherwise} \end{cases}\)

    if (* a x) < -0.0025438569f0

    1. Started with
      \[e^{a \cdot x} - 1\]
      0.2
    2. Using strategy rm
      0.2
    3. Applied add-cube-cbrt to get
      \[\color{red}{e^{a \cdot x} - 1} \leadsto \color{blue}{{\left(\sqrt[3]{e^{a \cdot x} - 1}\right)}^3}\]
      0.2
    4. Using strategy rm
      0.2
    5. Applied add-log-exp to get
      \[{\left(\sqrt[3]{\color{red}{e^{a \cdot x} - 1}}\right)}^3 \leadsto {\left(\sqrt[3]{\color{blue}{\log \left(e^{e^{a \cdot x} - 1}\right)}}\right)}^3\]
      0.2

    if -0.0025438569f0 < (* a x)

    1. Started with
      \[e^{a \cdot x} - 1\]
      21.3
    2. Applied taylor to get
      \[e^{a \cdot x} - 1 \leadsto \left(\frac{1}{2} \cdot \left({a}^2 \cdot {x}^2\right) + \left(1 + a \cdot x\right)\right) - 1\]
      19.9
    3. Taylor expanded around 0 to get
      \[\color{red}{\left(\frac{1}{2} \cdot \left({a}^2 \cdot {x}^2\right) + \left(1 + a \cdot x\right)\right)} - 1 \leadsto \color{blue}{\left(\frac{1}{2} \cdot \left({a}^2 \cdot {x}^2\right) + \left(1 + a \cdot x\right)\right)} - 1\]
      19.9
    4. Applied simplify to get
      \[\color{red}{\left(\frac{1}{2} \cdot \left({a}^2 \cdot {x}^2\right) + \left(1 + a \cdot x\right)\right) - 1} \leadsto \color{blue}{\left(\frac{1}{2} \cdot \left(x \cdot a\right)\right) \cdot \left(x \cdot a\right) + x \cdot a}\]
      0.1
  1. Removed slow pow expressions

Original test:


(lambda ((a default) (x default))
  #:name "NMSE section 3.5"
  (- (exp (* a x)) 1)
  #:target
  (if (< (fabs (* a x)) 1/10) (* (* a x) (+ 1 (+ (/ (* a x) 2) (/ (sqr (* a x)) 6)))) (- (exp (* a x)) 1)))