A biquadratic transformation maps two input variables using terms quadratic in each variable, including x²y², x²y, xy², and constant terms. Applying it to a standard square produces a flexible nonlinear shape, but determining its coefficients requires solving a nine-equation system for each output coordinate. section 8.1.1
The straightforward construction has practical limits:
- Solves two independent nine-equation systems instead of one system of eighteen equations. section 8.1.1
- Becomes harder to scale as polynomial degree rises: bicubic systems require sixteen equations, while biquartic systems require twenty-five. section 8.1.1
- Accumulates floating-point error as symbolic or numerical systems grow, motivating expression optimization with Herbie. section 8.1.1
Coefficient system
A typical output coordinate uses nine coefficients: a11x²y² + a12x²y + a13x² + a14xy² + a15xy + a16x + a17y² + a18y + a19. Substituting nine corresponding input and target points yields the equations solved symbolically. section 8.1.1
Use the figure to inspect how a biquadratic transformation reshapes the standard square. section 8.1.1
