Exact Taylor coefficients used by the executable witness.
Equations
Instances For
The cached coefficient array is the exact Taylor expansion.
The exact first Taylor coefficient.
Equations
- HexRootsMathlib.NKData.c1 p s = (HexRootsMathlib.NKData.coeffs p s).getD 1 (0, 0)
Instances For
Squared modulus of the first Taylor coefficient.
Equations
Instances For
Precision used for the executable reciprocal floor.
Equations
- HexRootsMathlib.NKData.invPrec p s = 8 + max 0 (Hex.Dyadic.ceilLog2 (HexRootsMathlib.NKData.normSq p s))
Instances For
Executable approximate inverse of the first Taylor coefficient.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Taylor coefficient after multiplication by the approximate inverse.
Equations
- HexRootsMathlib.NKData.residual p s k = (HexRootsMathlib.NKData.inverse p s).mul ((HexRootsMathlib.NKData.coeffs p s).getD k (0, 0))
Instances For
Exact sup norm of the constant Newton residual.
Equations
- HexRootsMathlib.NKData.y p s = (HexRootsMathlib.NKData.residual p s 0).lo
Instances For
Exact sup-operator norm of the first-order defect.
Equations
- HexRootsMathlib.NKData.z1 p s = Hex.Dyadic.abs (1 - (HexRootsMathlib.NKData.residual p s 1).1) + Hex.Dyadic.abs (HexRootsMathlib.NKData.residual p s 1).2
Instances For
One step of the running-power fold used by the executable radial bound.
Equations
Instances For
Running-power fold used by the executable radial derivative bound.
Equations
- HexRootsMathlib.NKData.z2Sum p s = (List.foldl (HexRootsMathlib.NKData.z2Step p s) (0, 1) (List.range (HexRootsMathlib.NKData.coeffs p s).size)).1
Instances For
Executable radial derivative-Lipschitz bound.
Equations
Instances For
Sup radius of the certified square.
Equations
Instances For
Exact representation of half the squared sup radius.
Equations
- HexRootsMathlib.NKData.halfRadiusSq s = Dyadic.ofIntWithPrec 1 (2 * s.prec + 1)
Instances For
Decode the Boolean witness into its three strict exact inequalities.
Casting the fold accumulator gives its mathematical partial sum and the next real power of the radial bound.
Real closed form of the executable running-power sum.
The zeroth executable Taylor coefficient is evaluation at the centre.
The first executable Taylor coefficient is derivative evaluation at the centre.
Casting a residual gives multiplication of the approximate inverse by the corresponding exact Taylor coefficient.
The sup-norm centre corresponding to the executable square centre.
Equations
Instances For
Complex scalar used as the executable approximate inverse.
Equations
Instances For
Approximate inverse as a real-linear operator in sup coordinates.
Equations
Instances For
Evaluation at the centre is the zeroth executable Taylor coefficient.
Multiplying an executable Taylor coefficient by the approximate inverse gives the executable residual.
The first Newton quantity is exact: no Euclidean norm conversion enters.
The approximate inverse composed with the frozen centre derivative is multiplication by the first executable residual.
The first-order defect bound is the exact sup operator norm.
A nonempty executable polynomial has complex degree below its stored coefficient count.
The exact Taylor shift at the square centre.
Equations
Instances For
Coefficients of the recentred polynomial are the executable Taylor coefficients.
The recentred derivative evaluates the original derivative at the translated point.
Radial Taylor identity for the derivative difference. The index is k,
not the mean-value overestimate k * (k - 1).
The derivative difference after applying the executable approximate inverse, still as an exact finite Taylor sum.
One radial Taylor term has the executable bound. There is one √2 from
the scalar operator norm and one from complex modulus versus sup norm; they
combine to the factor 2.
Complex displacement from the executable square centre.
Equations
Instances For
The complex displacement is at most √2 times the sup-norm
displacement from the centre.
Inside the half-width sup ball, the complex displacement stays below the executable upper radius bound.
Applying the approximate inverse to the derivative difference gives the
finite sum whose terms are bounded by z2.
The executable z2 is a radial Lipschitz bound for the transported
polynomial derivative on the certified sup ball.
The executable Lipschitz bound z2 is nonnegative.
Nonnegative Newton quantities in the format consumed by the generic theorem.
Equations
Instances For
The square's half-width as a nonnegative real.
Equations
Instances For
The strict radius and contraction inequalities encoded by the executable witness.
The generic Newton--Kantorovich theorem applied to the exact executable quantities, before transporting the result back from sup coordinates.
Closed-square membership is the sup-norm closed-ball inequality in the two-coordinate model.
Open-square membership is the strict sup-norm ball inequality in the two-coordinate model.
An executable Newton witness certifies exactly one root in the closed dyadic square.
Throughout the certified square, the approximate-inverse derivative defect has norm strictly below one.
Every root in a Newton-certified square is simple.
The strict radius check places every root in the certified closed square strictly inside that square.
Combined semantic contract of an executable Newton witness.