============================================================================== V1. E = (x^2 + y^5, 2), characteristic 2, ambient A^2 f = y^5 + x^2 ord_0 f = 2 in_0 f = x^2 Sing(E) ideal (f, df): [y, x] -> the origin only d_x^(2) f = 1 (a unit: so Diff^(2)(p(E,2)) = O, hence p^nega(E,-a)_0 = O for all a>0) step 1: chart y: f' = y^3 + x^2; ord_0 f' = 2; points of order>=2 over GF(4) on this chart: [(0, 0)] chart x: f' = x^3*y^5 + 1; points of order>=2 over GF(4): [] step 2: chart y: f' = x^2 + y; ord_0 f' = 1; points of order>=2 over GF(4) on this chart: [] chart x: f' = x*y^3 + 1; points of order>=2 over GF(4): [] -> resolved (order < 2). Observation: every point of order 2 of every transform lies on x'=0, the strict transform of {x=0}; the edge data (3-tuple (n, n-r, q)) of the p-algebra at the singular point is the same at every step (the only order-1 element of degree 1 is x), while the residual order ord(eps) drops 5 -> 3 -> 1. The manuscript's Thm 16.6(2) asserts a strict lexicographic drop of the Inv-string at every point of pi^{-1}(nabla)\nabla'. J#|_{x=0} (generators mod x): [y^5, y^10, y^9, y^9, y^8] -> F(Y) = (y^5, 2) on Y = A^1 ============================================================================== V2. Narasimhan's example, characteristic 2, ambient A^4 f = Y^7*W + Y*Z^3 + Z*W^3 + X^2 curve C: (x,y,z,w) = (t^32, t^7, t^19, t^15) f(C) = 0 d_x f = 0; restricted to C: 0; ord_0 = +Infinity d_y f = Y^6*W + Z^3; restricted to C: 0; ord_0 = 3 d_z f = Y*Z^2 + W^3; restricted to C: 0; ord_0 = 3 d_w f = Y^7 + Z*W^2; restricted to C: 0; ord_0 = 3 dimension of the critical scheme {f = df = 0}: 1 I(C) has 13 generators; minimal order of a generator at 0: 2 (no order-1 element: C lies in no smooth hypersurface at 0) semigroup <32,7,19,15>: minimal generators: [32, 7, 19, 15] (all four => embedding dimension 4 at 0) radical(crit) == I(C): True point eta on C over GF(4): (a + 1, a, a, 1) ord_eta f = 2 in_eta f = X^2 + (a + 1)*Y^2 + (a + 1)*Y*Z + (a + 1)*Z^2 + Y*W + Z*W + a*W^2 ord_eta(d_z f) = 1 -> a degree-1 element of p(E,1) of order 1 at eta; so the edge data at eta is (4, <=2, ...) < (4,3,2) = edge data at 0. ============================================================================== V3. Hauser's kangaroo sequence (arXiv:0811.4151, sect. G), characteristic 2, ambient A^3 f0 = y^7 + y*z^4 + x^2 Sing(f0,2) ideal radical: [x, y^3 + z^2] f1 = f0(xy,y,zy)/y^2 = y^3*z^4 + y^5 + x^2 f2 = f1(xz,yz,z)/z^2 = y^5*z^3 + y^3*z^5 + x^2 f3 = f2(xz,yz+z,z)/z^2 = y^5*z^6 + y^4*z^6 + y^3*z^6 + y^2*z^6 + x^2 ord at a0: 2, initial form: x^2 ord at a1: 2, initial form: x^2 ord at a2: 2, initial form: x^2 ord at a3: 2, initial form: x^2 a0: coefficient part w.r.t. x=0 (exceptional monomial removed) = y^7 + y*z^4, residual order = 5 a1: coefficient part w.r.t. x=0 (exceptional monomial removed) = z^4 + y^2, residual order = 2 a2: coefficient part w.r.t. x=0 (exceptional monomial removed) = y^2 + z^2, residual order = 2 a3: coefficient part w.r.t. x=0 (exceptional monomial removed) = y^5 + y^4 + y^3 + y^2, residual order = 2 a3: square part of eps = y^4*z^6 + y^2*z^6 = ( y^2*z^3 + y*z^3 )^2 ; non-square part h = y^5*z^6 + y^3*z^6 manuscript-style tail at a3: x + phi = y^2*z^3 + y*z^3 + x ; hypersurface {x + phi = 0} f3 in coordinates x -> x + phi: y^5*z^6 + y^3*z^6 + x^2 residual order w.r.t. the new hypersurface: 3 (was 2 at a2 w.r.t. {x=0}: INCREASE 2 -> 3) the transform of the old tail {x=0} still contains Sing(f3,2) but has residual order 2 which is NOT the residual order realised by the correct tail. ============================================================================== V4. Ridge (edge data of the tangent cone) at near points, random hypersurfaces, char 2 near points tested: 375; lexicographic increases of (n-r, q) found: 0