Alphabeta Math
Pipeline-generated
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

✓ 2 results · all verified · 2 also independently AI-judged
Every result on this page is machine-checked by a proof checker and read in full and owner-audited; the judge is an additional, independent cross-model AI review of the proofs; all 2 also cleared it.

One-step ODE Consistency and Local Error

1 · Prerequisites

2 · Summary

Summary

Define the local defect and uniform order of a one-step method, together with the one-step Lipschitz bound used to compare inputs in a compact time-state set. Prove the uniform O(h2) local defect for every admitted start time and step, and the 1+Lh stability bound for forward Euler under explicit C1 component and Jacobian bounds. This page makes no global convergence claim and does not infer higher order from a method's name.

3 · Logical flowchart

4 · Definitions, theorems and proofs

DefinitionDefinition: AI-adaptedProof: Not applicablejudge pass (gpt-6-luna)Open item page →

One-step ODE methods, local defect, uniform order, and step stability

Definition

Let d≥1, let D⊆R×Rd be open, and consider a first-order IVP and solution in the sense of First-order systems, initial value problems, and solutions on intervals. A one-step method is a family of maps Ψh(t,v), defined for admitted times t, positive step sizes h, and states v, whose value is the method's approximation at time t+h from the single input state v at time t. It does not use earlier numerical states.

If y is an exact solution on an interval containing [t,t+h], the local defect is

δh(t;y):=Ψh(t,y(t))−y(t+h).

Let K⊆D be a fixed compact time-state set. The method has uniform local order p on K along y if there are constants C≥0 and h0>0, independent of the admitted starting time t and step 0<h≤h0, such that

∥δh(t;y)∥2≤Chp+1

whenever the exact solution graph on [t,t+h] and the associated method time-state inputs, stages, and outputs under consideration lie in K. Here p is a positive integer and ∥⋅∥2 is the Euclidean norm of The Euclidean inner product ⟨x,y⟩=∑k<nxkyk on Rn.

The method has one-step Lipschitz stability on K with constant L≥0 if

∥Ψh(t,u)−Ψh(t,v)∥2≤(1+Lh)∥u−v∥2

for every admitted step size h, admitted time t, and input states u,v with (t,u),(t,v)∈K. This is a bound on the change in one step; it is separate from absolute or linear stability regions and does not itself assert a global error estimate.

LemmaStatement: AI-adaptedProof: AI-adaptedprecheck passjudge pass (gpt-6-luna)Open item page →

A vector field along an ODE solution has an explicit derivative bound

Statement

Let d≥1, let V⊆R×Rd be open, let f:V→Rd have continuous first partial derivatives, and let y:I→Rd solve y′=f(t,y) on an open interval I. Fix a<b with [a,b]⊂I such that (s,y(s))∈V for every s∈[a,b]. Suppose finite constants Fj,Mt,i,Mij≥0 satisfy for all s∈[a,b]

∣fj(s,y(s))∣≤Fj,∣∂tfi(s,y(s))∣≤Mt,i,∣∂zjfi(s,y(s))∣≤Mij(1≤i,j≤d).

Define gi(s):=fi(s,y(s)) and Bi:=Mt,i+∑j=1dMijFj. Then gi is continuous on [a,b], differentiable on (a,b), and

gi′(s)=∂tfi(s,y(s))+∑j=1d∂zjfi(s,y(s))fj(s,y(s)),∣gi′(s)∣≤Bi.

Facts & Assumptions

Given: The open ODE domain, continuous first partial derivatives, solution y′=f(t,y) on [a,b], and the displayed componentwise bounds.

[F1]

First-order systems, initial value problems, and solutions on intervals: an IVP solution is differentiable, its graph lies in the open domain, and it satisfies y′(s)=f(s,y(s)).

[F2]

If all partial derivatives exist on a neighbourhood and are continuous at a point, then the map is totally differentiable there with Jacobian derivative: continuous first partial derivatives give total differentiability with derivative matrix equal to the Jacobian.

[F4]

The chain rule for total derivatives: D(g∘f)(a)=Dg(f(a))∘Df(a): the derivative of a composition is the composition of the total derivatives.

Proof

1.1givenF1F2F3

By [F2]–[F3], f is continuous on V. Since y′=f(s,y(s)), the derivative of y is continuous on I; hence y is continuously differentiable and the path q(s):=(s,y(s)) has derivative q′(s)=(1,f(s,y(s))).

2.1F1F2F3F4algebra∎

For each i, the component map fi is totally differentiable, so [F4] applied to gi=fi∘q gives the displayed derivative formula. The bounds give ∣gi′(s)∣≤Mt,i+∑j=1dMijFj=Bi for every s∈(a,b). Continuity of gi on [a,b] follows from continuity of f and y.

TheoremStatement: AI-adaptedProof: AI-adaptedprecheck passjudge pass (gpt-6-luna)Open item page →

Forward Euler has uniform local order one and a one-step Lipschitz bound

Statement

Let d≥1, T>0, and let V⊆R×Rd be an open time-state domain whose fibers Vt:={z:(t,z)∈V} are open and convex. Let K⊂V be compact. Let f:V→Rd have continuous first partial derivatives, and suppose y is a solution of y′=f(t,y), y(0)=y0, on an open interval containing [0,T], with graph on [0,T] contained in K. For each local defect under consideration, require the Euler input and output time-state pairs to lie in K; the exact solution segment already lies in K.

For 1≤i,j≤d, assume finite constants Fj,Mt,i,Mij,L≥0 satisfy throughout V

∣fj(t,z)∣≤Fj,∣∂tfi(t,z)∣≤Mt,i,∣∂zjfi(t,z)∣≤Mij,∥Dyf(t,z)w∥2≤L∥w∥2(w∈Rd).

Here Dyf is the matrix with entries ∂zjfi, and the last inequality defines the operator bound used below. The forward Euler map is Ψh(t,v)=v+hf(t,v). Then, with Bi:=Mt,i+∑j=1dMijFj, for every t∈[0,T) and h>0 with t+h≤T whose Euler input and output pairs lie in K, its local defect satisfies

∥δh(t;y)∥2≤h22(∑i=1dBi2)1/2,

so it has uniform local order one on K along y∣[0,T], with h0=T. For every t∈[0,T], every admitted h>0, and every u,v∈Vt,

∥Ψh(t,u)−Ψh(t,v)∥2≤(1+Lh)∥u−v∥2.

This is a local defect and one-step stability result. It does not assert a global error estimate or an order for any other named method.

Facts & Assumptions

Given: The stated open ODE domain with open convex state fibers, continuous first partial derivatives and bounds, the solution through (0,y0) on an open interval containing [0,T], the compact set K containing its graph and the Euler time-state pairs under consideration, and the definition of uniform local order.

[F1]

One-step ODE methods, local defect, uniform order, and step stability: δh(t;y)=Ψh(t,y(t))−y(t+h); uniform local order one means a uniform O(h2) defect; one-step Lipschitz stability is the displayed 1+Lh bound.

[F2]

First-order systems, initial value problems, and solutions on intervals: an IVP has open time-state domain, and a solution is differentiable on an interval, has graph in the domain, and satisfies y′=f(t,y) and its initial value.

[F3]

If all partial derivatives exist on a neighbourhood and are continuous at a point, then the map is totally differentiable there with Jacobian derivative: continuous first partial derivatives give a total derivative whose matrix is the Jacobian of those partials.

[F5]

A convex subset of Rm contains every line segment between two of its points: a convex set contains the segment between each pair of its points; each open convex state fiber has this property.

[F6]

On a convex open set, a uniform bound ∥Df(z)v∥2≤M∥v∥2 implies ∥f(y)−f(x)∥2≤M∥y−x∥2: on an open convex Euclidean set, a uniform derivative bound M implies ∥g(u)−g(v)∥2≤M∥u−v∥2.

[F7]

Every Picard–Lindelöf initial value problem has one maximal solution on an open interval: a continuous vector field locally Lipschitz in the state variable has a unique maximal IVP solution. The given derivative bound and [F3]–[F6] provide that local Lipschitz condition on each state fiber.

[F8]

A vector field along an ODE solution has an explicit derivative bound: along an exact solution, gi(s)=fi(s,y(s)) has derivative bounded by Bi=Mt,i+∑k=1dMikFk.

[F9]

A first-order initial value problem is equivalent to its Volterra integral equation: for any initial time t on the solution interval, the solution satisfies y(t+h)=y(t)+∫tt+hf(s,y(s)) ds.

[F10]

The mean value theorem, as the case g(x)=x of Cauchy's: for f continuous on [a,b] with a<b and differentiable on (a,b) there is c∈(a,b) with f(b)−f(a)=f′(c)(b−a): if a real function is continuous on a closed interval and differentiable in its interior, its endpoint difference is its derivative at an interior point times the interval length.

[F11]

The Euclidean inner product ⟨x,y⟩=∑k<nxkyk on Rn: for x∈Rd, ∥x∥22=∑i=1dxi2.

Proof

1.1givenF2F3F4F5F6F7

By [F3]–[F4], f is totally differentiable and continuous on V, with state derivative matrix Dyf. On each open convex fiber Vt, the derivative bound and [F5]–[F6] make f(t,⋅) L-Lipschitz. Thus the hypotheses of [F7] hold and the IVP has a unique maximal solution; the given solution agrees with it throughout [0,T].

1.2givenF2F3F8

Fix any t∈[0,T) and h>0 with t+h≤T satisfying the stated compact-set condition, and fix a component i. Define gi(s):=fi(t+s,y(t+s)) for 0≤s≤h. The exact path segment lies in K⊂V, so [F8] gives ∣gi′(s)∣≤Bi on (0,h).

2.1step 1.2F10

For each s∈(0,h], [F10] applied to gi on [0,s] yields ∣gi(s)−gi(0)∣≤Bis; at s=0 the same inequality is equality.

3.1F1F9F11step 2.1algebra

Apply [F9] to the IVP restarted at (t,y(t)); then yi(t+h)−yi(t)=∫0hgi(s) ds. Euler's output from the exact state is yi(t)+hgi(0), so δh,i(t;y)=∫0h(gi(0)−gi(s)) ds. Using [step 2.1] gives ∣δh,i(t;y)∣≤∫0hBis ds=Bih2/2. Squaring the component bounds, summing, and using [F11] gives the asserted Euclidean norm estimate. Its constant is independent of t and h, proving uniform local order one under [F1].

4.1F1F3F5F6F12algebra∎

For any admitted t∈[0,T], h>0, and u,v∈Vt, [F3] identifies the derivative of z↦f(t,z) with Dyf(t,z); [F5]–[F6] give ∥f(t,u)−f(t,v)∥2≤L∥u−v∥2. Therefore [F12] gives ∥Ψh(t,u)−Ψh(t,v)∥2≤∥u−v∥2+h∥f(t,u)−f(t,v)∥2≤(1+Lh)∥u−v∥2, which is the one-step stability bound in [F1] for inputs in K.

Remarks

The defect proof is choice-free and uses only the time-along-flow chain rule and the componentwise integral remainder. The Taylor supplier is used by the scalar linear example, not needed to prove this theorem. The global error recurrence in the cited numerical-analysis sources is outside this item.

5 · Examples, counterexamples and false statements

None yet.

Sources