---
title: Laplace Sequences of Q-Nets
url: https://www.emergentmind.com/topics/laplace-sequences-of-q-nets
type: topic
---

# Laplace Sequences of Q-Nets

In projective discrete differential geometry, a Laplace sequence of a \(Q\)-net is the sequence obtained by iterating the two discrete Laplace transformations of a planar quadrilateral net. A \(Q\)-net is a map from \(\mathbb Z^2\) or a finite rectangular patch into projective space such that every elementary quadrilateral is planar; for a generic \(Q\)-net, the Laplace sequence is bi-infinite, whereas special degenerations collapse an iterate to a discrete curve or a point and thereby terminate the sequence. Recent work has treated finite Laplace sequences for discrete Koenigs nets, terminating Laplace sequences for \(Q\)-nets inscribed in quadrics and for circular nets with spherical parameter lines, and periodic Laplace cycles of period four as distinct but closely related manifestations of the same discrete projective mechanism [2508.02851][2312.04341][1104.3689].

## 1. Basic projective framework

A \(Q\)-net is a map
\[
P:\Sigma\to \mathbb{RP}^n,
\]
where \(\Sigma\) is either \(\mathbb Z^2\) or a finite rectangular patch \(\Sigma_{a,b}\), such that the image of each unit square is contained in a plane. In the non-degenerate setting, adjacent vertices are distinct and any three vertices of any face span a plane, so each elementary quadrilateral is a genuine planar quadrilateral [2508.02851].

For a non-degenerate \(Q\)-net, the two Laplace transforms are defined by opposite-edge intersections inside each face:
\[
\mathcal{L}_{+}P(i,j) := (P(i,j)\vee P(i,j+1))\cap (P(i+1,j)\vee P(i+1,j+1)),
\]
\[
\mathcal{L}_{-}P(i,j):= (P(i,j) \vee P(i+1,j))\cap (P(i,j+1) \vee P(i+1,j+1)).
\]
Because the four vertices of a face are coplanar, the relevant lines intersect. These transforms are again \(Q\)-nets and satisfy
\[
\mathcal{L}_+ \circ \mathcal{L}_- P(i,j) = \mathcal{L}_- \circ \mathcal{L}_+ P(i,j) = P(i+1,j+1).
\]
Thus the nontrivial dynamics comes from repeated transforms in the same direction rather than alternating directions.

Writing
\[
P_m := (\mathcal L_+)^m P,\qquad P_{-m}:=(\mathcal L_-)^m P,\qquad m\ge 0,
\]
one obtains the Laplace sequence
\[
\ldots \leftarrow P_{-3}\leftarrow P_{-2}\leftarrow P_{-1}\leftarrow P \rightarrow P_1\rightarrow P_2\rightarrow P_3\rightarrow \ldots
\]
For a generic \(Q\)-net on \(\mathbb Z^2\), all iterates exist and the sequence is bi-infinite. This generic bi-infiniteness is the baseline against which finite, terminating, and periodic cases are studied.

## 2. Degeneracy and termination

Termination occurs when some iterated transform ceases to be a genuine two-parameter net and collapses to a one-parameter object. In the formulation used for discrete Koenigs nets, \(P_m\) is Laplace degenerate if \(P_m(i,j)\) is independent of \(i\) for all \(j\), and \(P_{-m}\) is Laplace degenerate if \(P_{-m}(i,j)\) is independent of \(j\) for all \(i\). The transform has then become a discrete curve. Goursat degeneracy is the complementary curve-type degeneration: \(P_m\) is Goursat degenerate if it is independent of \(j\) for all \(i\) and is nowhere Laplace degenerate, with the analogous reversed condition for \(P_{-m}\). The genericity clause excludes mixed-type overlap and makes Laplace and Goursat degeneration distinct notions [2508.02851].

A basic algebraic diagnostic is provided by the Laplace invariants
\[
H(i,j):= \cro(P(i,j), P_{1}(i,j), P(i,j+1), P_{1}(i-1,j)),
\]
\[
K(i,j) := \cro(P(i,j), P_{-1}(i,j) ,P(i+1,j) , P_{-1}(i,j-1)).
\]
They satisfy the shift identities
\[
K_1(i,j)=H(i+1,j),\qquad H_{-1}(i,j)=K(i,j+1),
\]
and Doliwa’s recurrence
\[
H_1(i,j) = H_{-1}^{-1}(i,j) \frac{1-H(i+1,j)}{1-H^{-1}(i,j)} \frac{1-H(i,j+1)}{1-H^{-1}(i+1,j+1)}.
\]
A particularly useful criterion is
\[
P_1 \text{ is Laplace degenerate } \iff H(i,j)=1 \text{ for all } i,j,
\]
\[
P_{-1} \text{ is Laplace degenerate } \iff K(i,j)=1 \text{ for all } i,j.
\]
This makes first-step termination visible directly at the invariant level.

Higher-step degeneration also has an intrinsic projective description. For an extensive \(Q\)-net with \(P_m\) existing,
\[
P_m(0,0)=\bigcap_{k=0}^{m}\bigvee_{\ell=0}^{m} P(k,\ell),
\]
and more generally Laplace degeneracy after \(m\) steps is characterized by the intersections
\[
\bigcap_{k=i}^{i+m}\bigvee_{\ell=j}^{j+m} P(k,\ell)
\]
being points independent of \(i\). If that intersection is a \(d\)-dimensional projective subspace independent of \(i\), then \(P_{m-d}\) is Laplace degenerate. Goursat degeneracy is encoded by the dimensions of the vertical parameter spaces \(\underline P(i)\): if \(P_m\) exists and is nowhere Laplace degenerate, then
\[
P_m \text{ is Goursat degenerate } \iff \underline{P}(i)\text{ are }m\text{-dimensional.}
\]

A recurrent misconception is that termination in one direction should automatically imply termination in the other for arbitrary \(Q\)-nets. The general theory does not support that expectation. Finite behavior requires additional structure, and the strongest available results arise precisely for special classes such as discrete Koenigs nets and \(Q\)-nets inscribed in quadrics.

## 3. Koenigs nets and the diagonal-intersection correspondence

Two standard discrete versions of Koenigs nets organize the finite-sequence theory. A non-degenerate \(Q\)-net \(P\) is a BS-Koenigs net if its Laplace invariants satisfy
\[
H(i,j)\,H(i,j+1)=K(i,j+1)\,K(i-1,j+1),
\]
and a non-degenerate \(Q\)-net \(D\) is a D-Koenigs net if
\[
H(i,j)\,H(i+1,j)=K(i,j)\,K(i,j+1).
\]
These are the Bobenko–Suris and Doliwa discretizations, respectively [2508.02851].

The bridge between them is the diagonal intersection net
\[
D(i,j)= \big(P(i,j)\vee P(i+1,j+1)\big)\cap \big(P(i+1,j)\vee P(i,j+1)\big).
\]
If \(P\) is a BS-Koenigs net, then \(D\) is a \(Q\)-net, and if \(D\) is non-degenerate then \(D\) is a D-Koenigs net. This relation is not merely formal; it is the main symmetry device in the theory.

For BS-Koenigs nets there is also a hyperplane characterization: for an extensive \(P:\Sigma_{a,b}\to\mathbb{RP}^{a+b}\), there exist two distinct hyperplanes \(U_1,U_2\) such that
\[
P(i,j)\in 
\begin{cases}
U_1,& i+j\in 2\mathbb Z,\\
U_2,& i+j\in 2\mathbb Z+1.
\end{cases}
\]
Conversely, for an extensive \(Q\)-net this two-hyperplane condition implies the BS-Koenigs property. The lifted net therefore lies on the degenerate quadric
\[
\mathcal U:=U_1\cup U_2.
\]

The diagonal-intersection construction reverses Laplace-invariant data along the sequence. If \(P\) is BS-Koenigs and \(D\) is its diagonal intersection net, then
\[
H^{P}_{m}(i+1,j) = K^{D}_{-m}(i,j),\qquad H^{D}_{m}(i,j) = K^{P}_{-m}(i,j+1),
\]
whenever the relevant transforms exist. Consequently,
\[
P_m \text{ Laplace degenerate } \iff D_{-m}\text{ Laplace degenerate},
\]
and if \(P_m\) is Goursat degenerate then
\[
D_{-m-1}\text{ is Laplace degenerate.}
\]
This diagonal symmetry is the most compact formulation of the finite-sequence phenomenon for discrete Koenigs geometry.

## 4. Finite Laplace sequences

The central finite-length theorem states that if \(P:\Sigma\to\mathbb{RP}^n\) is either a BS-Koenigs net or a D-Koenigs net, then
\[
P_m \text{ Laplace degenerate } \Longrightarrow P_{-m-1}\text{ Laplace degenerate}
\]
and
\[
P_m \text{ Goursat degenerate } \Longrightarrow P_{-m-2}\text{ Laplace degenerate},
\]
assuming the opposite-side transforms exist [2508.02851]. One-sided termination therefore forces opposite-side termination after a controlled shift. In particular, for discrete Koenigs nets, termination implies finiteness.

For BS-Koenigs nets the result is sharpened by compatibility with the diagonal intersection net. If \(P_m\) is Laplace degenerate and \(P_{-m-1},D_{-m}\) exist, then both \(P_{-m-1}\) and \(D_{-m}\) are Laplace degenerate and
\[
P_{-m-1}(i)=D_{-m}(i)\qquad\forall i.
\]
For D-Koenigs nets the same implications follow by viewing every D-Koenigs net as essentially the diagonal intersection net of some BS-Koenigs net.

Two proof mechanisms coexist. The first is invariant-theoretic: Laplace degeneracy is detected by \(H=1\) or \(K=1\), and the symmetry between \(P\) and its diagonal intersection net turns forward degeneracy of one net into backward degeneracy of the other. The second is geometric: extensive BS-Koenigs nets lie on the degenerate quadric \(\mathcal U=U_1\cup U_2\), and a theorem on \(Q\)-nets inscribed in quadrics is applied to this degenerate setting. In the Laplace-degenerate case, polar-incidence constraints force the backward transform into the singular locus of \(\mathcal U\), which collapses it to a curve. In the Goursat-degenerate case, a suitable lift \(\hat P\) satisfies
\[
P_1 \text{ Goursat degenerate } \Longrightarrow \hat P_2 \text{ Laplace degenerate,}
\]
so Goursat degeneration behaves like a hidden Laplace degeneration one step later.

The finite theorem also clarifies the discrete departure from the smooth theory. In the smooth Koenigs setting, opposite-side termination classically occurs after the same number of steps. In the discrete setting, the generic shift is \(m\mapsto m+1\) for Laplace degeneracy and \(m\mapsto m+2\) for Goursat degeneracy. The paper also shows that this shift is generic rather than universal: for \(m\ge 2\) there exist BS-Koenigs nets such that
\[
P_m \text{ and } P_{-m}
\]
are both Laplace degenerate. A plausible implication is that the extra shift reflects discretization asymmetry rather than an unavoidable obstruction.

## 5. Quadrics, circular nets, and geometric realizations

A broader termination theory arises for \(Q\)-nets inscribed in quadrics. In one formulation, for a finite \(m\times m\) \(Q\)-net in \(\mathbb{RP}^n\) with iterated Laplace transforms well defined up to order \(m-1\), the terminal Laplace points \(\mathcal L_A^{m-1}P\) and \(\mathcal L_B^{m-1}P\) are points, and if all vertices except possibly \(P_{m,m}\) lie in a quadric \(\mathcal Q\), then
\[
P_{m,m}\in \mathcal Q \quad\Longleftrightarrow\quad \mathcal L_A^{m-1}P \text{ and } \mathcal L_B^{m-1}P \text{ are conjugate with respect to }\mathcal Q.
\]
This quadric-conjugacy principle is the main incidence tool in the projective theory of terminating sequences [2312.04341].

For \(Q\)-nets lying in a quadric, the two Laplace directions become coupled. If \(P:\mathbb Z^2\to \mathcal Q\subset \mathrm P^n\) is inscribed in a quadric and all lines \(P_{i,j}\vee P_{i+1,j}\) are not isotropic, then
\[
\mathcal L_A^mP \text{ Goursat degenerate } \Longrightarrow \mathcal L_B^mP \text{ Laplace degenerate},
\]
provided \(\mathcal L_B^dP\) exists for all \(d\le m\). If \(\mathcal L_A^mP\) is Laplace degenerate and \(\mathcal L_B^dP\) exists for all \(d\le m+n-1\), then
\[
\mathcal L_B^{m+n-1}P \text{ is Goursat degenerate.}
\]
Thus quadric inscription produces a discrete analogue of the classical Goursat principle: one-sided termination forces opposite-side termination, though not always at the same order.

Circular nets with spherical, circular, planar, or linear parameter lines give concrete geometric realizations of this mechanism through Möbius lifts. A circular net \(P:\mathbb Z^2\to \mathbb R^n\) lifts to a \(Q\)-net
\[
M:\mathbb Z^2\to \mathcal M^n\subset \mathrm P^{n+1}
\]
in the Möbius quadric. Then spherical or planar families of parameter lines become low-dimensional projective-span constraints. If the lifted parameter lines are \(m\)-spherical, then generically
\[
\mathcal L_A^{m+1}M \text{ is Goursat degenerate},\qquad
\mathcal L_B^{m+1}M \text{ is Laplace degenerate}.
\]
For planar parameter lines,
\[
\mathcal L_A^mP \text{ is Goursat degenerate},\qquad
\mathcal L_B^{m+1}P \text{ is Laplace degenerate},
\]
and the terminal transform lies in the hyperplane at infinity. When both parameter directions lie in \(d\)-dimensional projective subspaces of a non-degenerate quadric, \(\mathcal L_A^dP\) and \(\mathcal L_B^dP\) are both Laplace degenerate and Goursat degenerate, hence each collapses to a point.

These examples show that terminating Laplace sequences are not exceptional curiosities but encode concrete surface classes. In the Möbius setting the terminal opposite transforms correspond to polar data, often interpreted as common orthogonal spheres or hyperspheres. The same framework also explains Darboux cyclide geometry through pencils of quadrics. The later Lie-geometric reformulation adds an extra concurrency condition and recovers termination orders more faithful to the smooth theory, which suggests that the raw projective mechanism is necessary but not always sufficient to capture the intended differential-geometric limit.

## 6. Periodicity and adjacent sequence-type constructions

Finite termination is only one closure pattern. Another is periodicity. A discrete Laplace cycle of period four is defined by
\[
\mathcal L_{11}f=\mathcal L_{22}f,
\]
for a regular discrete conjugate net \(f:\mathbb Z^2\to\mathbb P^3\). With
\[
h=\mathcal L_1 f,\qquad g=\mathcal L_{11}f=\mathcal L_{22}f,\qquad k=\mathcal L_2 f,
\]
both Laplace sequences become 4-periodic:
\[
\mathcal L_1^{l+4}f=\mathcal L_1^l f,\qquad \mathcal L_2^{l+4}f=\mathcal L_2^l f.
\]
Opposite nets in the cycle satisfy
\[
\mathcal O_1 f=\mathcal O_2 g,\qquad \mathcal O_2 f=\mathcal O_1 g,
\]
so they are asymptotically related, and their connecting lines form the diagonal congruences
\[
K_i^j=f_i^j\vee g_i^j,\qquad L_i^j=h_i^j\vee k_i^j.
\]
A fundamental theorem states that if two nets are asymptotic transforms of each other, their common axis congruence is a \(W\)-congruence; hence in a period-four Laplace cycle both diagonal congruences are \(W\)-congruences [1104.3689].

This periodic case shows that closure of Laplace sequences need not occur by degeneration alone. It can also arise through a nontrivial cycle in which successive transforms remain genuine \(Q\)-nets but return after finitely many steps. The period-four theory gives a synthetic projective interpretation of this closure by linking Laplace transforms, osculating planes, asymptotic transforms, and line geometry on the Plücker quadric.

A related but distinct development appears in the theory of reduced \(Q\)-nets and generalized pentagram maps. There the paper explicitly states that it does not define a Laplace transformation or use the phrase “Laplace sequence,” but the row-by-row evolution
\[
(v^1,v^2)\mapsto (v^2,v^3)
\]
of a reduced \(Q\)-net is the closest analogous sequence-type construction in that setting. The next row is obtained by triple intersections of planes, and the resulting dynamics admits a refactorization description, a Lax form with spectral parameter, and invariant Poisson brackets [2412.08202]. This suggests that Laplace-sequence ideas extend beyond the classical opposite-edge-intersection framework, though in that literature the analogy is explicit rather than terminological.

Taken together, finite terminating sequences, period-four cycles, and adjacent row-propagation dynamics show that Laplace sequences of \(Q\)-nets are best understood as a family of discrete projective evolutions whose special behavior is controlled by additional structure: Koenigs constraints, quadric inscription, Möbius or Lie lifts, or periodic closure conditions.

Source: https://www.emergentmind.com/topics/laplace-sequences-of-q-nets