How strongly can a conformal map distort area

← All digests

Family 072 · Real and complex analysis
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-14

1. The question

A conformal map preserves angles, but it can stretch or compress lengths dramatically near a rough boundary. How much distortion can be concentrated there before the average becomes infinite?

Brennan’s conjecture asks for the exact range of powers of the derivative that remain integrable when an arbitrary simply connected plane domain is mapped onto the disk. The boundary may be fractal and the domain may be unbounded. The question is whether complex analyticity still imposes a universal limit on the total concentration of distortion.

The first manuscript in family 072 claims the full conjectured open range. A second manuscript studies a finer boundary average and claims a strict inequality contradicting Kraetzer’s predicted integral-means spectrum. These outcomes point in different directions: one confirms a classical integrability prediction, while the other rules out a more detailed conjectural formula.

2. Background and essential definitions

Derivatives, area and inverse maps

Let $\mathbb D=\{z\in\mathbb C:|z|<1\}$. A univalent map is an injective holomorphic function. Its derivative never vanishes. Locally, multiplication by $f'(z)$ scales lengths by $|f'(z)|$ and areas by $|f'(z)|^2$.

Suppose $\phi:W\to\mathbb D$ is a conformal bijection and $f=\phi^{-1}$. Changing variables gives

$$\int_W |\phi'(z)|^s\,dA(z) =\int_{\mathbb D}|f'(w)|^{2-s}\,dA(w),$$

where $dA$ is ordinary planar area. The factor $|f'|^2$ is the area Jacobian, while the inverse derivative contributes $|f'|^{-s}$. Thus powers of a derivative on a variable domain become powers of an inverse map on one fixed disk.

Write $t=2-s$. Positive $t$ emphasizes places where $f'$ is large; negative $t$ emphasizes places where it is small. The latter is especially important near the upper endpoint of Brennan’s proposed range.

Boundary averages and their growth exponents

The normalized schlicht class is

$$\mathcal S=\{f:\mathbb D\to\mathbb C\text{ univalent}:f(0)=0,\ f'(0)=1\}.$$

Normalization removes translations and changes of scale. For real $t$, define the radial integral mean

$$M_t[f'](r)=\frac1{2\pi}\int_{-\pi}^{\pi}|f'(re^{i\theta})|^t\,d\theta.$$

Its growth exponent records how quickly this average can increase as $r$ approaches one:

$$\beta_f(t)=\limsup_{r\uparrow1} \frac{\log M_t[f'](r)}{\log(1/(1-r))}.$$

The universal spectrum $B_{\mathcal S}(t)$ is the supremum of these exponents over $\mathcal S$. The bounded spectrum $B_b(t)$ uses only maps with bounded image. The image bound may depend on the map. These two classes must be distinguished, especially for positive powers.

An estimate $M_t(r)\le C(1-r)^{-b}$ with $b<1$ implies area integrability near the circle: the radial integral of $(1-r)^{-b}$ is finite. The exponent one is the threshold at which this simple conclusion stops working.

3. Development history and previous partial results

Brennan’s 1978 paper developed the derivative-integrability problem and pushed its upper range beyond the classical value three. The conjecture asks for all $4/3<s<4$. A slit-plane example shows why one should not expect the endpoints themselves.

Later work by Pommerenke, Carleson and Makarov connected area integrability with radial integral means and the distribution of harmonic measure. In particular, the linear formula for the spectrum was known sufficiently far into the negative-power range. Reaching its proposed transition at $t=-2$ remained a sharper issue.

Bertilsson’s 1999 thesis reported the range $4/3<s<3.421$ and developed equivalent formulations of the inverse-square endpoint. Hedenmalm and Shimorin used weighted Bergman spaces, Hilbert spaces of analytic functions with weighted area norms, to improve bounds for the spectrum. The present manuscript recalls their upper bound $B_{\mathcal S}(-2)\le1.218$, followed by Sola’s bound of $1.206$. Both still exceeded the target value one.

Special families of conformal maps also admit stronger results than the unrestricted class. Examples discussed in the manuscript include certain holomorphic semigroups, polynomial basins and rational univalent functions. Success on such families does not automatically cover every possible boundary geometry.

A separate conjectural development came from Kraetzer’s 1996 numerical experiments. They suggested a quadratic formula $B_b(p)=p^2/4$ for $|p|\le2$, joined to a linear tail outside that interval. At $p=-1$, the predicted value is $1/4$. The companion manuscript claims an upper bound strictly smaller than that value.

4. What the new manuscripts claim

Brennan’s full open interval

The first paper states that every conformal bijection $\phi:W\to\mathbb D$ from a simply connected plane domain satisfies

$$\int_W |\phi'(z)|^s\,dA(z)<\infty \qquad\text{for every }\frac43<s<4.$$

No boundedness or boundary regularity is assumed. The existence of the conformal bijection excludes degenerate cases such as the whole plane. Equivalently, every univalent disk map satisfies

$$\int_{\mathbb D}|f'(w)|^t\,dA(w)<\infty \qquad\text{for }-2<t<\frac23.$$

The central radial claim is uniform over normalized maps: for every $\varepsilon>0$, there is $C_\varepsilon<\infty$ such that

$$M_{-2}[f'](r)\le C_\varepsilon(1-r)^{-1-\varepsilon} \quad(f\in\mathcal S,\ 1/2\le r<1).$$

It states $B_{\mathcal S}(-2)=1$, and derives $B_{\mathcal S}(t)=|t|-1$ for every $t\le-2$. The arbitrary positive loss in the radial estimate does not assert integrability at the endpoint.

A strict failure of the predicted spectrum

The independent companion states that some constants $0<\eta<1/4$ and $C<\infty$ satisfy

$$M_{-1}[f'](r)\le C(1-r)^{-1/4+\eta} \quad(f\in\mathcal S,\ 1/2\le r<1).$$

Consequently $B_b(-1)<1/4$, contradicting Kraetzer’s formula at that parameter. The bound actually applies to the unrestricted normalized class. The paper asserts existence of a positive gap without giving its numerical size. It does not settle the prediction at $p=1$.

5. Why the advance matters

Importance: 5/5. The stated Brennan theorem settles a central extremal problem in geometric function theory for arbitrary simply connected domains. Its range quantifies how much boundary irregularity a conformal change of coordinates can tolerate while preserving integral control.

The companion would also change the expected shape of a universal spectrum used to organize derivative growth and boundary geometry. Knowing that a plausible formula is strictly too large at one parameter is more informative than another upper bound approaching it from above: it changes the target that future sharp estimates must reach.

These claims are related, but the second is not merely a restatement of the first. Interpolation from the inverse-square estimate yields a much weaker inverse-first-power estimate than the strict $1/4$ threshold. The manuscripts develop related machinery for two separate conclusions.

6. The main idea

Replace exceptional maps by a distribution of maps

Pointwise distortion estimates control how small a derivative can be at a single point. They do not adequately control how frequently very small derivatives occur around a nearly boundary circle. Integral means require this second kind of information.

The papers move from the disk to the upper half-plane and normalize each map at a chosen interior point. Translating and rescaling the source produces another normalized map. Averaging two such changes of base point defines a positive operator: it sends nonnegative functions of maps to nonnegative functions.

Repeated application samples successively smaller scales. Its growth rate therefore summarizes the possible growth of the relevant inverse-derivative means. If that rate were too large, the manuscripts construct a probability distribution on normalized maps with an exact scaling identity. This distribution is an analytical device, not an assumption that the original conformal domain is random.

Compare saddles with maxima

For a target point $\xi$, the arguments introduce potentials of the form

$$V_\xi(x+iy)=\frac{y^k}{|F(x+iy)-\xi|},$$

with $k$ chosen from the hypothetical growth exponent. The potential combines distance from the real boundary with distance to a target in the image.

Critical points of this potential can be maxima or saddles. A planar index argument is used to pair each relevant saddle with a maximum at least as high. The pairing is designed to behave consistently under the source translations and rescalings used by the probability law.

The exact scaling identity gives one sign for an averaged Jacobian, a signed local area-change factor. Transporting weights between the paired critical points gives the opposite sign when the assumed growth is excessive. That contradiction is the papers’ proposed route to the upper bounds.

The companion must do more than exclude exponents above $1/4$. It also excludes equality. Its stated mechanism uses strict inequality between the values at paired critical points for almost every target, followed by a separate argument ruling out the resulting equality case. A uniform numerical improvement for each individual pair is not claimed or needed for this description.

7. Scope, limitations, and open questions

The open endpoints are essential. The Koebe function

$$f(z)=\frac{z}{(1-z)^2}$$

has divergent area integrals at both $t=-2$ and $t=2/3$. Its inverse gives failure at $s=4$ and $s=4/3$. Thus the claimed theorem does not conceal an unproved endpoint extension.

The ordinary area integral and the radial growth exponent answer different questions. An exact exponent one can coexist with a divergent area integral. Likewise, a universal spectrum takes a supremum after the limiting exponent for each individual map; a uniform estimate over all maps is an additional form of control that the manuscripts explicitly state.

The strict inverse-first-power theorem does not determine the exact value of $B_b(-1)$, the full bounded spectrum, or its positive-power counterpart. The two papers also do not assert general derivative integrability for arbitrary multiply connected domains. Their common setting is injective holomorphic maps of the disk and their inverses.

8. Sources and further reading

Brennan’s original paper introduces the integrability problem. Hedenmalm and Shimorin develop the weighted Bergman-space approach. Kraetzer’s paper records the experimental prediction addressed by the companion. The historical numerical bounds above are reported in the source manuscripts and their cited thesis literature.

For optional background, see univalent functions and the Koebe function and distortion theorems.

Source manuscripts

• M1. OpenAI, Brennan’s conjecture and sharp inverse-square integral means (September 24, 2026).
• M2. OpenAI, A strict inverse-first-power bound for univalent functions (September 24, 2026).

Sources consulted on 14 October 2026 at repository revision 1581d4c. Download the manuscript citations.