Deep Hole Problem in Lattices & Black Holes

• Deep Hole Problem is a dual-context challenge found in both lattice theory and supersymmetric black-hole microstate geometry, focusing on extremal points and smooth capping.
• In lattice theory, it involves identifying points farthest from a lattice, using a distinguished deep hole to generate a new lattice via iterative geometric operations.
• In the black-hole context, it addresses constructing smooth, horizonless supergravity solutions that replicate the long throat of macroscopic supersymmetric black holes.

The phrase Deep Hole Problem has distinct technical meanings in different research areas. In lattice theory, it concerns points farthest from a lattice and, in the planar case, can be reformulated through a canonical transformation that sends a lattice $L$ to an affiliated deep hole lattice $H(L)$ generated by a shortest vector and a distinguished deep hole. In supersymmetric black-hole microstate geometry, the same phrase is used for the problem of constructing smooth horizonless supergravity solutions that have the same asymptotic conserved charges as a macroscopic black hole while lying deep inside the black-hole regime, meaning both that the classical black hole exists and that the microstate reproduces its throat over a parametrically large region before capping off smoothly (Fukshansky et al., 2023, Bena et al., 2016).

1. Terminological scope and problem statements

In the planar lattice setting, a deep hole of a lattice $L\subset \mathbb R^2$ is a point of $\mathbb R^2$ that is farthest from the lattice, and the common distance from a deep hole to its nearest lattice points is the covering radius $\mu$. The relevant problem is not merely to identify deep holes, but to use a distinguished one to define a new lattice

$$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$

where $\mathbf x_1$ is a shortest vector and $\mathbf z$ is the fundamental deep hole, the unique deep hole contained in the triangle with vertices $\mathbf 0,\mathbf x_1,\mathbf x_2$ for shortest basis vectors $\mathbf x_1,\mathbf x_2$. This converts the geometric notion of a deep hole into an explicit operation on similarity classes of planar lattices (Fukshansky et al., 2023).

In the black-hole setting, the problem is formulated for supersymmetric rotating D1-D5-P black holes in five dimensions. The central question is whether one can construct smooth, horizonless supergravity solutions that have the same asymptotic conserved charges as a macroscopic supersymmetric black hole and lie deep inside the black-hole regime, namely a region of charge and angular-momentum space where the corresponding classical black hole has a large regular horizon and the microstate geometry reproduces its throat over a parametrically large region. The relevant classical regime is characterized by the cosmic censorship inequality

$H(L)$0

In this context, “deep inside” also has a geometric meaning: the solution should be well approximated by the black-hole geometry over a long throat region and differ only very near where the black-hole singularity or horizon would otherwise appear (Bena et al., 2016).

A common misconception is that the phrase denotes a single standard problem across disciplines. The cited literature instead uses it in two sharply different senses: one geometric-arithmetic and discrete, the other supergravity-based and holographic.

2. Planar lattices, deep holes, and the affiliated lattice

Let $H(L)$1 be a lattice with successive minima $H(L)$2 and corresponding minimal basis vectors $H(L)$3. By changing signs if necessary, one can ensure that the angle $H(L)$4 between $H(L)$5 lies in

$H(L)$6

This angle is an invariant of the lattice and is called the angle of $H(L)$7. The lattice is well-rounded (WR) if $H(L)$8, and semi-stable if $H(L)$9. For planar lattices, WR implies semi-stable. Two lattices $L\subset \mathbb R^2$0 are similar, written $L\subset \mathbb R^2$1, if

$L\subset \mathbb R^2$2

for some $L\subset \mathbb R^2$3 and $L\subset \mathbb R^2$4 (Fukshansky et al., 2023).

Every planar lattice is similar to a unique lattice

$L\subset \mathbb R^2$5

with

$L\subset \mathbb R^2$6

Thus $L\subset \mathbb R^2$7 parameterizes similarity classes of planar lattices. In this model, WR similarity classes correspond to

$L\subset \mathbb R^2$8

and semi-stable similarity classes correspond to

$L\subset \mathbb R^2$9

This gives a moduli-theoretic description of the planar problem in terms of the upper half-plane (Fukshansky et al., 2023).

The planar theory is tractable because the distinguished deep hole is geometrically explicit. If $\mathbb R^2$0 is the triangle with vertices

$\mathbb R^2$1

then there is a unique deep hole $\mathbb R^2$2 of $\mathbb R^2$3 contained in $\mathbb R^2$4, and $\mathbb R^2$5 is the center of the circumcircle of $\mathbb R^2$6. Accordingly,

$\mathbb R^2$7

The covering radius is

$\mathbb R^2$8

If $\mathbb R^2$9 is the basis matrix of $\mu$0, then

$\mu$1

For $\mu$2 with $\mu$3, $\mu$4, $\mu$5, the fundamental deep hole is

$\mu$6

and hence

$\mu$7

This formula is the basic recursion underlying the deep-hole sequence (Fukshansky et al., 2023).

3. Dynamics of the planar deep-hole operator

The deep-hole operator has a precise geometric structure in dimension $\mu$8. If $\mu$9 is a lattice in the plane with angle $$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$0 and successive minima $$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$1, $$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$2, $$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$3, then the first principal theorem gives four basic properties: a criterion for $$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$4 to be WR, the fact that semi-stability implies that one deep-hole step yields a WR lattice, the fact that WR lattices are fixed up to similarity, and preservation of the field of definition (Fukshansky et al., 2023).

The exact sufficient criterion is

$$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$5

which implies that $$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$6 is WR. This arises from the angle $$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$7 between $$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$8 and $$H(L):=\operatorname{span}_{\mathbb Z}\{\mathbf x_1,\mathbf z\},$$9. Since $\mathbf x_1$0 is the circumcenter of $\mathbf x_1$1, the triangle with sides $\mathbf x_1$2 is isosceles, and one gets

$\mathbf x_1$3

If $\mathbf x_1$4, then $\mathbf x_1$5 is well-rounded, because $\mathbf x_1$6 and $\mathbf x_1$7 have equal length and form a minimal basis. After rearrangement, this leads exactly to the inequality above (Fukshansky et al., 2023).

A particularly strong specialization is the semi-stable case. If $\mathbf x_1$8 with

$\mathbf x_1$9

and $\mathbf z$0 is semi-stable, then

$\mathbf z$1

so the criterion automatically holds. Hence, if $\mathbf z$2 is semi-stable, then $\mathbf z$3 is WR (Fukshansky et al., 2023).

The operator also has a fixed-point description. If $\mathbf z$4 is WR, then $\mathbf z$5, and the proof shows that the angle of $\mathbf z$6 agrees with the angle of $\mathbf z$7; therefore $\mathbf z$8. In the moduli space $\mathbf z$9, WR similarity classes are precisely the fixed points of the deep-hole operation modulo similarity. This suggests that, in the planar setting, the deep-hole problem admits a canonical terminal locus rather than merely a set of extremal points (Fukshansky et al., 2023).

4. Iteration, elliptic curves, CM, and counting

The second major innovation is to iterate the construction. Given

$\mathbf 0,\mathbf x_1,\mathbf x_2$0

one constructs a finite sequence

$\mathbf 0,\mathbf x_1,\mathbf x_2$1

with

$\mathbf 0,\mathbf x_1,\mathbf x_2$2

The first step already sends any $\mathbf 0,\mathbf x_1,\mathbf x_2$3 to

$\mathbf 0,\mathbf x_1,\mathbf x_2$4

so after one deep-hole step all subsequent representatives lie on the vertical line $\mathbf 0,\mathbf x_1,\mathbf x_2$5. The proof gives the crucial estimate

$\mathbf 0,\mathbf x_1,\mathbf x_2$6

The stopping criterion is $\mathbf 0,\mathbf x_1,\mathbf x_2$7, and since $\mathbf 0,\mathbf x_1,\mathbf x_2$8, this occurs once

$\mathbf 0,\mathbf x_1,\mathbf x_2$9

Therefore, if

$\mathbf x_1,\mathbf x_2$0

then $\mathbf x_1,\mathbf x_2$1, so $\mathbf x_1,\mathbf x_2$2, and at that point $\mathbf x_1,\mathbf x_2$3 is WR. The theorem states

$\mathbf x_1,\mathbf x_2$4

Thus every planar similarity class reaches the WR locus after finitely many deep-hole steps (Fukshansky et al., 2023).

The same paper gives an elliptic-curve interpretation. A point $\mathbf x_1,\mathbf x_2$5 corresponds to the elliptic curve

$\mathbf x_1,\mathbf x_2$6

The full standard modular domain

$\mathbf x_1,\mathbf x_2$7

parameterizes isomorphism classes of elliptic curves, while $\mathbf x_1,\mathbf x_2$8 is half of it, adapted to lattice similarity classes. The deep-hole sequence therefore becomes a controlled motion in moduli space: the first step sends $\mathbf x_1,\mathbf x_2$9 to the boundary line $H(L)$00, and later steps move vertically downward until the WR boundary $H(L)$01 is reached (Fukshansky et al., 2023).

For quadratic irrational $H(L)$02, equivalently for CM elliptic curves, the arithmetic consequences are stronger. A lattice $H(L)$03 is arithmetic if its Gram matrix $H(L)$04 is a scalar multiple of an integer matrix, and $H(L)$05 is arithmetic iff

$H(L)$06

for some $H(L)$07, i.e. iff $H(L)$08 is a quadratic irrationality. This is equivalent to saying that $H(L)$09 has complex multiplication by the imaginary quadratic field $H(L)$10. If $H(L)$11 is quadratic irrational and $H(L)$12 is its deep-hole sequence, then all corresponding elliptic curves $H(L)$13 are isogenous. The proof uses a torsion statement about deep holes from Forst–Fukshansky: under arithmeticity hypotheses, the deep hole $H(L)$14 has finite order $H(L)$15 in $H(L)$16, which yields an isogeny $H(L)$17. The theorem also gives an explicit upper bound on the degree of an isogeny between consecutive terms: $H(L)$18 where

$H(L)$19

The same work also studies the inverse problem of counting planar lattices with a prescribed deep hole lattice. For

$H(L)$20

lying over a number field $H(L)$21, the set

$H(L)$22

is infinite, and the preimages lie on the circular arc

$H(L)$23

For primitive height $H(L)$24, the height-bounded set

$H(L)$25

satisfies

$H(L)$26

This gives a quantitative asymptotic upper estimate for planar similarity classes over $H(L)$27 with prescribed deep-hole image (Fukshansky et al., 2023).

5. Deep-hole problem for D1-D5-P black holes

In the supersymmetric black-hole literature, the deep-hole problem is posed for Type IIB on

$H(L)$28

with $H(L)$29 or $H(L)$30, microscopic $H(L)$31, and macroscopic $H(L)$32 of radius $H(L)$33. The charges are carried by $H(L)$34 D1-branes on $H(L)$35, $H(L)$36 D5-branes on $H(L)$37, and momentum $H(L)$38 along $H(L)$39. The supersymmetric rotating black hole compared to is the five-dimensional D1-D5-P BMPV black hole, or equivalently its six-dimensional black string lift (Bena et al., 2016).

The cited construction gives the first family of supersymmetric six-dimensional D1-D5-P supergravity solutions obtained by adding specific momentum-carrying deformations to a two-charge maximally rotating D1-D5 supertube seed using superstratum technology and coiffuring. The fluctuations are labeled by integers

$H(L)$40

with mode amplitude $H(L)$41. The metric is written as

$H(L)$42

where the flat $H(L)$43 base is

$H(L)$44

with

$H(L)$45

The $H(L)$46-fibration is

$H(L)$47

Momentum is added via deformations with phase

$H(L)$48

The fluctuating ansatz takes

$H(L)$49

$H(L)$50

with profile

$H(L)$51

Coiffuring is used so that, although the tensor fields depend on $H(L)$52, the metric does not (Bena et al., 2016).

These solutions have the same conserved quantum numbers as general supersymmetric rotating D1-D5-P black holes in five dimensions. The explicit expressions are

$H(L)$53

where

$H(L)$54

A key feature is that the family permits arbitrarily small finite angular momenta. For the subclass with $H(L)$55,

$H(L)$56

Thus taking $H(L)$57 with $H(L)$58 adjusted appropriately sends both angular momenta to zero while retaining momentum charge, producing the first microstate geometries of the non-rotating D1-D5-P Strominger-Vafa black hole (Bena et al., 2016).

6. Throat geometry, holography, and limitations

The black-hole comparison geometry is the supersymmetric rotating D1-D5-P black hole, whose six-dimensional near-horizon geometry is $H(L)$59 fibered over extremal BTZ with metric

$H(L)$60

The parameters obey

$H(L)$61

and

$H(L)$62

The microstate geometries are well approximated by the corresponding black-hole solution, and in particular they exhibit the same near-horizon throat. When

$H(L)$63

the cap lies deep in the AdS$H(L)$64 region. For $H(L)$65, the leading terms in the metric coincide with those of the corresponding black hole; for $H(L)$66, the geometry caps off smoothly. The proper length of the throat is

$H(L)$67

so $H(L)$68 gives an arbitrarily deep throat. The family lies within the cosmic censorship bound whenever

$H(L)$69

This is the explicit criterion showing that smooth horizonless geometries exist in the same parameter region where a large BMPV black hole exists (Bena et al., 2016).

Regularity is controlled by the constraints

$H(L)$70

together with the condition that $H(L)$71 vanish at

$H(L)$72

To ensure $H(L)$73, the construction uses the bound

$H(L)$74

which implies $H(L)$75, and hence $H(L)$76. The explicit $H(L)$77 family is stated to be easy to verify as regular and CTC-free everywhere (Bena et al., 2016).

The holographic dual is the $H(L)$78 D1-D5 orbifold CFT with central charge

$H(L)$79

The states are supersymmetric Ramond-sector states at the symmetric-orbifold point $H(L)$80. In strand language, momentum is carried on $H(L)$81 strands by commuting operators

$H(L)$82

and the identified coherent superposition is

$H(L)$83

with strand budget

$H(L)$84

The average strand numbers

$H(L)$85

reproduce the supergravity charges $H(L)$86 exactly. At the same time, the work is explicit that these are not generic typical black-hole states. They are protected, supersymmetric, Ramond-sector states counted by the elliptic genus, but they are highly special coherent states built from a restricted set of generators. The paper therefore partially resolves and substantially advances the deep-hole problem: it establishes that smooth horizonless microstates can exist deep inside the classical black-hole regime and with arbitrarily small angular momenta, but it does not prove that generic black-hole microstates are smooth horizonless geometries, nor does it address non-extremal or non-supersymmetric cases (Bena et al., 2016).

A plausible implication is that the two uses of the term share a structural theme despite their very different mathematical content. In both cases, the “deep-hole” perspective turns an extremal geometric notion into a dynamical or constructive framework: in the planar lattice setting, repeated passage to $H(L)$87 drives every similarity class to the well-rounded locus, while in the D1-D5-P setting, the problem is to realize microstates that sit deep down the classical throat yet terminate in a smooth cap rather than a horizon.