ON THE FUNDAMENTAL REGULARITY THEOREM FOR

MASS-MINIMIZING FLAT CHAINS

arXiv:2408.04083v1 [math.DG] 7 Aug 2024

BRIAN WHITE

Abstract. In the theory of flat chains with coefficients in a normed abelian

group, we give a simple necessary and sufficient condition on a group element
g in order for the following fundamental regularity principle to hold: if a mass-
minimizing chain is, in a ball disjoint from the boundary, sufficiently weakly
close to a multiplicity g disk, then, in a smaller ball, it is a C 1,α perturbation
with multiplicity g of that disk.

1. Introduction
The fundamental regularity theorem for mass-minimizing integral flat chains can
be stated in various ways, one of which is the following:
Theorem 1. Let 0 < α < 1. Suppose M is a a mass-minimizing integral m-
chain in Rd such that M has no boundary in the open unit ball B(0, 1). Suppose
also that M xB(0, 1) is weakly sufficiently close to a multiplicity-1 disk through the
origin. Then M xB(0, 1/2) is an m-manifold of multiplicity 1. Furthermore, the
m-manifold is the graph of a function f over a domain in the given disk with
kf kC 2,α ≤ 1.
More generally, the theorem is also true for “almost minimizing” chains. In
particular, it applies to chains that are homologically mass minimizing in smooth
Riemannian manifolds, or, more generally, in sets of positive reach. See §3.
If G is any normed abelian group, then the theory of integral flat chains gener-
alizes to flat chains with coefficients in G. In this paper, we address the question:
for which coefficient groups G and for which multiplicities g ∈ G does Theorem 1
hold?
It is natural to require that G be complete with respect to its norm | · |, i.e.,
that G be a complete metric space with respect to the metric d(x, y) = |x − y|.
(Otherwise, replace G by its metric space completion.) It is also natural to require
that
(1) If R < ∞, then {x ∈ G : |x| ≤ R} is compact.
Condition (1) is the necessary and sufficient condition for the fundamental com-
pactness theorem for flat chains to hold. (This follows easily from the Deformation
Theorem [Fle66, Theorem 7.3] or [Whi99a].)
For such normed abelian coefficient groups G, we prove

Date: 7 August, 2024.

2020 Mathematics Subject Classification. Primary 49Q15.
Key words and phrases. mass-minimizing, flat chain.
1
2 BRIAN WHITE

Theorem 2. Suppose g ∈ G. Theorem 1 holds for m-chains with coefficients in

G and with “multiplicity g” in place of “multiplicity 1” if and only g satisfies the
following strong triangle inequality:
(2) |g| < inf{|a| + |b| : a, b ∈ G \ {0}, a + b = g}.
We remark that (as is easily shown) the strong triangle inequality (2) is equiva-
lent to
(i) inf x∈G\{0} |x| > 0, and
(ii) |g| < |a| + |b| for all a, b ∈ G \ {0} for which a + b = g.
Simple examples show that the strong triangle inequality is necessary for the
regularity theorem to hold; see §5. Most of the paper is devoted to showing that
sufficiency of the strong triangle inequality follows from the Allard Regularity The-
orem [All72]. See [DPGS24] for a greatly simplified proof of Allard’s theorem.
Theorem 1 holds more generally for integral flat chains that minimize (or almost
minimize) parametric elliptic functionals. See [SS82], [Bom82], [Fed69, 5.3.14], or
(for the original, slightly different theorem) [Alm68]. I conjecture that Theorem 2
also holds for parametric elliptic functionals. (If the group has elements of order 2,
then the parametric elliptic integrand needs to be even for the functional to make
sense.) The proofs in this paper rely on monotonicity and on the Allard Regularity
Theorem and hence only work for mass.
In a different direction, De Pauw and Züst [DPZ19] have proved a regularity the-
orem for mass-minimizing or almost mass-minimizing flat chains (with coefficients
in an abelian group) in possibly infinite dimensional Hilbert spaces. Specifically,
they prove that if {|x| : x ∈ G} is a discrete set, then the regular set of the chain
is a dense open subset of the support. (In the finite dimensional setting, and with
the notion of almost minimizing used in this paper, their result follows immediately
from the Allard Regularity Theorem.)

2. Preliminaries
Throughout the paper, G is a normed abelian group, i.e., an abelian group with
a translation invariant metric; the norm of an element is its distance to 0. We
will always assume that G is complete and satisfies the compactness property (1).
In this paper, “m-chain” will mean “flat m-chain with coefficients in G”. We do
not require that the chains have compact support. Thus, in the terminology of
Federer’s book [Fed69], they would be called “locally flat chains”. The appendix
of [Whi09] indicates how to extend the theory of compactly supported chains to the
general case. See [Whi96] for a quick introduction to flat chains with coefficients in
a normed abelian group. Fleming’s original paper [Fle66] is the standard reference.
See also [Whi99a] and [Whi99b], or, for a different approach, [DPH12].
If M is an m-chain in Rd , we let |M | denote its mass. If M has locally finite
mass (which will be the case for all m-chains that arise in this paper) and if S
is a Borel subset of Rd , then M has a well-defined portion in S, denoted M xS.
Furthermore, M determines a Radon measure µM on Rd such that
µM (S) = |M xS|
for every Borel set S.
Except in §5, we will be working with groups that also have the property:
(3) inf |x| > 0.
x∈G, x6=0
ON THE FUNDAMENTAL REGULARITY THEOREM FOR MASS-MINIMIZING FLAT CHAINS3

(If G does not satisfy (3), then no element of G satisfies the strong triangle in-
equality (2).) For coefficient groups satisfying (3), every compact set is finite. In
particular, by (1),
If R < ∞ then {x ∈ G : |x| ≤ R} is a finite set.
For such normed groups G, every m-chain M of finite (or locally finite) mass is
rectifiable [Whi99b]. That is, M can be written as
∞
X
M= gi [Si ],
i=1
where the gi ∈ G, where the Si are disjoint Borel sets, and and where each Si is
contained in a C 1 , oriented m-manifold Σi . If W is a Borel subset of the ambient
space, then
∞
X
M xW = gi [Si ∩ W ],
i=1
∞
X
|M | = |gi | Hm (Si ),
i=1
∞
X
µM (W ) = |gi | Hm (Si ∩ W ).
i=1

We also define flat norms F (·) and F (·, K) and flat seminorms FW (·) as follows.
Definition 3. Let M be a flat m-chain in Rd . We define F (M ) to be the infimum
of
(4) |M − ∂Q| + |Q|.
over all (m + 1)-chains Q.
If K is a closed subset of Rd , we define F (M ; K) to be the infimum of (4) among
all (m + 1)-chains Q supported in K.
If U is an open subset of Rd , we let
FU (M )
be the infimum of
|(M − ∂Q)xU | + |QxU |
over all finite-mass (m + 1)-chains Q.
The term (M − ∂Q)xU needs explanation, since we have only defined AxU when
A has locally finite mass. Suppose A is any m-chain and U is an open set. If there
is a m-chain B supported in U c such that A − B has locally finite mass, then we
let AxU = (A − B)xU ; otherwise we leave AxU undefined. If AxU is undefined, we
take |AxU | to be infinite.
Note that if U ⊂ W , then
FU (M ) ≤ FW (M ).
If U is compact, then FU (M ) < ∞. It follows that if M is compactly supported,
then F (M ) < ∞. We say that Mn converges (weakly) to M and write Mn → M
provided
FU (Mn − M ) → 0
4 BRIAN WHITE

for all bounded open sets U of Rd . If Mi and M are supported in a compact set,
then Mn → M if and only if F (Mn − M ) → 0.

3. λ-Minimizing Chains
Definition 4. For λ ∈ [0, ∞), an m-chain M of locally finite mass in Rd is called
λ-minimizing provided it has the following property. If K is a compact subset of
U and if Q is an (m + 1)-chain compactly supported in
{x : dist(x, spt M ) < r},
then
(5) (1 − λr) |M xK| ≤ |M xK + ∂Q|.
Thus “0-minimizing” is the same as “mass-minimizing”.
Remark 5. If M is λ-minimizing, then (5) also follows from the slightly weaker
hypothesis
spt Q ⊂ {x : dist(x, spt M ) ≤ r},
for, in that case,
(1 − λR) |M xK| ≤ |M xK + ∂Q|
holds for every R > r and therefore also for R = r.
If C is a closed subset of Rd , if x ∈ Rd , and if there is a unique point y ∈ C
closest to x, we let πC (x) = y; otherwise πC (x) is not defined (i.e., x is not in the
domain of πC .) We define reach(C) to be the smallest number R such that
{x : dist(x, C) < R}
is contained in the domain of πC .
Theorem 6. Suppose C ⊂ Rd is a set with R := reach(C) > 0. Suppose that M
is homologically minimizing in C i.e., that M is supported in C, and that if K is a
compact subset of spt M , then
|M xK| ≤ |M xK + ∂Q|
for every (m + 1)-chain Q supported in C. Then M is (m/R)-minimizing.
Proof. For r < R, the restriction π of πC to {x : dist(x, C) ≤ r} is Lipschitz with
Lipschitz constant ≤ R/(R − r). (See [Fed59, Theorem 4.8(8)].) Suppose that Q is
an (m + 1)-chain supported in
{x : dist(x, K) ≤ r} ⊂ {x : dist(x, C) ≤ r}.
Then
|M xK| ≤ |M xK + ∂(π# Q)|
= |π# (M xK + ∂Q)|
 m
R
≤ |M xK + ∂Q|
R−r
 r −m

= 1− |M xK + ∂Q|,
R
ON THE FUNDAMENTAL REGULARITY THEOREM FOR MASS-MINIMIZING FLAT CHAINS5

so
 r m
|M xK + ∂Q| ≥ 1 − |M xK|
R 
 m
≥ 1 − r |M xK|.
R

d
Theorem 7. Suppose M is a rectifiable, λ-minimizing m-chain in R . Then in
Rd \ spt(∂M ), the varifold V associated to M has mean curvature bounded by λ.
Proof. Let X be smooth vectorfield supported in a compact subset C of Rd \
spt(∂M ) with |X(p)| ≤ 1 for all p. Let K = C ∩ spt M . Let φt (p) = p + tX(p). Let
Σt = φ# (M xK). Let Qt be the image of
(M xK) × [0, t]
under the map (p, τ ) 7→ φτ (p).
Then
(1 − λt)|M xK| ≤ |Σt |
Thus, differentiating, and using the first variation formula,
−λµV (K) ≤ δ(V ; X).
Since the same holds for −X, we see that
|δ(V ; X)| ≤ λµV (K),
which is equivalent to the assertion of the theorem. 
Theorem 8. Suppose that Mi and M are m-chains of locally finite mass in Rd ,
that Mi is λ-minimizing, and that Mi → M . Then M is λ-minimizing and
µMi → µM .
Proof. By passing to a subsequence, we may suppose that spt Mi converges to a
closed set X. Thus
dist(·, spt Mi ) → dist(·, X).
In particular, if V is any compact subset of U , then
max dist(q, spt Mi ) → max dist(q, X).
q∈V q∈V

Let 0 < r < 1/λ. Let Q be an (m + 1)-chain compactly supported in

{x ∈ U : dist(x, X) < r}.
Thus
max dist(·, X) < r.
spt Q
Hence
max dist(·, Mi ) < r
spt Q
for all sufficiently large i.
Now let K be a compact subset of X, let f : Rd → R be the distance function
to K, and let
K[s] = {f ≤ s} = {x : dist(x, K) ≤ r},
M [s] = M xK[s],
Mi [s] = Mi xK[s].
6 BRIAN WHITE

By passing to a subsequence, we can assume that for almost every s ∈ (0, r),
F (Mi [s] − M [s]; K[s]) → 0.
(See Theorem 15 in the appendix.) For such an s, there exist (m + 1)-chains Qi
supported in K[s] such that
|Mi [s] − M [s] + ∂Qi | + |Qi | → 0.
Now
max dist(·, X) ≤ s < r.
K[s]

Thus
max dist(·, spt Mi ) < r
K[s]

for all sufficiently large i. In particular,

max dist(·, spt Mi ) < r
spt Qi

for all sufficiently large i.

Now
(1 − λr)|Mi [s]| ≤ |Mi [s] − ∂(Q + Qi )|
≤ |M [s] − ∂Q| + |Mi [s] − M [s] − ∂Qi |
Thus
(6) (1 − λr) lim sup |Mi [s]| ≤ |M [s] − ∂Q|.
i

In particular (letting Q = 0), we see that every compact set K in X is contained

in the interior of a set K[s] for which lim supi |M xK[s]| < ∞. Thus, after passing to
a subsequence, the Radon measures µi = µMi converge weakly to a Radon measure
µ on U .
By (6) and lower semicontinuity of mass with respect to flat convergence,
(1 − λr)|M [s]| ≤ (1 − λr) lim inf |Mi [s]|
i

(7) ≤ (1 − λr) lim sup |Mi [s]|

i
≤ |M [s] − ∂Q|.
Letting s → 0 gives
(1 − λr)|M [0]| ≤ |M [0] − ∂Q|.
Hence M is λ-minimizing.
Now consider (7) in the case Q = 0. For almost every s, µM (K[s]) → µ(K[s]).
For such s, we can rewrite (7) as
(1 − λr)µM (K[s]) ≤ (1 − λr)µ(K[s]) ≤ µM (K[s]).
Letting s → 0 and then r → 0 gives
µM (K) ≤ µ(K) ≤ µM (K).

ON THE FUNDAMENTAL REGULARITY THEOREM FOR MASS-MINIMIZING FLAT CHAINS7

4. Main Theorem
Throughout this section, we assume that g satisfies the strong triangle inequality,
and we prove the fundamental regularity theorem under that assumption. Note that
the strong triangle inequality for g is equivalent to the existence of an ǫ > 0 such
that
(8) |a| + |b| ≥ |g| + ǫ if a + b = g and a, b ∈ G \ {0}.
This implies that
(9) inf |x| > 0.
x∈G, x6=0

(Indeed, if x 6= 0, then |x| ≥ min{ǫ/2, |g|}.)

The condition (9) implies that every flat chain of locally finite mass is rectifi-
able [Whi99b].
Lemma 9. Suppose M is a rectifiable, compactly supported m-chain such that
π# M = g[Ω],
where Ω is a Borel subset of an oriented m-plane P and where π : M → P is the
orthogonal projection. Suppose also that M has no points of multiplicity g. Then
|M | ≥ (|g| + ǫ)Hm (Ω).
Here, “point of multiplicity g” means “point that has a plane of multiplicity g
as a tangent cone”.

Proof. Since M is rectifiable, it can be written as

X
M= gi [Si ],
i

where the Si are disjoint Borel sets and each Si is contained in an oriented C 1
m-manifold Σi . By subdividing the Si and Σi and reorienting (if necessary), we
can assume that if p ∈ Si and if
π| Tan(Σi , p) : Tan(Σi , p) → P
is an isomorphism, then it preserves orientation. Let S = ∪Si and define
γ : S → G,
γ(p) = gi for p ∈ Si .
Since M has no points of multiplicity g, we can assume that γ(p) is never equal
to g or −g.
Since π# M = g[Ω], we see that
X
γ(p) = g
p∈S∩π −1 x

for almost every x ∈ Ω, and therefore that

X
|γ(p)| ≥ |g| + ǫ
p∈π −1 x

since γ(p) 6= g.
8 BRIAN WHITE

Thus
Z
|M | = |γ(p)| dHm p
ZS X
≥ |γ(p)| dx
x∈Ω p∈S∩π −1 x
Z
≥ (|g| + ǫ) dHm
Ω
≥ (|g| + ǫ)Hm (Ω).


If M is an m-chain of locally finite mass, and if B(p, r) is a ball, we let

Φ(M, p, r)
be the minimum of  
1
FB(0,1) (M − p), g[P ]
r
among all oriented m-planes P with 0 ∈ P .
We also let
|M xB(p, r)|
Θ(M, p, r) = ,
ωm r m
Θ(M, p) = lim Θ(M, p, r) (if the limit exists), and
r→0
Θ(M ) = sup Θ(M, p, r).
p∈Rd , r>0

Lemma 10. There is a δ > 0 with the following property. Suppose M is a mass-
minimizing m-chain in Rd such that
(1) ∂M = 0.
(2) Θ(M ) ≤ |g|.
(3) Φ(M, 0, r) ≤ δ for all r > 1.
Then M = g[P ] for some oriented m-plane P .
Proof. Suppose that Mi is a sequence of m-chains satisfying (1) and (2), and such
that
δi := sup F (Mi , 0, r) → 0.
r>1
It suffices to prove that Mi is a m-plane of multiplicity g for all sufficiently large i.
By rotating the Mi , we can assume that
(10) FB(0,1) (Mi − g[P ]) → 0
where P = Rm × RN −m .
It follows that the Mi converge weakly to g[P ].
(If this is not clear, note that if R > 1, then, after passing to a subsequence,
there is a P ′ such that FB(0,R) (Mi − g[P ′ ]) → 0. By (10), P ′ = P , and thus it was
not necessary to pass to a subsequence.)
Let
K[r] = B m (0, r) × B d−m (0, r).
ON THE FUNDAMENTAL REGULARITY THEOREM FOR MASS-MINIMIZING FLAT CHAINS9

That is, K[r] = {f ≤ r}, where

f : Rd → R,
f (x, y) = max{|x|, |y|} (x ∈ Rm , y ∈ Rd−m ).
After passing to a subsequence, we can assume (by Theorem 15) that
Mi xΩ(r) → g[Dr ]
for almost every r, where Dr = P ∩ B(0, r). Fix such an r, and let Mi′ = Mi xK[r]
and D = Dr .
Since spt(Mi ) → P , if follows that
spt(Mi′ ) ⊂ B(0, r) × B(0, ηi )
where ηi → 0. For all sufficiently large i, ηi < r, and thus
spt(∂Mi′ ) ⊂ ∂B m (0, r) × B(0, ηi ).
Consequently,
π# Mi = gi [D]
for some gi . Since Mi → g[D], we see that |gi − g| → 0 and therefore (since
inf x∈G, x6=0 |x| > 0) that gi = g for all sufficiently large i.
The λ-minimizing property implies (see Theorem 8) that
|Mi′ | → |g[D]| = |g|Hm (D).
Hence, by Lemma 9, for large i, Mi′ must have a point pi at which the tangent cone
is a multiplicity m-plane. Since Θ(Mi′ ) ≤ |g|, it follows from monotonicity that Mi′
is a cone with vertex p′i , and therefore a multiplicity g plane. 

Corollary 11. If M is a λ-minimizing m-chain in Rd , if p is a point in spt(M ) \

spt(∂M ), and if
σ := lim sup F (M, p, r) < δ,
r→0
then every tangent cone to M at p is a multiplicity-g plane.
Proof. Let M ′ be a tangent cone to M at p. Then F (M ′ , 0, r) ≤ δ for all r, so M ′
is a multiplicity-g plane by Lemma 10. 

Theorem 12. Suppose that G is a complete normed abelian group satisfying the
compactness property (1), and that g ∈ G satisfies the strong triangle inequality (8).
Suppose that U is an open subset of Rd and that Mi is a sequence of λ-minimizing
m-chains in Rd with spt(∂Mi ) ⊂ U c . Suppose also that Mi xU converges to g[Σ],
where Σ is a properly C 1 -embedded oriented m-manifold in U . Then there exists
an exhaustion W1 ⊂ W2 ⊂ . . . of U by open subsets and C 1,α properly embedded,
oriented m-manifolds Σi in Wi such that
Mi xWi = g[Σi ]
and such that the Σi converge in C 1,α with multiplicity 1 to Σ.
Proof. By the Allard Regularity Theorem ([All72] or [DPGS24]), it suffices to show
that if W ⊂⊂ U , then for all sufficiently large i,
Θ(M, p) ≥ |g| for every p ∈ W ∩ spt Mi .
10 BRIAN WHITE

Suppose not. Then (after passing to a subsequence), there exist pi ∈ W ∩ spt Mi

for which
Θ(Mi , pi ) < |g|.
By passing to a further subsequence, we can assume that the pi converge to a point
p ∈ W.
Choose Ri converging to 0 sufficiently slowly that
1
(Mi − pi ) → g[Tan(Σ, p)].
Ri
Hence if
ρi ρi
0 < lim inf ≤ lim sup < ∞,
Ri Ri
then
1
(Mi − pi ) → g[Tan(Σ(p)],
ρi
and therefore
(11) Φ(Mi , pi , ρi ) → 0.
Let δ be as in Lemma 10. Let
1
Mi′ := (Mi − pi ),
ri
where ri is the supremum of r ∈ (0, Ri ] such that
F (Mi , pi , r) ≥ δ.
Note that, by Corollary 11, there exists such an r, so ri > 0. By (11),
ri
→ 0.
Ri
Let M ′ be a subsequential limit of Mi′ . Then
F (M ′ , 0, r) ≤ δ
for all r ≥ 1, so M ′ is a multiplicity g plane through the origin (by Lemma 10),
and therefore F (M ′ , 0, r) = 0 for all r > 0. But F (M ′ , 0, 1) ≥ δ by choice of ri , a
contradiction. 

5. Examples
Here we show that the hypothesis
(12) |g| < inf (|a| + |b|)
a,b∈G\{0}, a+b=g

is necessary in Theorem 12.

Recall that (12) is equivalent to
(1) inf x∈G\{0} |x| > 0, and
(2) |g| < |a| + |b| for all a, b ∈ G \ {0} such that a + b = g.
Let U = B m (0, 1) × R and let Σ be the unit m-ball Bm (0, 1) × {0} with the
standard orientation.
If inf x∈G\{0} |x| = 0, choose an ∈ G \ {0} with an → 0. Then (g + an )[Σ]
is area-minimizing and converges as n → ∞ to g[Σ], violating the conclusion of
Theorem 12.
ON THE FUNDAMENTAL REGULARITY THEOREM FOR MASS-MINIMIZING FLAT CHAINS
11

Now suppose that there exist a, b ∈ G \ {0} such that

g = a + b,
|g| = |a| + |b|.
Then
Mn := a[Σ] + b[Σ + (1/n)em+1 ]
is mass-minimizing and converges to g[P ] as n → ∞, violating the conclusion of
Theorem 12.
Here, Mn is mass-minimizing according to the following lemma:
Lemma 13. Suppose that a, b ∈ G and that |a + b| = |a| + |b|. Suppose also that Ω
is an open subset of an oriented m-plane P , and let
M = a[Ω] + b[Ω + u],
where u is a vector perpendicular to P . Then M is mass-minimizing.
Proof. Let M ′ be a compactly supported chain with ∂M ′ = ∂M . Let π denote
orthogonal projection onto P . Then
π# (M ′ ) = (a + b)[Ω],
so
|M ′ | ≥ |π# M ′ |
= |a + b|Hm (Ω)
= (|a| + |b|)Hm (Ω)
= (|a|)Hm (Ω) + |b|Hm (Ω + u)
= |M |.


6. appendix
Lemma 14. Suppose that M is a finite mass m-chain in Rd , that f : Rd → R is
a Lipschitz function with Lipschitz constant ≤ 1. For s ∈ R, let
K[s] = {f ≤ s},
M [s] = M xK[s].

If W ⊂ Rd is an open set that contains {f ≤ b}, then

Z b
(13) F (M [s]; K[s]) ds ≤ (1 + (b − a))F (M ).
s=a

Proof. If A is a chain with |A| < ∞ and if s ≤ b, we let

A[s] := Ax{f ≤ s} = AxK[s].
Let Q be an (m + 1)-chain such that
|M − ∂Q| + |Q| < ∞.
12 BRIAN WHITE

Since |M | < ∞, it follows that |∂Q| < ∞. Note that

F (M [s]; K[s]) ≤ |M [s] − ∂(Q[s])| + |Q[s]|
= |M [s] − (∂Q)[s] + |Q(s)| + |(∂Q)[s] − ∂(Q[s])|
(14)
= |(M − ∂Q)[s]| + |Q(s)| + |(∂Q)[s] − ∂(Q[s])|
= |(M − ∂Q| + |Q| + |(∂Q)[s] − ∂(Q[s])|
Also
Z b
|(∂Q)[s] − ∂(Q[s])| ds ≤ |Q|
a
by [Fle66, Theorem 5.7]. Thus integrating (14) gives
Z b
F (M [s], K[s]) ds ≤ (1 + b − a)(|M − ∂Q| + |Q|).
a
Taking the infimum over (m + 1)-chains Q gives (13). 
d
Theorem 15. Suppose that f : R → R is Lipschitz with Lipschitz constant ≤ 1
and that K[s] := {f ≤ s} is compact for every s < ∞. If Ti are m-chains of locally
finite mass such that Ti → 0, then (after passing to a subsequence)
(15) F (Ti xK[s]; K[s]) → 0
for almost every s ≥ 0.
Proof. For almost every R,
Ti xB(0, R) → 0.
(Indeed, this holds for any R such that µT ∂B(0, R) = 0.) Thus, after passing to a
subsequence, there exist Rn → ∞ such that
1
F (Tn xB(0, Rn )) < n
2
and therefore X
F (Tn xB(0, Rn )) < ∞.
n
Let Tn′ = Tn xB(0, Rn ).
Fix an R < ∞. For all sufficiently large n, say n ≥ N , K[R] is contained in
B(0, Rn ). Thus if r ≤ R and n ≥ N , then
Tn xK[r] = Tn′ xK[r],
so
Z R X XZ R Z
F (Tn xK[r]; K[r]) dr = F (Tn′ xK[s]; K[s]) ds
0 n≥N n≥N 0
X
≤ (1 + R) F (Tn′ )
n≥N
< ∞.
Hence, for almost all s ∈ (0, R), we have
X
F (Ti [s]; K[s]) < ∞,
i
and therefore (15) holds. Since R is arbitrary, (15) holds for almost every s ∈
[0, ∞). 
ON THE FUNDAMENTAL REGULARITY THEOREM FOR MASS-MINIMIZING FLAT CHAINS
13

References
[All72] William K. Allard, On the first variation of a varifold, Ann. of Math. (2) 95 (1972),
417–491, DOI 10.2307/1970868. MR0307015 ↑2, 9
[Alm68] F. J. Almgren Jr., Existence and regularity almost everywhere of solutions to ellip-
tic variational problems among surfaces of varying topological type and singularity
structure, Ann. of Math. (2) 87 (1968), 321–391, DOI 10.2307/1970587. MR0225243
↑2
[Bom82] Enrico Bombieri, Regularity theory for almost minimal currents, Arch. Rational Mech.
Anal. 78 (1982), no. 2, 99–130, DOI 10.1007/BF00250836. MR0648941 ↑2
[DPH12] Thierry De Pauw and Robert Hardt, Rectifiable and flat G chains in a metric space,
Amer. J. Math. 134 (2012), no. 1, 1–69, DOI 10.1353/ajm.2012.0004. MR2876138 ↑2
[DPZ19] Thierry De Pauw and Roger Züst, Partial regularity of almost minimizing rectifi-
able G chains in Hilbert space, Amer. J. Math. 141 (2019), no. 6, 1591–1705, DOI
10.1353/ajm.2019.0044. MR4030524 ↑2
[DPGS24] Guido De Philippis, Carlo Gasparetto, and Felix Schulze, A short proof of Allard’s and
Brakke’s regularity theorems, Int. Math. Res. Not. IMRN 9 (2024), 7594–7613, DOI
10.1093/imrn/rnad281. MR4742836 ↑2, 9
[Fed69] Herbert Federer, Geometric measure theory, Die Grundlehren der mathematischen
Wissenschaften, vol. Band 153, Springer-Verlag New York, Inc., New York, 1969.
MR0257325 ↑2
[Fed59] , Curvature measures, Trans. Amer. Math. Soc. 93 (1959), 418–491, DOI
10.2307/1993504. MR0110078 ↑4
[Fle66] Wendell H. Fleming, Flat chains over a finite coefficient group, Trans. Amer. Math.
Soc. 121 (1966), 160–186, DOI 10.2307/1994337. MR0185084 ↑1, 2, 12
[SS82] Richard Schoen and Leon Simon, A new proof of the regularity theorem for rectifiable
currents which minimize parametric elliptic functionals, Indiana Univ. Math. J. 31
(1982), no. 3, 415–434, DOI 10.1512/iumj.1982.31.31035. MR0652826 ↑2
[Whi96] Brian White, Existence of least-energy configurations of immiscible fluids, J. Geom.
Anal. 6 (1996), no. 1, 151–161, DOI 10.1007/BF02921571. MR1402391 ↑2
[Whi99a] , The deformation theorem for flat chains, Acta Math. 183 (1999), no. 2, 255–
271, DOI 10.1007/BF02392829. MR1738045 ↑1, 2
[Whi99b] , Rectifiability of flat chains, Ann. of Math. (2) 150 (1999), no. 1, 165–184, DOI
10.2307/121100. MR1715323 ↑2, 3, 7
[Whi09] , Currents and flat chains associated to varifolds, with an application to mean
curvature flow, Duke Math. J. 148 (2009), no. 1, 41–62, DOI 10.1215/00127094-2009-
019. MR2515099 ↑2

Brian White
Department of Mathematics
Stanford University
Stanford, CA 94305, USA
E-mail address: [email protected]