Pith. sign in

REVIEW 3 major objections 7 minor 21 references

Murray's law for discrete and continuum models of biological networks

T0 review · 3 major / 7 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proves that Murray's law—the classical scaling relation between branch conductivity and metabolic cost—holds at every critical point of a discrete transport-network energy and carries over to continuum and diffusion-augmented…

desk verdict Solid discrete generalization of Murray's law, with continuum analogues that are formally derived and explicitly conditional on unproved flux regularity; the abstract oversells the rigorous content. read the letter →

arxiv 1908.01197 v1 pith:BJU3C3JY submitted 2019-08-03 math.AP

Jan Haskovec , Peter Markowich , Giulia Pilli This is my paper · ORCID
classification math.AP MSC 92C3505C2176S05
keywords Murray'slawbiologicaltransportationnetworkscontinuumlimitenergyminimizationconductivityscalingKirchhoffmetaboliccost3/4-law
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper shows that Murray's law—the rule that branch radii or conductivities in a transport network scale to balance pumping cost against metabolic upkeep—follows from energy minimization in a general discrete network model and survives passage to continuum descriptions. In the discrete setting, every critical point of the total energy satisfies a generalized relation: at each node, the sum of incoming conductivities raised to the power $(\gamma+1)/2$, plus the scaled source strength, equals the same sum of outgoing conductivities. For blood vessels ($\gamma=1/2$) this reduces to the classical $3/4$-power law. The same balance is derived, at least formally, for two families of continuum models—one obtained as the continuum limit of the discrete model on a rectangular mesh, and one phenomenological model of network formation under Darcy flow—and for their diffusion-augmented versions. The continuum results require the flux to be locally Lipschitz continuous, a regularity property the paper does not prove, leaving that step as the main open gap.

What carries the argument

The load-bearing object is the derivative identity (11): $\partial_{C_{kl}}\sum_{(i,j)\in E}(Q_{ij}^2/C_{ij})L_{ij}=-Q_{kl}^2/C_{kl}^2 L_{kl}$, which holds because the Kirchhoff-constrained pressures respond to conductivity changes so that all cross terms cancel. It turns the stationarity condition into the pointwise relation $Q^2=\nu C^{\gamma+1}$, and Kirchhoff's law then converts the flux balance at each node into the conductivity-power identity (15). In the continuum analogues, the same role is played by the Euler–Lagrange equations (24) and (45), which tie pressure gradients to powers of the conductivity field, and by the De Giorgi–Federer Green formula, which transplants the node balance into a boundary-flux balance over arbitrary regions $\Lambda$ (equations (31), (38), (50)).

What would settle it

One concrete test: construct, for $\gamma>1$ and a smooth source $S\in L^2(\Omega)$, the unique minimizer $p$ of (34) (or of (52)) on a rectangle, and compute the flux $q=-(rI+c)\nabla p$; if $q$ has an interior discontinuity or cusp where the pressure gradient changes sign, the Green-formula step behind (31), (38), and (50) is not justified for that solution, so the continuum Murray law would not follow for the constructed solution.

Watch

Extended reading notes

Core claim

The central claim is that Murray's law is not a special feature of single tubes or symmetric trees but a generic consequence of minimizing the energy $E[C]=\sum_{(i,j)\in E}(Q_{ij}^2/C_{ij} + \nu C_{ij}^\gamma/\gamma)L_{ij}$ under Kirchhoff's law. At any critical point one has $Q_{ij}^2=\nu C_{ij}^{\gamma+1}$ on every edge, and Kirchhoff balance then yields the generalized Murray law (15), $\sum_{j\in N^+(i)}C_{ij}^{(\gamma+1)/2}+S_i/\sqrt{\nu}=\sum_{j\in N^-(i)}C_{ij}^{(\gamma+1)/2}$. The paper further claims the analogous balance holds in the continuum: for the diagonal-tensor model (19)–(23) the Euler–Lagrange relation $(\partial_{x_k}p)^2=\nu|c_k|^{\gamma-1}$ converts the integrated Poisson equation over any region $\Lambda$ into the boundary-flux identity (31), and the same structure yields (50) for the $m\otimes m$ model when restricted to linearly stable steady states. These derivations are formal, justified only under the assumption that the flux is locally Lipschitz continuous.

Load-bearing premise

The continuum versions of Murray's law depend on the flux through the network being locally Lipschitz continuous so that integrating the pressure equation over arbitrary regions and applying the boundary-flux formula is legitimate; the paper constructs only less regular weak solutions and leaves this smoothness unproved.

Editorial extensions

If this is right

  • In any network that minimizes (10) under Kirchhoff balance, the generalized Murray law (15) holds at every node, so optimal conductivities obey a universal power-law relation regardless of graph topology.
  • Setting $\gamma=1/2$ recovers the classical $3/4$-law $C_0^{3/4}=\sum_i C_i^{3/4}$ for laminar blood flow, placing the classical observation inside a general variational framework.
  • The continuum limit of the discrete model preserves the Murray balance in the form (31): for any region $\Lambda$, the difference of in- and out-fluxes through $\partial\Lambda$ equals the integrated source strength.
  • For the phenomenological $m\otimes m$ model, only linearly stable steady states are admissible, and on those states the Murray balance (50) holds; unstable states with $m=0$ on a positive-measure set require the modified law in Remark 1.
  • Adding diffusion to either continuum model does not destroy the Murray balance; it only replaces the power $|c_k|^{\gamma-1}$ (or the analogous $m$-expression) by $\nu|c_k|^{\gamma-1}-D^2\Delta c_k$, so the balance persists in (38) and (58).

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same variational argument likely yields Murray-type relations for other cost exponents and graph ensembles, since only (11), stationarity, and Kirchhoff's law are used; one testable extension is to random graphs with random source placements, where (15) should hold at every node individually.
  • The regularity gap suggests a concrete numerical experiment: compute minimizers of (34) and measure the Hölder exponent of the flux near points where $\partial_{x_k}p$ changes sign; if the exponent is strictly below 1, the Green-formula step fails and a different proof or a counterexample would be needed.
  • For $\gamma=1$, the free-boundary structure of (54)–(56) means the Murray law might serve as a sharp interface condition: the set where $|\nabla p|^2=\nu$ is the network, and (50) becomes a transmission condition across that interface—an interpretation the paper does not develop.
  • The discrete law (15) has a diagnostic corollary the authors do not state: given measured conductivities and sources in a real vascular network, the residual at each node quantifies how far that network is from optimality, offering a statistical test of Murray's law in vivo.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 7 minor

Summary. This paper derives Murray's law-type scaling relations from variational models of biological transportation networks. In the discrete setting (Section 2), for a graph with conductivities C_ij and energy E[C] = sum_{edges} (Q_ij^2/C_ij + nu/gamma C_ij^gamma) L_ij subject to Kirchhoff's law, the authors prove, via the derivative identity in Lemma 1, that every critical point satisfies Q_ij^2 = nu C_ij^(gamma+1), and hence the generalized Murray law (15): at each node, sqrt(nu) times the sum of incoming C_ij^((gamma+1)/2) plus the source S_i equals sqrt(nu) times the sum of outgoing C_ij^((gamma+1)/2); for gamma = 1/2 this reduces to the classical 3/4-law. In Section 3, for the continuum limit model (19)--(23), the paper formally derives the Euler-Lagrange relation (24) and, using the De Giorgi-Federer Green formula, the boundary-flux Murray law (31), together with a diffusive variant (38). In Section 4, for the phenomenological model (42)--(44), the paper derives the Euler-Lagrange equation (45) and, assuming linear stability so that m is nonzero almost everywhere, the Murray law (50). The manuscript explicitly labels the continuum derivations as formal and states that the required flux regularity is beyond its scope.

Significance. The discrete part is the strongest contribution: Lemma 1 is proved in detail, and the resulting identity (15) is a clean and nontrivial generalization of Murray's law to arbitrary graphs with multiple branching nodes and a general metabolic exponent gamma. The continuum sections are less rigorous: equations (31), (38), and (50) are conditional identities that hold provided the flux is locally Lipschitz on the integration set, and the weak solutions constructed in Lemmas 3 and 5 do not provide this regularity. As formal derivations the material is useful, and the manuscript is transparent about the gap, but the abstract's wording 'prove an analogue' overstates the rigorous content. The paper also relies on external results for the continuum limit and for the linear-stability criterion; these are supporting ingredients rather than circular assumptions, so no circularity concern arises.

major comments (3)
  1. [Sections 3, 3.1, 4, 4.1; Eqs. (29)--(31), (49)--(50)] The continuum Murray laws are not established as theorems. The derivations integrate the PDE over an arbitrary C^{0,1} subset Lambda and apply the De Giorgi-Federer formula to the flux q = -(rI+c) grad p (or q = -(rI+m tensor m) grad p), which requires q to be locally Lipschitz continuous on Lambda. Lemma 3 provides only p in H^1, and Lemma 5 provides p in H^1 intersect W^{1,2gamma/(gamma-1)} on A_+ union A_-; neither regularity implies the existence of well-defined traces of n dot q on partial Lambda. The paper acknowledges this in Sections 3 and 4 and in the introduction, but the abstract's claim to 'prove an analogue' is stronger than the content. The authors should either add the missing regularity proof or explicitly present (31), (38), and (50) as conditional formal identities and adjust the abstract accordingly. This is the main load-bearing gap.
  2. [Section 3.1, Lemma 3; Eqs. (33)--(34)] The proof of Lemma 3 asserts that the functional F in (34) is uniformly convex on H^1 and obtains a minimizer p in H^1. However, the term integral |grad p|^{2gamma/(gamma-1)} is not finite for every H^1 function when d >= 3 and 2gamma/(gamma-1) > 2, so F is not defined on all of H^1 and cannot be uniformly convex there. The natural solution space is W^{1,2gamma/(gamma-1)}(Omega), or H^1 intersect W^{1,2gamma/(gamma-1)}, and Lemma 3 should be restated in that space. This is a fixable but nontrivial correction.
  3. [Section 4.2, around Eq. (58)] The sentence 'On the set A we have (grad p tensor grad p)m = 0, which is equivalent to m dot grad p = 0' contradicts the Euler-Lagrange equation (grad p tensor grad p)m = -D^2 Delta m + nu |m|^{2(gamma-1)}m and the definition of A. On the set A the scalar product of the right-hand side with m is strictly positive, so (grad p dot m)^2 > 0 and hence (grad p tensor grad p)m cannot vanish. The statement should presumably refer to the complement A^c, where indeed m dot grad p = 0. As written this is an internal inconsistency in the derivation of the diffusive Murray law and needs correction.
minor comments (7)
  1. [Section 3, first paragraph] The text reads 'the formal continuum limit of of the discrete model'; 'of of' should be 'of'.
  2. [Section 4, before Eq. (50)] The phrase 'with the flux := -(rI + m tensor m) grad p' is missing the symbol q before ':='.
  3. [Remark 1] In the displayed formula of Remark 1, the last integral is printed as + int_{A^c cap partial Lambda_+} r |grad p dot n| ds, but to match the outflow term it should presumably be over A^c cap partial Lambda_-; otherwise the same boundary segment appears twice with the same sign.
  4. [Section 3.2, Eq. (38)] In equation (38), the characteristic functions chi_{A_+} and chi_{A_-} should carry the subscript k to match the sets A_k^+ and A_k^- defined in the previous paragraph, since the sets depend on the coordinate direction k.
  5. [Section 4.1, Eqs. (53)--(56)] The symbol 'c2' in equations (53), (54), and (56) is used without definition; it appears to encode the constraint |grad p|^2 <= nu (for instance c2 = 1/nu), and it should be defined explicitly or removed in favor of nu.
  6. [Section 4.2, Eq. (58)] The expression (nu |m|^{2gamma} - D^2 m dot Delta m)^{-1/2} is only meaningful where the radicand is positive; the paper should state explicitly that this is precisely the definition of A and explain how the formula is interpreted on A^c, as it later does in the final boundary integral.
  7. [Abstract and Introduction] The introduction's statement that the continuum calculations are formal is useful and should be mirrored in the abstract, which currently says the paper 'prove[s] an analogue' for the continuum systems.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the discrete Murray law is derived from critical-point equations and the continuum analogues from Euler-Lagrange plus Green's formula; only regularity is formal, not circular.

full rationale

The derivation chain is not circular. In Section 2, Lemma 1 proves the derivative identity (11) directly from the flux definition (7) and Kirchhoff law (8); the critical-point condition (13) then gives Q_ij^2 = \nu C_ij^{\gamma+1}, and substituting this into the flux balance (14) produces the generalized Murray law (15). Nothing equivalent to (15) is assumed in the energy functional (10) or in the constraints. The continuum sections are likewise consequences, not assumptions: the Euler-Lagrange equations (24) and (45) are first-order conditions for the stated energies, and the Murray-law formulas (31), (38), and (50) follow by integrating the Poisson equation over an arbitrary Lipschitz subset and inserting those first-order conditions. The target law is never an input. The paper does rely on prior work by the same authors for the continuum limit [12, 13] and for the linear-stability criterion [8], but those are supporting model and stability inputs, not the target relation, and the discrete argument in Section 2 is independent of them. The honest weakness is rigor, not circularity: Section 3 states that the calculation becomes rigorous only if the flux q = -(rI+c)\nabla p is locally Lipschitz, Section 3.1 says 'the proof of higher regularity is beyond the scope of this paper', and Section 4 similarly notes the central assumption that q = -(rI + m \otimes m)\nabla p is locally Lipschitz. Remark 1 further shows that (50) needs the stability-derived assumption meas(A)=meas(Omega). These are acknowledged regularity and conditionality gaps; they do not reduce the claimed laws to their own inputs. No fitted parameter is renamed as a prediction, and no uniqueness or ansatz result is imported to force the conclusion.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

No parameters are fitted to data and no new physical entities are introduced. The metabolic coefficient nu, exponent gamma, diffusivity D, and background permeability r are model inputs inherited from prior literature. The central claim depends on the background assumptions listed above, especially the unproved regularity assumption for the continuum derivations.

assumptions (6)
  • domain assumption Network flow follows Kirchhoff's law with flux proportional to pressure drop divided by edge length, and permeability follows Darcy or Hagen-Poiseuille type constitutive relations.
    Equations (7)-(8), (18), (41)-(42) model biological flows this way; the paper does not derive these physical laws.
  • domain assumption Total network cost is the sum of a pumping term Q^2/C and a metabolic term (nu/gamma) C^gamma per edge, or their continuum analogues.
    The energy functionals (10), (23), (44) and (57) are postulated based on prior literature [14], [10], [11], [12]; Murray's law is a consequence of this cost structure.
  • domain assumption The continuum energy functional (16) is the correct continuum limit of the discrete model on rectangular grids for gamma > 1.
    Section 3 takes this limit from Ref. [12] without reproving it; the present paper relies on that result.
  • standard math For the phenomenological model, linearly stable steady states must satisfy meas(A) = meas(Omega), i.e., m is nonzero almost everywhere.
    This follows from Theorem 7 and Remark 6 of Ref. [8], which states that if meas(A) < meas(Omega) then the solution is linearly unstable. The paper cites this theorem rather than proving it.
  • ad hoc to paper The flux q = -(rI + c) grad p (or q = -(rI + m tensor m) grad p) is locally Lipschitz continuous on the arbitrary subset Lambda, so the De Giorgi-Federer Green formula applies.
    This regularity is assumed in Sections 3 and 4 to pass from weak formulations to boundary integrals (31), (38), (50). The paper states it is not proved and defers it to future work.
  • domain assumption The isotropic background permeability r(x) >= r0 > 0 is introduced to make the elliptic problems uniformly solvable.
    Regularization (20) and (42) is taken from prior work to handle degeneracy of the conductivity tensor.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Murray's law for discrete and continuum models of biological networks." pith.science (2026). https://pith.science/paper/BJU3C3JY

@misc{pith2026190801197,
  author       = {Pith},
  title        = {Pith review of: Murray's law for discrete and continuum models of biological networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BJU3C3JY}},
  note         = {Machine review of arXiv:1908.01197}
}
read the original abstract

We demonstrate the validity of Murray's law, which represents a scaling relation for branch conductivities in a transportation network, for discrete and continuum models of biological networks. We first consider discrete networks with general metabolic coefficient and multiple branching nodes and derive a generalization of the classical 3/4-law. Next we prove an analogue of the discrete Murray's law for the continuum system obtained in the continuum limit of the discrete model on a rectangular mesh. Finally, we consider a continuum model derived from phenomenological considerations and show the validity of the Murray's law for its linearly stable steady states.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 19 canonical work pages

  1. [1]

    G. Albi, M. Artina, M. Fornasier and P. Markowich: Biological transportation net- works: modeling and simulation. Analysis and Applications. Vol. 14, Issue 01 (2016)

    work page 2016

  2. [2]

    G. Albi, M. Burger, J. Haskovec, P. Markowich and M. Schlo ttbom: Continuum Mod- elling of Biological Network Formation. In: N. Bellomo, P. Degond, and E. Tamdor (Eds.), Active Particles Vol.I - Theory, Models, Applications , Series: Modelling and Simulation in Science and Technology, Birkh¨ auser-Springer (Boston), 2017

    work page 2017

  3. [3]

    Alston and E.R

    M.E. Alston and E.R. Barber: Leaf venation, as a resistor, to optimize a switchable IR absorber. Sci Rep. 6:31611 (2016)

    work page 2016

  4. [4]

    A mesoscopic model of biological transportation networks

    M. Burger, J. Haskovec, P. Markowich and H. Ranetbauer: A mesoscopic model of biological transportation networks. Preprint, arXiv:1806.00120v2 (2018)

    work page Pith review arXiv 2018

  5. [5]

    Emerson and K

    D.R. Emerson and K. Cie´ slicki, X. Gu, R.W. Barber: Biomimetic design of mi- crofluidic manifolds based on a generalised Murray’s law. Lab Chip. 6(3), pp. 447–54 (2006)

    work page 2006

  6. [6]

    Evans: Partial differential equations

    L.C. Evans: Partial differential equations. Graduate Studies in Mathematics Vol. 19, American Mathematical Society (2010). 16

    work page 2010

  7. [7]

    Evans and R.F

    L.C. Evans and R.F. Gariepy: Measure Theory and Fine Properties of Functions. CRC press (2015)

    work page 2015

  8. [8]

    Haskovec, P

    J. Haskovec, P. Markowich and B. Perthame: Mathematical Analysis of a PDE System for Biological Network Formation. Comm. PDE 40:5, pp. 918-956 (2015)

    work page 2015

Show all 21 references
  1. [9]

    Haskovec, P

    J. Haskovec, P. Markowich, B. Perthame and M. Schlottbom : Notes on a PDE system for biological network formation. Nonlinear Analysis 138 (2016), pp. 127–155

    2016

  2. [10]

    Hu: Optimization, Adaptation, and Initialization of Biologic al Transport Networks

    D. Hu: Optimization, Adaptation, and Initialization of Biologic al Transport Networks. Notes from lecture (2013)

    2013

  3. [11]

    Hu and D

    D. Hu and D. Cai: Adaptation and Optimization of Biological Transport Netwo rks. Phys. Rev. Lett. 111 (2013), 138701

    2013

  4. [12]

    Haskovec, L

    J. Haskovec, L. M. Kreusser and P. Markowich: ODE and PDE based modeling of biological transportation networks . arXiv:1805.08526 (2018)

    2018 arXiv

  5. [13]

    Haskovec, L

    J. Haskovec, L. M. Kreusser and P. Markowich: Rigorous continuum limit for the discrete network formation problem. arXiv: (2018)

    2018

  6. [14]

    Murray: The Physiological Principle of Minimum Work: I

    C. Murray: The Physiological Principle of Minimum Work: I. The Vascu- lar System and the Cost of Blood Volume . PNAS 12 (3): 207–214 (1926). doi:10.1073/pnas.12.3.207

    1926 doi

  7. [15]

    Murray: The Physiological Principle of Minimum Work: II

    C. Murray: The Physiological Principle of Minimum Work: II. Oxygen Exc hange in Capillaries. PNAS 12 (5): 299–304 (1926). doi:10.1073/pnas.12.5.299

    1926 doi

  8. [16]

    Razavi, E

    M. Razavi, E. Shirani and G. Kassab: Scaling Laws of Flow Rate, Vessel Blood Vol- ume, Lengths, and Transit Times With Number of Capillaries . Frontiers in Physiology 9:581 (2018)

    2018

  9. [17]

    Rogers: Laminar flow analysis

    D. Rogers: Laminar flow analysis. Cambridge University Press, 1992

    1992

  10. [18]

    Sherman: On connecting large vessels to small

    T. Sherman: On connecting large vessels to small. The meaning of Murray’ s law . The Journal of General Physiology. 78 (4): 431–453 (1981). doi: 10.1085/jgp.78.4.431

    1981 doi

  11. [19]

    Stephenson, A

    D. Stephenson, A. Patronis, D. Holland and D. Lockerby: Generalizing Murray’s law: An optimization principle for fluidic networks of arbitrary shape and scale . J. Appl. Phys. 118:174302 (2015)

    2015

  12. [20]

    Whitaker: Flow in porous media I: A theoretical derivation of Darcy’s l aw

    S. Whitaker: Flow in porous media I: A theoretical derivation of Darcy’s l aw. Trans- port in porous media, 1 (1986), pp. 3–25

    1986

  13. [21]

    Zheng et al: Bio-inspired Murray materials for mass transfer and activi ty

    X. Zheng et al: Bio-inspired Murray materials for mass transfer and activi ty. Nature Communications 8:14921 (2017)

    2017

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.