# Random Spin Networks & Generalized Diffusion

Low-frequency magnetic noise often looks deceptively simple: a spectrum close to an inverse power of frequency. A microscopic explanation must account for the exponent, its range of validity, and the way a real probe couples to the fluctuating degrees of freedom. Agarwal, Demler, and Martin studied this problem using random Heisenberg spin networks. Their work connects collective disordered dynamics to noise, while leaving open which microscopic defects dominate any particular device.

### Figure to read

Noise and dynamical exponents across random-bond spin-chain regimes. [Figure 1 in the source paper](https://arxiv.org/pdf/1506.00643#page=2). Agarwal, Kartiek; Demler, Eugene; Martin, Ivar. *$1/f^\alpha$ noise and generalized diffusion in random Heisenberg spin systems*, Phys. Rev. B 92, 184203 (2015). Figure linked rather than reproduced; the archived manuscript does not provide an explicit open reproduction license.

Separate zero- and high-temperature results, and notice the parameter region inaccessible to the renormalization method. The changing bond bias tests collective dynamics rather than a universal noise exponent.

### From fluctuators to interacting clusters

One familiar way to produce broad noise is to combine many independent fluctuators with a wide distribution of relaxation times. An interacting spin network offers a different possibility: the broad hierarchy can emerge from collective dynamics and a distribution of exchange couplings. Spins strongly coupled to one another form effective objects on lower energy scales. The remaining degrees of freedom then interact through renormalized couplings. A real-space renormalization approach tracks that hierarchy. [1](https://doi.org/10.1103/PhysRevB.92.184203)

The important methodological feature of the 2015 work is that the probe is renormalized along with the Hamiltonian. A sensor does not observe every spin with equal strength. Its geometry defines a weighted operator, and the weight carried by an effective cluster changes as the network is coarse-grained. Computing the energy spectrum alone would therefore be insufficient to predict the measured noise. This is a useful principle well beyond the particular spin model.

### What the spectrum says

The authors relate the dynamical structure factor at finite wavevector to generalized diffusion and find linked low- and high-frequency power laws in the studied regimes. Their analysis treats zero and high temperature separately, because the available transitions and statistical weights differ. For the spin-chain settings examined, the low-frequency behavior can produce inverse-frequency exponents between one-half and one. That is a model result with stated regimes, not a universal exponent for every source of flux noise. [1](https://doi.org/10.1103/PhysRevB.92.184203)

A helpful schematic is to write a probe signal as a weighted sum of spin components. Its spectrum then contains both diagonal fluctuations and cross-correlations between distinct spins. Those cross-correlations are exactly where collective dynamics can distinguish itself from an independent-fluctuator picture. Changing the sensor geometry or distance changes the weighting, potentially exposing information that a single frequency trace cannot isolate.

### Connections to rare regions and localization

The broader Griffiths literature explains why distributed time scales arise near disordered transitions. However, a random-bond Heisenberg network is not interchangeable with the random-field XXZ chain used in the anomalous-diffusion study. Symmetries, conservation laws, dimensionality, and the renormalization approximation all matter. The appropriate comparison is between mechanisms and observables, rather than transferring a phase diagram from one model to another. [2](https://doi.org/10.1103/PhysRevB.93.134206) [3](https://doi.org/10.1103/PhysRevLett.114.160401)

The network calculation also has geometric limitations. The renormalization procedure is more controlled in the one-dimensional or effectively narrow settings analyzed than in general two-dimensional networks. The authors explicitly discuss the difficulty of extending it. That limitation is important when invoking the work to interpret spins distributed on a real material surface, where connectivity may be neither a simple chain nor a clean two-dimensional lattice. [1](https://doi.org/10.1103/PhysRevB.92.184203)

### The experimental landscape

Quantum-sensing methods provide several ways to vary a probe's spectral response, while scanning or height-dependent measurements change its spatial sensitivity. Agarwal and collaborators' later two-dimensional magnetometry theory illustrates how such spatial filtering can be calculated explicitly for current noise. The same experimental philosophy—vary the coupling kernel rather than collecting only one spectrum—helps distinguish candidate spin-noise mechanisms. [4](https://doi.org/10.1103/RevModPhys.89.035002) [5](https://doi.org/10.1103/PhysRevB.95.155107)

The principal unresolved issue is microscopic identification. A measured power law could reflect interacting spins, activated defects, environmental processes, or overlapping mechanisms. Useful tests would compare temperature dependence, probe geometry, field dependence, and deviations from a pure power law. A model that explains all of those with consistent parameters is substantially more informative than one that fits a single exponent. Read this page alongside [local magnetometry](https://knowen.org/nodes/33550): both emphasize that a noise spectrum is jointly determined by the material and the operator through which it is observed.

### References

1. Agarwal, Kartiek; Demler, Eugene; Martin, Ivar. [$1/f^\alpha$ noise and generalized diffusion in random Heisenberg spin systems](https://doi.org/10.1103/PhysRevB.92.184203). Phys. Rev. B 92, 184203 (2015). [Open manuscript](https://arxiv.org/abs/1506.00643).
2. Gopalakrishnan, Sarang; Agarwal, Kartiek; Demler, Eugene; Huse, David A.; Knap, Michael. [Griffiths effects and slow dynamics in nearly many-body localized systems](https://doi.org/10.1103/PhysRevB.93.134206). Phys. Rev. B 93, 134206 (2016). [Open manuscript](https://arxiv.org/abs/1511.06389).
3. Agarwal, Kartiek; Gopalakrishnan, Sarang; Knap, Michael; Mueller, Markus; Demler, Eugene. [Anomalous diffusion and Griffiths effects near the many-body localization transition](https://doi.org/10.1103/PhysRevLett.114.160401). Phys. Rev. Lett. 114, 160401 (2015). [Open manuscript](https://arxiv.org/abs/1408.3413).
4. Degen, C. L.; Reinhard, F.; Cappellaro, P.. [Quantum sensing](https://doi.org/10.1103/RevModPhys.89.035002). Reviews of Modern Physics 89, 035002 (2017). [Open manuscript](https://arxiv.org/abs/1611.02427).
5. Agarwal, Kartiek; Schmidt, Richard; Halperin, Bertrand; Oganesyan, Vadim; Zaránd, Gergely; Lukin, Mikhail D.; Demler, Eugene. [Magnetic noise spectroscopy as a probe of local electronic correlations in two-dimensional systems](https://doi.org/10.1103/PhysRevB.95.155107). Phys. Rev. B 95, 155107 (2017). [Open manuscript](https://arxiv.org/abs/1608.03278).

*Independent research synthesis. Literature checked 5 September 2026; preprints are identified in the references.*

<!-- research-map:kartiek-context-2026-09-v2:spinnoise -->

# Parents

* Disorder, Localization & Slow Quantum Dynamics
* Quantum Sensing, Spectroscopy & Information Diagnostics