WorksPhysics and Astronomy
Universal Statistics of Random Waves in Multi-Port non-Hermitian Scattering Systems
Abstract
We study the statistical properties of wave transport of a fixed multi-port non-Hermitian scattering system through wavefront shaping to explore all possible coherent illumination conditions. This is a departure from traditional approaches utilizing cavity shaping to create an ensemble of systems with different detailed scattering arrangements that lead to uncorrelated field configurations. Using microwave measurements in the low channel-number limit, we empirically demonstrate universal statistics in the probability density functions (PDF) of reflectivity, transmissivity, and absorptivity, Fisher information and the difference between forward propagating and time reversed transmissivity. Because the PDFs depend only on the number of input channels and the eigenvalues of Hermitian operators, we demonstrate how cavity shaping can be used to manipulate these eigenvalues, even creating Diabolic Point degeneracies, giving us direct control over the shapes and bounds of transport phenomena PDFs. We further investigate the PDFs when the excitations are finite bandwidth pulses as opposed to monochromatic waves, and determine the behavior in the limits of two channels and many channels. Our results establish a new paradigm in statistical investigations and cavity shaping applications by focusing on outcomes arising from all possible excitations, rather than creating conditions for one specific wavefront that optimally achieves a desired purpose.
Introduction
The study of wave propagation in complex multi-scattering media and ray-chaotic billiards has been quantified through statistical descriptions of observable properties such as transmission, reflection, and conductance, informed by theoretical approaches based on random matrix theory (RMT) [1, 2, 3, 4, 5] and maximum entropy principles [6]. Historically these studies have been focused on creating a random ensemble of such systems that differ in detail but share the properties of a single universality class. The extreme sensitivity of the wave properties to sub-wavelength boundary or scatterer perturbations serves to create a sampling of random wave configurations within their interior. Examples include reverberant chambers for electromagnetic compatibility studies with random plane waves [7, 8], the use of tuning electrodes to change the shape of quantum dots to sample different values of conductance [9], the use of variable phase shifters to alter wave interference at the nodes of metric graphs [10, 11, 12, 13, 14, 15], and the scrambling of scatterers in a wave scattering medium [16, 17]. The statistical properties studied in such systems include eigenvalue [18, 19, 20, 21] and eigenfunction [22, 23, 24] statistical properties, as well as the properties of impedance and scattering matrices and derived quantities such as conductance [25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39].
In addition to variations in the structure of scattering systems, one can also employ variations in the frequency of injected waves, in the limit that the system is large compared to the wavelength and has complex ray trajectories [40, 41, 42]. This effective equivalence of shape-change and spectral variation is fundamental to many statistical studies of complex systems [43]. Other parametric quantities can be utilized to scramble wave properties as well, such as a magnetic field to alter trajectories of charged particles within the system [44, 45, 46, 47].
The time-honored paradigm is that single-realizations of a complex system will be dominated by system-specific and realization-specific details that are not of interest because they cannot be universal. We now call this assumption into question and examine the statistical properties of single realizations of arbitrary scattering systems. Remarkably, there are clear quantitative and universal statistical properties that have eluded experimental examination up to now, and we will present them here.
The statistical questions that we now ask concern the outcome of illuminating a given scattering system with all possible randomly-selected input waves, and considering the range of possible outputs in terms of reflection, transmission, and absorption. In the past, wavefront shaping has been used almost exclusively for identifying one specific excitation that achieves a particular purpose or goal, such as maximizing transmission through a strongly-scattering system [48, 49, 50, 51, 52, 53, 54], focusing or depositing energy at a location within a system [55, 56, 57, 58, 59, 60], or maximizing absorption [61, 62, 63, 64, 65, 66].
We test the prediction of Wang and Guo [67], which upsets this paradigm by using wavefront shaping to explore every possible input excitation and examine the statistical properties of the corresponding output states. The prediction is that if an input wavefront on channels results in an observable , being a Hermitian operator, then the probability density function of is a B-spline piecewise polynomial of degree with knots at the eigenvalues of . We find that the predicted distribution functions are universal, in the sense that their shape is independent of the microscopic configuration and global properties of the system, including the degree of loss (incorporating non-Hermitian scattering systems). Further, these predictions hold for systems both large compared to the wavelength (such as those studied through ensemble statistics), and small compared to the wavelength.
Results—Our experiments are conducted in the microwave scattering regime, where we have the advantage of easily accessing the low channel-number limit, where the PDF of any observable most dramatically change with and are not well approximated by a Gaussian, as opposed to what is seen in the many channel limit [68]. We explore the predicted PDFs using microwave graph networks with different topologies [10, 12, 13, 64, 69, 70, 71, 72, 73, 74], and a ray-chaotic quarter bowtie billiard [22, 43, 47, 66, 75, 76, 77, 78, 79]. The scattering matrices we measure can be block-decomposed as
where is and corresponds to reflection at the “input” channels, and which is and corresponds to transmission from the “input” channels to the “output” channels. In our experimental systems, all the channels are treated on equal footing, so which ones are used for input or output is an arbitrary choice. For a given system, we can vary from 2 to , or keep fixed and change which channels are in the set of inputs.
Measurements are performed using an channel, single source Vector Network Analyzer (VNA), and another channel source VNA. With the first VNA, we can measure scattering matrices that are up to . With the second VNA, we can measure up to , but we can also inject a wavefront with independent amplitude and phase control on channels. We then measure the amplitude and phase of waves leaving the cavity on all channels, with the wavefront returning on the input channels being and the wavefront transmitted to the “output” channels is . The scalar reflectivity , transmissivity , and absorptivity of a wavefront are calculated via:
A schematic of this measurement process is shown in Figure 1(g), with system details suppressed. See Supp. Mat. section S1 for a complete discussion and schematics of the systems measured in this work. To arrive at an estimate of the PDF of these quantities, we repeat the injection process for approximately 100,000 input wavefronts with amplitudes randomly chosen from the Dirichlet distribution and phases randomly chosen from a uniform distribution.
Figure 1. (a-c) Experimentally determined PDFs (orange) of reflectivity from an port microwave billiard in the cases of number of input ports . Black curves depict theoretical prediction of B-spline (piecewise polynomial) with knots at the eigenvalues of . The location of the eigenvalues are marked in vertical dashed lines in ascending order: red, green, blue, pink. (d-e) PDFs of transmissivity (blue) and absorptivity (green) from the same measurement as panel (b). (f) Experimental PDF of Fisher information (FI) (purple) of a voltage applied to a metasurface embedded within the port microwave billiard. For all panels, the billiard was measured at 8 GHz. (g) schematic of measurement process.
Figure Figure 1 shows statistical PDFs for resulting output quantities measured for a fixed configuration of the microwave billiard at a fixed frequency for a broad set of input wavefronts. In Figure 1(a–Figure 1c), we show that the experimentally determined PDFs of reflectivity (orange areas) match the prediction of a B-spline (black curve) with knots at the eigenvalues of the reflectance matrix (colored dashed lines) for the cases of [67]. For brevity, we only show the PDFs of transmissivity and absorptivity for the case of in panels (d) and (e), which come from the same measurement as panel (b). Panel (d) shows that the transmittance matrix has two degenerate eigenvalues at 0. This is due to there being eigenvalues but for , at most only eigenvalues can be non-zero and in this case . Notably, this Diabolic Point degeneracy [80, 81] skews the PDF, and in non-trivial cases () we can create such degeneracies to further control the PDFs, which we demonstrate in the Supp. Mat. Sec. S2. Panel (e) demonstrates that the billiard is a highly absorptive system, in which every possible incident wavefront is over 90% absorbed. Measured PDFs from a much lower loss microwave graph are available in Supp. Mat. Sec. S3.
In panel (f), we demonstrate the universality of this result beyond the simple transport observables of , , and . We do this by tuning the voltage applied to a globally biased varactor diode-loaded metasurface [77, 82] between and . We then form the Fisher information operator (FIO) [79, 83]. The Fisher information returned by a wavefront can be experimentally determined from
where we calculate from the finite difference . If measurement of the full scattering matrix is not possible, the Fisher information operator can also be formed using or , and the FI calculated from and . The PDF of FI shown in Figure 1(f) also agrees with the predicted universal B-spline form.
It is known that the eigenvector associated with the largest eigenvalue of the Fisher information operator is the optimal input for maximizing the returned information about the associated parameter [79, 83]. In this billiard, we find that there are orders of magnitude between the eigenvalues of (, , and ). If we ask how much information will be returned by a random input wavefront, we see that for this situation there is an 83.8% probability of the returned FI being within an order of magnitude of the optimal amount. In Ref. [79], it was found that the optimal wavefront was not required to achieve certain tasks, just a wavefront associated with Fisher information above some threshold, which for this same billiard and metasurface was within an order of magnitude of the largest eigenvalue. These two results together suggest that an arbitrary wavefront could be used with good odds of accomplishing the desired task. However, this would not be true in the case of large as the bulk of the PDF becomes solely defined by the interior eigenvalues [67], and so most wavefronts would return FI much less than the largest eigenvalue, and the useful wavefronts would reside in the tail of the probability density function.
More observables that display PDFs with this universal behavior are provided in Appendix § 2, with a figure demonstrating some examples. We emphasize at this point that these experimental verifications of the theory in Ref. [67] demonstrate that knowledge of only the eigenvalues of a Hermitian operator is sufficient to completely describe the statistical outcome of wave excitations, but does not determine what happens to a particular wavefront.
Past work on cavity shaping has been mostly limited to creating conditions required for a specific wavefront to demonstrate various phenomena, such as reflectionless scattering [84, 85, 86, 87, 88], Coherent Perfect Extinction [89, 90], and Coherent Perfect Absorption [47, 65, 66, 77, 78, 91, 92], etc. Some works have demonstrated CPA for all wavefronts with a single cavity, but this utilizes a designed, degenerate cavity that is not reconfigurable [93, 94]. With perfect knowledge of the statistical outcome of all wave excitations deduced from knowledge of the scattering matrix , we introduce a new paradigm of cavity shaping based on optimizing the PDF of the observable of interest such that a majority of input wavefront exhibits a desired behavior. We provide an example of this new paradigm in the context of minimizing and maximizing the absorption experienced by all waves entering a scattering system.
Panels (a-c) of Figure 2 are PDFs of absorptivity that come from experimental ensembles of microwave networks with the topology of the Petersen graph. Each individual realization of the graph in these ensembles has the same total electrical length of 3.92 m, with the differences coming from small (millimeter to single centimeter) changes to six of the fifteen bond lengths accomplished through programmable analog phase shifters [74, 95]. Therefore the uniform attenuation rate of each realization should be the same, with only the details of the wave interferences at the nodes of the graph changing. We see from the pairs of distributions in Figure 2 that just by making these small changes, we are able to, at arbitrary frequency, completely change the degree to which all waves are absorbed.
Figure 2. Absorptivity PDFs demonstrating the new cavity shaping paradigm. Panels (a-c) are from microwave Pettersen graphs with channels. In each panel, both PDFs are formed from data taken at the same frequencies (7.106 GHz, 7.791 GHz, and 7.136 GHz respectively), with the only differences being small local perturbations to the lengths of some of the bonds of the graph such that the overall length is fixed. One of the PDFs shows the lowest average absorptivity measured while the other shows the highest. (d) Data from the port microwave billiard at 8 GHz where the average absorptivity was maximized and minimized through an optimization function.
Panel (a) of Figure 2 comes from an port Petersen graph at 7.106 GHz. The two solid green areas show the B-spline PDF of for the realization with the smallest average and largest average . Panels (b) and (c) show the same things but for an port Petersen graph at 7.791 GHz and an port Petersen graph at 7.136 GHz, respectively. Note that an optimization algorithm was not utilized to arrive at these cavity configurations, they are simply random settings of the phase shifters taken from large ensembles used for traditional cavity-shaping statistical analysis. In contrast, panel (d) shows two PDFs of from the quarter bowtie billiard at 8 GHz for which the average was optimized using a surrogate optimizer to have the smallest and largest value possible [66]. In the case of largest average , we were able to configure the system such that there was a 90% probability of a random wavefront being at least 95% absorbed. However, because the billiard is so absorptive, the dynamic range of the absorptivity matrix eigenvalues is much smaller with a hard lower bound, and even with optimization we were unable to get any configuration that had any waves that would be absorbed less than 70%.
In Supp. Mat. Sec. S2, we also show PDFs from the same ensembles and frequencies as panels (a-c) where instead of looking at the average , we consider the variance of . These results show the utility and adaptability of this new form of cavity shaping, as not just the average or variance, but also the bounds (smallest and largest eigenvalue) on any observable can be controlled. We also show examples of cavity shaping to create degeneracies in the eigenvalues which are Diabolic Points (DP), not Exceptional Points, because the operators are Hermitian. These are actually easier and more common to create than scattering singularities such as CPA [66, 90, 95], which further argues that cavity shaping to control scattering outcome PDFs is advantaged over standard methods. There are many relevant tasks, such as maximizing or minimizing a transport observable [47, 48, 49, 50, 51, 52, 53, 54, 65, 66, 77, 78, 87, 88, 91, 92] or focusing energy at a location [60, 79, 83, 96, 97], for which the optimal wavefront is the eigenvector of a Hermitian operator associated with it’s largest eigenvalue. For these tasks, we suggest cavity shaping such that a high percentage, if not all incident wavefronts, can accomplish the task within an acceptable tolerance.
Taken a step further, it should be possible to cavity shape a system such that all inputs achieve multiple desired objectives, such as specified values of reflectance, transmittance, and absorptance simultaneously, as was done for a single input in a fixed system in Ref. [98].
There are, however, important limitations on what can be accomplished with passive complex scattering systems [99], and one already discussed is the degree of homogeneous loss of the scattering system. Intrinsic homogeneous loss puts a lower bound on and an upper bound on and (we predict also on ), whereas inhomogeneous loss in the form of an attenuator, lumped element, or unmonitored open channel, can be avoided to varying degrees by different wavefronts. In general, lumped loss could lead to a wider dynamic range for the absorptivity matrix eigenvalues. The coupling strength of the channels to the cavity modes can also limit the effectiveness of any cavity shaping efforts, as imperfect coupling puts a lower bound on (and consequently an upper bound on and ).
Now we consider the case of exciting a scattering system with a finite bandwidth pulse as opposed to a monochromatic wave. The resulting PDF of any observable becomes the sum of the PDFs at each frequency within the bandwidth. We highlight the extreme case of and then discuss what happens as increases.
For , the PDF of any observable produced by a monochromatic wave has been shown to be a uniform distribution. For a finite bandwidth pulse, we find that there is still a region of the PDF that is uniform, bounded below (above) by the largest (smallest) value of (). in the frequency band. If , then the PDF develops a sharp point, and no uniform region remains. These two cases are shown in Figure 3, panels (a) and (c), both of which have PDFs of from an experimental channel, six node complete bipartite graph , with the bandwidth of the pulse being one mean mode spacing MHz. The center frequency for both is 15.55 GHz, but the graph itself has been modified between the two panels through the use of programmable analog phase shifters to change multiple bond lengths.
Figure 3. (a) and (c), PDFs of transmissivity for two input channels and channel complete bipartite microwave graph where the input waves include frequencies covering a complete mean mode spacing centered around GHz. (b), six input channel PDF of reflectivity for the same graph as panel (c), showing the change from single frequency (solid curve) to a bandwidth of (dashed curve). The overlapping region of the two PDFs is colored pink, and the difference is plotted as a blue curve. (d) ten input channel PDF of reflectivity from an RMT simulation, demonstrating that as gets larger, the change from single frequency to bandwidth PDFs becomes negligible.
As the number of input channels grows, the effect of having a finite bandwidth excitation lessens. In panel (b), we show the single frequency PDF of , , outlined in a solid curve from the same graph and center frequency as panel (c), and the frequency bandwidth PDF outlined in a dashed curve. Both PDFs are colored orange, and their overlap is represented in pink. The difference is shown as a blue curve corresponding to the right y-axis. Panel (d) shows a similar plot, but this time taken from an channel Random Matrix Theory (RMT) generated scattering matrix, using the algorithm outlined in Ref. [95]. Already at , the single frequency and finite bandwidth PDFs are nearly indistinguishable. Since the single frequency PDFs converge to a Gaussian in the large limit, and the sum of Gaussians is also a Gaussian, we predict that finite bandwidth PDFs at large converge to the PDF of the center frequency of the excitation band.
**Conclusions—**In this Letter, we have empirically demonstrated universal statistical features in the observables of a fixed system, establishing a new paradigm in the statistical study of multi-port Non-Hermitian scattering systems. Our work is in the interesting low- limit, and we demonstrate universality in transmissivity , reflectivity , absorptivity , as well as generalizing to other observables such as Fisher information. We not only validate the predictions of Wang and Guo [67], but further generalize them to finite bandwidth excitations, and show the effects on the PDFs in the limits of and large . We show how controlling the eigenvalues of a Hermitian operator through cavity shaping methods can completely change how the system responds to the set of all input wavefronts. With clever tuning, including controlling the extremal eigenvalues, manipulating the eigenvalue mean, and creating Diabolic Point degeneracies, one can create a situation in which most, if not all, incident wavefronts yield the desired response. These results provide guidance for new cavity and medium shaping techniques and objective functions.
Acknowledgements We are grateful to Prof. Cheng Guo (UT Austin) for valuable discussions. This work was partially supported by NSF/RINGS under grant No. ECCS-2148318, ONR under grant N000142312507, and DARPA WARDEN under grant HR00112120021.
Universality Extended to Other Observables
In Ref. [67], Wang and Guo apply their theory to , , and , but claim it extends naturally to other measured observables. We demonstrated that this is true for Fisher information, and in this section suggest more observables with statistics described by the B-spline prediction. First, we consider how non-reciprocal or time-reversed transmission differs from the “forward propagating” transmission in the case of . This is related to the concept of transmission time delay difference introduced in Refs. [74, 100], but generalized to sub-blocks of the scattering matrix as opposed to individual elements. By applying the identical wavefront on the input channels and then the output channels, we can measure the “forward” transmissivity and the “reversed” transmissivity and calculate the difference .
Figure 4(a) shows from the same graph as Figure 3(c) in blue ( is identical), and in yellow. This graph was perfectly reciprocal (), yet for a generic wavefront we see a range for the difference . Only in the trivial case of would and be generically identical. However, we do find that for reciprocal systems the mean of will always be 0, though as shown, it does not necessarily have to be symmetric about zero. Panel (b) shows experimental data where we performed the measurement on a 2.38 meter long non-reciprocal microwave network with the topology of the Petersen graph with . The blue areas are the PDFs of and , their overlap shown in pink, and the yellow area is . Because this graph is non-reciprocal, as can be inferred from , the mean of is not zero, but instead . One can also consider , though in general no relationship between and can be assumed so no special features, such as a mean of zero, can be attributed to .

Figure 4. (a) PDF of transmissivity (teal) and of non-reciprocal / difference from time reversed transmissivity (yellow) from the same graph and at the same single frequency GHz as Figure 3(a-b). (b) Experimental demonstration of non-reciprocal transmissivity in a 4 channel Petersen graph incorporating a microwave circulator that is 2.38 meter long measured at 1.2715 GHz. PDFs of and are indicated by the light blue areas bounded by the solid and dashed black curves, their overlap is marked in pink, and is marked in yellow. (c-d) PDFs of real (blue) and imaginary (red) time delays experienced by a wavefront injected into the 3 and 4 channel Petersen graphs from Figure 2.
In Hermitian scattering, the Wigner-Smith operator (WSO): [101, 102, 103, 104], and the generalized Wigner-Smith operator (GWSO): [96, 105, 106] are Hermitian operators, and their respective observables of time delay [107, 108] and momentum (when is a displacement) [109, 110] can be measured for an ensemble of wavefronts . B-Spline PDFs for were suggested in the Supplemental Material of Ref. [110], and it is not unreasonable to expect the same for .
Perhaps more interestingly, in non-Hermitian scattering where and are not Hermitian, the real part of time delay can become negative, and there is an imaginary component as well [74, 111, 112, 113]. We then predict that the PDFs of and will each individually have the B-spline form, and these are quantities that can be independently measured in the time domain through pulse propagation [114, 115]. In panels (c) and (d) of Figure 4, we show the expected PDFs of and for the and channel Petersen graphs featured in Figure 2. Note that the values of are truly unbounded, though we know there is a power-law suppression of extreme positive and negative values of both and [74]. While the interpretation of the complex GWSO and what the observable can be measured is less well understood, we expect the PDFs of and to also share the B-spline form.
Supplementary Material
Nadav Shaibe, Jared M. Erb, and Steven M. Anlage
Experimental Details
In this section, we provide a more complete discussion of the multiport wave systems measured, schematics for which are provided in Figure S1. Various different microwave graphs were used in this work, but all were built using standard coaxial SMA cables for the bonds and three way tee-junctions for the nodes. To connect a node of the graph to the Vector Network Analyzer (VNA), we attach two tee-junctions together to make a four way junction. Two different graph topologies were used, the first being the Petersen graph and the second being the bipartite graph.
Figure S1. (a) Schematic of Petersen graph with 6 variable phase shifters (green lines), one of which is labeled . The internal nodes are labeled as blue dots, and the port nodes connected to the VNA are labeled with red dots. (b) Schematic of Petersen graph with a circulator on one of the nodes, marked in purple, which introduces scattering non-reciprocity. (c) Schematic of the bipartite graph with two variable phase shifters labeled in green and two of the ports having circulators. (d) Schematic of quarter bowtie billiard with three metasurfaces labeled , , and .
The Petersen graph has ten vertices and fifteen edges. It is useful for our purposes as it is a cubic graph (each vertex having degree three) which we can build with tee-junctions, and it is the smallest cubic graph that can be formed such that we can separate the nodes into “internal nodes”, ones not connected to the VNA, and “port nodes”, connected to the VNA, and not have any port node directly connect to any other port node. This gives more opportunities for wave interference before the excitations leave the system, providing greater tunability and more complex scattering. A schematic with six variable phase shifters as was used for Figure 2 and Figure 4(c-Figure 4d) is shown in Figure S1, and panel (b) has a schematic of the Petersen graph with a circulator on one of the nodes to break reciprocity, which was used for Figure 4(b). Note that while we label four ports in panels (a) and (b), some of the Petersen graph measurements were done with two or three ports. For those measurements, the extra unused port nodes are replaced with single tee-junctions are become internal nodes.
The graph is the complete bipartite graph with six vertices and nine edges. The vertices can be separated into two sets, and , such that a vertex in set is connected to all the vertices in set and to no other vertices in . The data in Figure 3(a-Figure 3c) and Figure 4(a) comes from a graph with two phase shifters ( and ) and ports. Two of the ports have circulators instead of tee-junctions. A schematic of this graph is shown in Figure S1.
We also measure a ray-chaotic quarter bowtie billiard with four ports and three tunable metasurfaces, shown in Figure S1. The metasurfaces are globally biased with an externally generated DC voltage that can be controlled to the level. The metasurfaces are comprised of 18 unit cell varactor diode-loaded elements arranged linearly, with each cell being 1 cm long. As the bias voltage applied to a metasurface is increased, the effective capacitance of the elements decreases, changing both the amplitude and phase of the reflection spectrum. Because the billiard has a perimeter of 1.574 meters, each metasurface occupies about 11% of the wall space, and within their tunability range of 8 – 12 GHz provide a great deal of control over the billiard scattering matrix.
Cavity Shaping and Diabolic Points
In this section, we show more examples of cavity shaping a system such that an observable of interest has the desired PDF. In Figure 2(a-Figure 2c) of the main text, we showed how the average absorptivity can be controlled so that a system with a fixed uniform attenuation rate can be made more or less absorbing. In Figure S2, we demonstrate in the same systems, how the variance of can be controlled to make a system have a very large or small range of possible responses to an input excitation. We calculate the variance of the PDFs as the variance of the eigenvalues divided by .
Figure S2. Showing an example of the new cavity shaping paradigm by controlling the variance of absorptivity . PDFs are formed from absorptivity matrix eigenvalues taken from experimental data of microwave Petersen graphs with total length 3.92 m. Top (bottom) row shows the PDFs from the phase shifter configurations with the largest (smallest) variance of . Panels (a) and (d) come from the channel Petersen graph measured at 7.106 GHz, and have and respectively. Panels (b) and (e) come from the channel Petersen graph measured at 7.791 GHz, and have and respectively. Panels (c) and (f) come from the channel Petersen graph measured at 7.136 GHz, and have and respectively.
Panels (a) and (d) of Figure S2 show the PDFs of the port, 3.92 m Petersen graph measured at 7.106 GHz, as in Figure 2(a). Six of the fifteen bonds of the graph are comprised of programmable analog phase shifters that can be tuned to have an effective electrical length between 25 and 30 cm. By ensuring the net length change of any realization is zero, we can create an ensemble of graphs with a nominally fixed loss parameter. However, sub-centimeter changes of the lengths of a subset of the bonds of the graph can change the absorptivity matrix eigenvalues from panel (a)’s 0.18 and 0.97 to panel (d)’s 0.31 and 0.34. This completely changes the absorption experienced by any waves entering this graph. In panels (b) and (e), we show the same things but for the port Petersen graph at 7.791 GHz, and panels (c) and (f) have the PDFs from the port Petersen graph measured at 7.136 GHz.
Note that in the top row of Figure S2, because the eigenvalues are bounded above by 1 and below by the minimum absorption possible in the system, which for this graph is about 15%, the configurations with the largest variances not only have the largest range of absorption response, but also cover almost the full range of the physically possible absorption. We reiterate that these distributions were found without optimization, instead they were identified by choosing the best realizations of an ensemble with the fixed length constraint.
For Figure S3, we let an optimizer function vary three voltage-controlled metasurfaces within the port quarter bowtie billiard in order to create the desired condition, as we did in Figure 2(d). All the measurements were performed at 8 GHz. In panels (a) and (b), we maximize and minimize the variance of , respectively. In panels (c) and (d), we create Diabolic Point (DP) degeneracies of two of absorptivity eigenvalues, which has a profound effect on the shape of the PDF. Because the piecewise polynomials that define the PDF meet at the eigenvalues, by having degenerate eigenvalues we reduce the number of spans (B-spline segments).
Figure S3. PDFs of absorptivity from the port microwave billiard at 8 GHz. For each panel, an optimization function was used to create the condition given in the text: (a) largest variance of equal to 0.0035, (b) smallest variance of equal to , (c) , and (d) . The cases of largest and smallest average for the same system at the same frequency are shown in Figure 2(d).
Further, at a knot, a B-spline is times differentiable where is the degree and is the multiplicity of the knot. If there is DP degeneracy of order , then the PDF is actually only times continuously differentiable at the degenerate eigenvalues, as opposed to the times continuously differentiable at a knot corresponding to a unique eigenvalue. That’s why there is a cusp in Figure S3, as in this case and the PDF cannot be differentiated at the knot corresponding to the degenerate eigenvalues and .
The cavity shaping demonstrations in this section all focused on absorptivity, but the principal is general and will hold for any observable. Absorptivity was chosen because loss is present in most physical systems of interest and therefore it is almost always of interest and concern to control it.
More Experimental PDFs
In this section, we present more experimentally generate PDFs where we perform multi-source injection and directly measure the outgoing waves. This section proves that the theory does not just hold for highly absorptive systems such as the quarter bowtie billiard used in the main text. We also show the experimental PDFs in two different situations, large and small variance, demonstrating that the cavity shaping ideas discussed in the main text and Sec. S2 are sound.
The experimental platform used in this section is the port, input channel Petersen graph with a total electrical length of 2.38 m measured for the data in Figure 4(b), and shown in Figure S1. No phase shifters are included in this graph, so it is not tunable, but by measuring at different frequencies we can see different distributions for the eigenvalues of , , and . In the top row of Figure S4, we show , , and
Figure S4. (a-c) Experimental PDFs of reflectivity , transmissivity and absorptivity of a port, input channel microwave Petersen graph measured at 1.2715 GHz. (d-e) The same PDFs as the row above, but now measured at 1.1283 GHz. The locations of the eigenvalues of , , and are marked in vertical dashed lines in ascending order: red, green, blue.
measured at 1.2715 GHz. At this frequency, the variance of both and is relatively large, with and . However, when we measure at 1.1283 GHz (Figure S4 bottom row), which is just over two mean mode spacings away, we find the variance of and particularly have decreased significantly. Notably, doesn’t change as much between panels (c) and (f). The center eigenvalue changes from to , but the extremal eigenvalues hardly move: and .
This tells us that purely spectral shaping (changing frequencies) is less effective for controlling the PDF of than cavity shaping. This makes sense as cavity shaping should have a much more pronounced effect on the interference conditions of wave excitations within the system. Destructive interference at the channels and constructive interference at sources of loss are instrumental for increasing the absorptivity experience by a wavefront without directly tuning the lossiness of the system. While spectral shaping of course also effects wave interference by changing the wavelength, it is in a much smoother and global manner which offers less control. Therefore with spectral shaping one is much more constrained by the intrinsic attenuation rate of the system.
References
- [1]F. J. Dyson, Statistical theory of the energy levels of complex systems. i, Journal of Mathematical Physics 3, 140–156 (1962).
- [2]T. A. Brody, J. Flores, J. B. French, P. A. Mello, A. Pandey, and S. M. Wong, Random-matrix physics: spectrum and strength fluctuations, Reviews of Modern Physics 53, 385–479 (1981).DOI
- [3]O. Bohigas, M. J. Giannoni, and C. Schmit, Characterization of chaotic quantum spectra and universality of level fluctuation laws, Physical Review Letters 52, 1 (1984).DOI
- [4]C. W. J. Beenakker, University in the random-matrix theory of quantum transport, Physical Review Letters 70, 1155–1158 (1993).
- [5]H. Li, S. Suwunnarat, R. Fleischmann, H. Schanz, and T. Kottos, Random matrix theory approach to chaotic coherent perfect absorbers, Phys. Rev. Lett. 118, 044101 (2017).DOI
- [6]P. A. Mello, P. Pereyra, and T. H. Seligman, Information-theory and statistical nuclear-reactions .1. general-theory and applications to few-channel problems, Annals of Physics 161, 254–275 (1985).
- [7]D. A. Hill, Electromagnetic fields in cavities : deterministic and statistical theories, IEEE Press series on electromagnetic wave theory (IEEE, Piscataway, N.J, 2009).
- [8]R. Serra, A. C. Marvin, F. Moglie, V. M. Primiani, A. Cozza, L. R. Arnaut, Y. Huang, M. O. Hatfield, M. Klingler, and F. Leferink, Reverberation chambers a la carte: An overview of the different mode-stirring techniques, IEEE Electromagnetic Compatibility Magazine 6, 63–78 (2017).
- [9]C. M. Marcus, A. J. Rimberg, R. M. Westervelt, P. F. Hopkins, and A. C. Gossard, Conductance fluctuations and chaotic scattering in ballistic microstructures, Physical Review Letters 69, 506–509 (1992).
- [10]B. Dietz, V. Yunko, M. Białous, S. Bauch, M. Ławniczak, and L. Sirko, Nonuniversality in the spectral properties of time-reversal-invariant microwave networks and quantum graphs, Phys. Rev. E 95, 052202 (2017).
- [11]A. Rehemjiang, M. Richter, U. Kuhl, and H.-J. Stöckmann, Spectra and spectral correlations of microwave graphs with symplectic symmetry, Phys. Rev. E 97, 022204 (2018).
- [12]J. Che, J. Lu, X. Zhang, B. Dietz, and G. Chai, Missing-level statistics in classically chaotic quantum systems with symplectic symmetry, Phys. Rev. E 103, 042212 (2021).
- [13]J. Che, X. Zhang, W. Zhang, B. Dietz, and G. Chai, Fluctuation properties of the eigenfrequencies and scattering matrix of closed and open unidirectional graphs with chaotic wave dynamics, Phys. Rev. E 106, 014211 (2022).
- [14]M. Ławniczak, A. Akhshani, O. Farooq, M. Białous, S. Bauch, B. Dietz, and L. Sirko, Distributions of the wigner reaction matrix for microwave networks with symplectic symmetry in the presence of absorption, Phys. Rev. E 107, 024203 (2023).
- [15]M. Ławniczak, A. Akhshani, O. Farooq, S. Bauch, and L. Sirko, Experimental distributions of the reflection amplitude for networks with unitary and symplectic symmetries, Acta Physica Polonica A 144, 469 (2024).
- [16]S. Zhang, Y. Lockerman, and A. Z. Genack, Mesoscopic speckle, Phys. Rev. E 82, 051114 (2010).DOI
- [17]J. Wang and A. Z. Genack, Transport through modes in random media, Nature 471, 345–351 (2011).
- [18]H. J. Stockmann and J. Stein, Quantum chaos in billiards studied by microwave-absorption, Physical Review Letters 64, 2215–2218 (1990).pubmed.ncbi.nlm.nih.gov/10041617
- [19]H. D. Graf, H. L. Harney, H. Lengeler, C. H. Lewenkopf, C. Rangacharyulu, A. Richter, P. Schardt, and H. A. Weidenmuller, Distribution of eigenmodes in a superconducting stadium billiard with chaotic dynamics, Physical Review Letters 69, 1296–1299 (1992).
- [20]Y. Fyodorov, On the density of complex eigenvalues of wigner reaction matrix in a disordered or chaotic system with absorption, Acta Physica Polonica A 144, 447 (2023).
- [21]Y. V. Fyodorov, M. A. Skvortsov, and K. S. Tikhonov, Resonances in a single-lead reflection from a disordered medium: σ-model approach, Annals of Physics 460, 169568 (2024).
- [22]S.-H. Chung, A. Gokirmak, D.-H. Wu, J. S. A. Bridgewater, E. Ott, T. M. Antonsen, and S. M. Anlage, Measurement of wave chaotic eigenfunctions in the time-reversal symmetry-breaking crossover regime, Phys. Rev. Lett. 85, 2482 (2000).DOI
- [23]U. Kuhl, O. Legrand, and F. Mortessagne, Microwave experiments using open chaotic cavities in the realm of the effective hamiltonian formalism, Fortschritte der Physik 61, 404 (2013).
- [24]S. Ma, T. M. Antonsen, and S. M. Anlage, Eigenfunction and eigenmode-spacing statistics in chaotic photonic crystal graphs, Phys. Rev. E 106, 054215 (2022).
- [25]R. Méndez-Sánchez, U. Kuhl, M. Barth, C. Lewenkopf, and H.-J. Stöckmann, Distribution of reflection coefficients in absorbing chaotic microwave cavities, Physical Review Letters 91, 174102 (2003).
- [26]Y. V. Fyodorov and D. V. Savin, Statistics of impedance, local density of states, and reflection in quantum chaotic systems with absorption, Journal of Experimental and Theoretical Physics Letters 80, 725 (2004).
- [27]U. Kuhl, M. Martínez-Mares, R. A. Méndez-Sánchez, and H.-J. Stöckmann, Direct processes in chaotic microwave cavities in the presence of absorption, Phys. Rev. Lett. 94, 144101 (2005).
- [28]S. Hemmady, X. Zheng, E. Ott, T. M. Antonsen, and S. M. Anlage, Universal impedance fluctuations in wave chaotic systems, Phys. Rev. Lett. 94, 014102 (2005).
- [29]S. Hemmady, X. Zheng, T. M. Antonsen, E. Ott, and S. M. Anlage, Universal statistics of the scattering coefficient of chaotic microwave cavities, Phys. Rev. E 71, 056215 (2005).
- [30]S. Hemmady, J. Hart, X. Zheng, T. M. Antonsen, E. Ott, and S. M. Anlage, Experimental test of universal conductance fluctuations by means of wave-chaotic microwave cavities, Phys. Rev. B 74, 195326 (2006).
- [31]Y. V. Fyodorov, D. V. Savin, and H.-J. Sommers, Scattering, reflection and impedance of waves in chaotic and disordered systems with absorption, Journal of Physics A: Mathematical and General 38, 10731 (2005).
- [32]D. Savin, Y. Fyodorov, and H.-J. Sommers, Correlation functions of impedance and scattering matrix elements in chaotic absorbing cavities, Acta Physica Polonica A 109, 5364 (2006).DOI
- [33]J.-H. Yeh, J. A. Hart, E. Bradshaw, T. M. Antonsen, E. Ott, and S. M. Anlage, Experimental examination of the effect of short ray trajectories in two-port wave-chaotic scattering systems, Phys. Rev. E 82, 041114 (2010).DOI
- [34]J.-H. Yeh, J. A. Hart, E. Bradshaw, T. M. Antonsen, E. Ott, and S. M. Anlage, Universal and nonuniversal properties of wave-chaotic scattering systems, Phys. Rev. E 81, 025201 (2010).pubmed.ncbi.nlm.nih.gov/20365617
- [35]M. Ławniczak, S. Bauch, O. Hul, and L. Sirko, Experimental investigation of the enhancement factor for microwave irregular networks with preserved and broken time reversal symmetry in the presence of absorption, Phys. Rev. E 81, 046204 (2010).pubmed.ncbi.nlm.nih.gov/20481804
- [36]S. Kumar, A. Nock, H.-J. Sommers, T. Guhr, B. Dietz, M. Miski-Oglu, A. Richter, and F. Schäfer, Distribution of scattering matrix elements in quantum chaotic scattering, Phys. Rev. Lett. 111, 030403 (2013).DOI
- [37]A. Nock, S. Kumar, H.-J. Sommers, and T. Guhr, Distributions of off-diagonal scattering matrix elements: Exact results, Annals of Physics 342, 103 (2014).DOI
- [38]M. Ławniczak and L. Sirko, Investigation of the diagonal elements of the wigner’s reaction matrix for networks with violated time reversal invariance, Scientific Reports 9, 5630 (2019).DOI
- [39]Y. V. Fyodorov and D. V. Savin, Random matrix theory for chaotic wave scattering and transport (2026), arXiv:2606.10957 [nlin.CD].arxiv.org/abs/2606.10957
- [40]D. A. Hill, Electronic mode stirring for reverberation chambers, Electromagnetic Compatibility, IEEE Transactions on 36, 294–299 (1994).DOI
- [41]Z. Xing, T. Antonsen, and E. Ott, Statistics of impedance and scattering matrices in chaotic microwave cavities: Single channel case, Electromagnetics 26, 33p (2006).
- [42]S. Hemmady, X. Zheng, E. Ott, T. M. Antonsen, and S. M. Anlage, Universal impedance fluctuations in wave chaotic systems, Physical Review Letters 94, 014102 (2005).DOI
- [43]S. Ma, T. M. Antonsen, S. M. Anlage, and E. Ott, Short-wavelength reverberant wave systems for physical realization of reservoir computing, Phys. Rev. Res. 4, 023167 (2022).DOI
- [44]C. W. J. Beenakker, Random-matrix theory of quantum transport, Rev. Mod. Phys. 69, 731 (1997).DOI
- [45]F. Castañeda-Ramírez, J. L. Martínez-Argüello, T. Hofmann, A. Rehemanjiang, M. Martínez-Mares, J. A. Méndez-Bermúdez, U. Kuhl, and H.-J. Stöckmann, Microwave graph analogs for the voltage drop in three-terminal devices with orthogonal, unitary, and symplectic symmetry, Phys. Rev. E 105, 014202 (2022).DOI
- [46]X. Zhang, W. Zhang, J. Che, and B. Dietz, Experimental test of the rosenzweig-porter model for the transition from poisson to gaussian unitary ensemble statistics, Phys. Rev. E 108, 044211 (2023).DOI
- [47]J. Erb, N. Shaibe, R. Calvo, D. P. Lathrop, T. M. Antonsen, T. Kottos, and S. M. Anlage, Topology and manipulation of scattering singularities in complex non-Hermitian systems: Two-channel case, Phys. Rev. Res. 7, 023090 (2025).DOI
- [48]I. M. Vellekoop and A. P. Mosk, Universal optimal transmission of light through disordered materials, Phys. Rev. Lett. 101, 120601 (2008).DOI
- [49]M. Kim, Y. Choi, C. Yoon, W. Choi, J. Kim, Q.-H. Park, and W. Choi, Maximal energy transport through disordered media with the implementation of transmission eigenchannels, Nature Photonics 6, 581 (2012).
- [50]H. Yu, T. R. Hillman, W. Choi, J. C. Lee, M. S. Feld, R. R. Dasari, and Y. Park, Measuring large optical transmission matrices of disordered media, Phys. Rev. Lett. 111, 153902 (2013).pubmed.ncbi.nlm.nih.gov/24160602
- [51]B. Gérardin, J. Laurent, A. Derode, C. Prada, and A. Aubry, Full transmission and reflection of waves propagating through a maze of disorder, Phys. Rev. Lett. 113, 173901 (2014).DOI
- [52]S. M. Popoff, A. Goetschy, S. F. Liew, A. D. Stone, and H. Cao, Coherent control of total transmission of light through disordered media, Phys. Rev. Lett. 112, 133903 (2014).DOI
- [53]H. Cao, A. P. Mosk, and S. Rotter, Shaping the propagation of light in complex media, Nature Physics 18, 994–1007 (2022).DOI
- [54]K. Y. Bliokh, E. Karimi, M. J. Padgett, M. A. Alonso, M. R. Dennis, A. Dudley, A. Forbes, S. Zahedpour, S. W. Hancock, H. M. Milchberg, S. Rotter, F. Nori, . K. Özdemir, N. Bender, H. Cao, P. B. Corkum, C. Hernández-García, H. Ren, Y. Kivshar, M. G. Silveirinha, N. Engheta, A. Rauschenbeutel, P. Schneeweiss, J. Volz, D. Leykam, D. A. Smirnova, K. Rong, B. Wang, E. Hasman, M. F. Picardi, A. V. Zayats, F. J. Rodríguez-Fortuño, C. Yang, J. Ren, A. B. Khanikaev, A. Alù, E. Brasselet, M. Shats, J. Verbeeck, P. Schattschneider, D. Sarenac, D. G. Cory, D. A. Pushin, M. Birk, A. Gorlach, I. Kaminer, F. Cardano, L. Marrucci, M. Krenn, and F. Marquardt, Roadmap on structured waves, Journal of Optics 25, 103001 (2023).DOI
- [55]I. M. Vellekoop and A. P. Mosk, Focusing coherent light through opaque strongly scattering media, Opt. Lett. 32, 2309 (2007).pubmed.ncbi.nlm.nih.gov/17700768
- [56]X. Cheng and A. Z. Genack, Focusing and energy deposition inside random media, Opt. Lett. 39, 6324 (2014).DOI
- [57]R. Horstmeyer, H. Ruan, and C. Yang, Guidestar-assisted wavefront-shaping methods for focusing light into biological tissue, Nature Photonics 9, 563 (2015).pubmed.ncbi.nlm.nih.gov/27293480
- [58]I. M. Vellekoop, Feedback-based wavefront shaping, Opt. Express 23, 12189 (2015).DOI
- [59]M. Durand, S. M. Popoff, R. Carminati, and A. Goetschy, Optimizing light storage in scattering media with the dwell-time operator, Phys. Rev. Lett. 123, 243901 (2019).DOI
- [60]R. McIntosh, A. Goetschy, N. Bender, A. Yamilov, C. W. Hsu, H. Yılmaz, and H. Cao, Delivering broadband light deep inside diffusive media, Nature Photonics 18, 744–750 (2024).DOI
- [61]Y. D. Chong, L. Ge, H. Cao, and A. D. Stone, Coherent perfect absorbers: Time-reversed lasers, Physical Review Letters 105, 053901 (2010).DOI
- [62]W. Wan, Y. Chong, L. Ge, H. Noh, A. D. Stone, and H. Cao, Time-reversed lasing and interferometric control of absorption, Science 331, 889 (2011).DOI
- [63]D. G. Baranov, A. Krasnok, T. Shegai, A. Alù, and Y. Chong, Coherent perfect absorbers: linear control of light with light, Nature Reviews Materials 2, 17064 (2017).DOI
- [64]L. Chen, T. Kottos, and S. M. Anlage, Perfect absorption in complex scattering systems with or without hidden symmetries, Nature Communications 11, 5826 (2020).DOI
- [65]P. del Hougne, K. B. Yeo, P. Besnier, and M. Davy, On-demand coherent perfect absorption in complex scattering systems: Time delay divergence and enhanced sensitivity to perturbations, Laser & Photonics Reviews 15, 2000471 (2021).DOI
- [66]J. Erb, N. Shaibe, and S. M. Anlage, Creation and control of scattering singularities in non-hermitian systems (2026), arXiv:2609.13085 [physics.optics].
- [67]Y. Wang and C. Guo, Probability distribution for coherent transport of random waves, Newton 2, 100478 (2026).arxiv.org/abs/2511.04602
- [68]M. Davy, Z. Shi, J. Wang, and A. Z. Genack, Transmission statistics and focusing in single disordered samples, Opt. Express 21, 10367 (2013).DOI
- [69]T. Kottos and U. Smilansky, Quantum chaos on graphs, Phys. Rev. Lett. 79, 4794 (1997).
- [70]O. Hul, S. Bauch, P. Pakoński, N. Savyt skyy, K. Życzkowski, and L. Sirko, Experimental simulation of quantum graphs by microwave networks, Phys. Rev. E 69, 056205 (2004).DOI
- [71]O. Hul, O. Tymoshchuk, S. Bauch, P. M. Koch, and L. Sirko, Experimental investigation of wigner's reaction matrix for irregular graphs with absorption, Journal of Physics A: Mathematical and General 38, 10489 (2005).DOI
- [72]M. Lawniczak, S. Bauch, O. Hul, and L. Sirko, Experimental investigation of properties of hexagon networks with and without time reversal symmetry, Physica Scripta 2009, 014050 (2009).DOI
- [73]M. Lawniczak, S. Bauch, O. Hul, and L. Sirko, Experimental investigation of microwave networks simulating quantum chaotic systems: the role of direct processes, Physica Scripta 2012, 014018 (2012).DOI
- [74]N. Shaibe, J. M. Erb, and S. M. Anlage, Superuniversal statistics of complex time delays in non-Hermitian scattering systems, Phys. Rev. Lett. 134, 147203 (2025).DOI
- [75]P. So, S. M. Anlage, E. Ott, and R. N. Oerter, Wave chaos experiments with and without time reversal symmetry: GUE and GOE statistics, Phys. Rev. Lett. 74, 2662 (1995).pubmed.ncbi.nlm.nih.gov/10057986
- [76]D. H. Wu, J. S. A. Bridgewater, A. Gokirmak, and S. M. Anlage, Probability amplitude fluctuations in experimental wave chaotic eigenmodes with and without time-reversal symmetry, Phys. Rev. Lett. 81, 2890 (1998).DOI
- [77]J. Erb, D. Shrekhenamer, T. Sleasman, T. Antonsen, and S. Anlage, Control of the scattering properties of complex systems by means of tunable metasurfaces, Acta Physica Polonica A 144, 421 (2024).DOI
- [78]J. Erb, N. Shaibe, T. Kottos, and S. M. Anlage, Robust wave splitters based on scattering singularities in complex non-Hermitian systems, Applied Physics Letters 128, 071105 (2026).arxiv.org/abs/2504.18775
- [79]N. Shaibe, J. Erb, and S. M. Anlage, Measurement and optimal targeting of a hidden scatterer in a complex environment utilizing fisher information (2026), arXiv:2608.19442 [physics.optics].arxiv.org/abs/2608.19442
- [80]M. Yang, H.-Q. Zhang, Y.-W. Liao, Z.-H. Liu, Z.-W. Zhou, X.-X. Zhou, J.-S. Xu, Y.-J. Han, C.-F. Li, and G.-C. Guo, Realization of exceptional points along a synthetic orbital angular momentum dimension, Science Advances 9, eabp8943 (2023).arxiv.org/abs/2209.07769
- [81]J. Chen, W. Liu, J. Wang, C. Li, R.-Y. Zhang, X. Cui, F. Guan, L. Shi, J. Zi, and C. T. Chan, Continuous lines of topological singularities in metasurface scattering matrices: From nodal to exceptional, ACS Photonics 12, 3208 (2025).DOI
- [82]T. Sleasman, R. Duggan, R. S. Awadallah, and D. Shrekhenamer, Dual-resonance dynamic metasurface for independent magnitude and phase modulation, Phys. Rev. Appl. 20, 014004 (2023).DOI
- [83]D. Bouchet, S. Rotter, and A. P. Mosk, Maximum information states for coherent scattering measurements, Nature Physics 17, 564 (2021).DOI
- [84]W. R. Sweeney, C. W. Hsu, and A. D. Stone, Theory of reflectionless scattering modes, Phys. Rev. A 102, 063511 (2020).DOI
- [85]M. F. Imani, D. R. Smith, and P. del Hougne, Perfect absorption in a disordered medium with programmable meta-atom inclusions, Advanced Functional Materials 30, 2005310 (2020).DOI
- [86]A. D. Stone, W. R. Sweeney, C. W. Hsu, K. Wisal, and Z. Wang, Reflectionless excitation of arbitrary photonic structures: a general theory, Nanophotonics 10, 343 (2021).DOI
- [87]J. Sol, A. Alhulyami, A. D. Stone, and P. del Hougne, Reflectionless programmable signal routers, Science Advances 9, eadf0323 (2023).DOI
- [88]X. Jiang, S. Yin, H. Li, J. Quan, H. Goh, M. Cotrufo, J. Kullig, J. Wiersig, and A. Alù, Coherent control of chaotic optical microcavity with reflectionless scattering modes, Nature Physics 20, 109 (2024).DOI
- [89]C. Guo, J. Li, M. Xiao, and S. Fan, Singular topology of scattering matrices, Phys. Rev. B 108, 155418 (2023).DOI
- [90]A. H. Alhulyami, N. Pyvovar, P. del Hougne, O. D. Miller, and A. D. Stone, Coherent control of wave scattering via coincidences of complex spectra (2025), arXiv:2510.13035 [physics.optics].
- [91]K. Pichler, M. Kühnmayer, J. Böhm, A. Brandstötter, P. Ambichl, U. Kuhl, and S. Rotter, Random anti-lasing through coherent perfect absorption in a disordered medium, Nature 567, 351 (2019).DOI
- [92]B. W. Frazier, T. M. Antonsen, S. M. Anlage, and E. Ott, Wavefront shaping with a tunable metasurface: Creating cold spots and coherent perfect absorption at arbitrary frequencies, Phys. Rev. Res. 2, 043422 (2020).DOI
- [93]Y. Slobodkin, G. Weinberg, H. Hörner, K. Pichler, S. Rotter, and O. Katz, Massively degenerate coherent perfect absorber for arbitrary wavefronts, Science 377, 995 (2022).DOI
- [94]H. Hörner, L. Wild, Y. Slobodkin, G. Weinberg, O. Katz, and S. Rotter, Coherent perfect absorption of arbitrary wavefronts at an exceptional point, Phys. Rev. Lett. 133, 173801 (2024).DOI
- [95]N. Shaibe, J. M. Erb, and S. M. Anlage, Superuniversal statistics with topological origins for non-Hermitian scattering singularities, Phys. Rev. Res. 7, 043185 (2025).DOI
- [96]P. Ambichl, A. Brandstötter, J. Böhm, M. Kühnmayer, U. Kuhl, and S. Rotter, Focusing inside disordered media with the generalized Wigner-Smith operator, Phys. Rev. Lett. 119, 033903 (2017).DOI
- [97]P. del Hougne, R. Sobry, O. Legrand, F. Mortessagne, U. Kuhl, and M. Davy, Experimental realization of optimal energy storage in resonators embedded in scattering media, Laser & Photonics Reviews 15, 2000335 (2021).arxiv.org/abs/2001.04658
- [98]S. Li, D. Kim, S. Fan, and C. Guo, Joint control of coherent transmission, reflection, and absorption, Phys. Rev. B 112, 245421 (2025).DOI
- [99]C. Guo and S. Fan, Passivity constraints on the relations between transmission, reflection, and absorption eigenvalues, Phys. Rev. B 110, 205431 (2024).arxiv.org/abs/2410.04565
- [100]L. Chen, I. L. Giovannelli, N. Shaibe, and S. M. Anlage, Asymmetric transmission through a classical analogue of the Aharonov-Bohm ring, Phys. Rev. B 110, 045103 (2024).DOI
- [101]E. P. Wigner, Lower limit for the energy derivative of the scattering phase shift, Phys. Rev. 98, 145 (1955).
- [102]F. T. Smith, Lifetime matrix in collision theory, Phys. Rev. 118, 349 (1960).
- [103]P. W. Brouwer, K. M. Frahm, and C. W. J. Beenakker, Quantum mechanical time-delay matrix in chaotic scattering, Phys. Rev. Lett. 78, 4737 (1997).DOI
- [104]F. D. Cunden, F. Mezzadri, N. Simm, and P. Vivo, Correlators for the Wigner–Smith time-delay matrix of chaotic cavities, Journal of Physics A: Mathematical and Theoretical 49, 18LT01 (2016).DOI
- [105]M. Horodynski, M. Kühmayer, A. Brandstötter, K. Pichler, Y. V. Fyodorov, U. Kuhl, and S. Rotter, Optimal time delay fields for micromanipulation in complex scattering environments, Nature Photonics 14, 149 (2020).
- [106]P. del Hougne, K. B. Yeo, P. Besnier, and M. Davy, Coherent wave control in complex media with arbitrary wavefronts, Phys. Rev. Lett. 126, 193903 (2021).DOI
- [107]U. Smilansky, Delay-time distribution in the scattering of time-narrow wave packets. (i), J. Phys. A: Math. Theor. 50, 215301 (2017).DOI
- [108]U. Smilansky and H. Schanzе, Delay-time distribution in the scattering of time-narrow wave packets (ii)—quantum graphs, J. Phys. A: Math. Theor. 51, 075302 (2018).DOI
- [109]B. Orazbayev, M. Malléjac, N. Bachelard, S. Rotter, and R. Fleury, Wave-momentum shaping for moving objects in heterogeneous and dynamic media, Nature Physics 20, 1441 (2024).DOI
- [110]T. Nerson, J. Hüpfl, C. Ferise, D. Globosits, M. Hudler, M. Malléjac, S. Rotter, and R. Fleury, Multiobjective tweezers in scattering media, Phys. Rev. Res. 8, 023254 (2026).DOI
- [111]M. Asano, K. Y. Bliokh, Y. P. Bliokh, A. G. Kofman, R. Ikuta, T. Yamamoto, Y. S. Kivshar, L. Yang, N. Imoto, Ş. K. Özdemir, and F. Nori, Anomalous time delays and quantum weak measurements in opti-micros-resonators, Nature Communications 7, 13488 (2016).DOI
- [112]L. Chen, S. M. Anlage, and Y. V. Fyodorov, Generalization of Wigner time delay to subunitary scattering systems, Phys. Rev. E 103, L052003 (2021).DOI
- [113]L. Chen and S. M. Anlage, Use of transmission and reflection complex time delays to reveal scattering matrix poles and zeros: Example of the ring graph, Physical Review E 105, 054210 (2022).DOI
- [114]I. L. Giovannelli and S. M. Anlage, Physical interpretation of imaginary time delay, Phys. Rev. Lett. 135, 043801 (2025).DOI
- [115]I. L. Giovannelli, S. M. Anlage, and T. M. Antonsen, Chirped pulse analysis and control in non-hermitian scattering systems using complex time delay, Journal of Physics: Photonics 8, 035006 (2026).DOI






