DMRadio-Core: A new approach for GUT-scale axion searches
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: High Energy Physics: Experiment. The day in high-energy experiment: detectors, searches and limits.
Kai: We're looking at the paper: "DMRadio-Core: A new approach for GUT-scale axion searches".
Mira: Here's the gist of the paper. Searches for QCD axions with masses in the neV/c2 mass range are strongly motivated by new physics at the GUT scale and by well-motivated pre-inflationary axion symmetry breaking scales.
Kai: In this paper, we propose a new experimental geometry based on a narrow-bore, segmented solenoid that optimizes the collection of the axion-induced signal using LC resonators outside the high-field region of the magnet bore.
Mira: This alternative optimization significantly reduces the required stored magnetic energy while preserving sensitivity, thus enabling a near-term experiment in the 30–200 MHz (120–830 neV/c2) range.
Kai: Foundational Concepts of the Core Geometry and DMRADIO The primary figure of merit of resonant axion haloscopes is the scan rate: the rate at which the experiment tunes its resonance frequency such that it can achieve sensitivity to axions at a given axion-photon coupling gaγγ and a given signal-to-noise ratio (SNR).
Paper summary: Kai: So we're looking at this paper called "DMRadio-Core: A new approach for GUT-scale axion searches". It seems like the main thing they're proposing is a new geometry for how you actually build these experiments to look for axions.
Mira: Right. They start by saying that searching for QCD axions with masses in the neV over c2 range is really motivated because of new physics at the grand unified theory scale and also by some pre-inflationary axion symmetry breaking scales.
Lev: So, what's their big claim about this DMRadio-Core thing? Is it just a slight tweak to an existing setup?
Kai: It's more than that. They propose a new experimental geometry based on a narrow-bore, segmented solenoid. The main goal is to optimize the collection of the axion signal using LC resonators placed outside the high-field region of the magnet bore.
Mira: What does this optimization actually achieve in terms of practical results? They claim this alternative approach significantly reduces the required stored magnetic energy while still keeping sensitivity high.
Lev: That sounds like it tackles a major engineering hurdle, which is always something when you move from theory to actual hardware. So, how are they doing that optimization mathematically?
Kai: They reformulate the equation for scan rate by using quantifiable experimental measurements. They use finite element method simulations to get numbers for things like the volume of the pickup, Q factor, and effective length.
Mira: And what are some of the key design choices they highlight in that optimization process? What makes this geometry work better than a standard setup?
Kai: First, they use an enclosed structure that intersects the magnetic field lines produced by a solenoid. They also bring in bulk electron shuttling, which is something described in appendix D of reference forty <ref:2604.16602#pg3>.
Lev: Electron shuttling—that’s about driving physical electron currents on the walls of the pickup from the high magnetic field region into where the DC magnetic field is low. So, it's using that low-field space to help drive current flow?
Kai: Exactly. And they segment these pickups by stacking them axially inside a solenoid that is also segmented axially to add coherently induced signals. It’s about making sure you get those coherent signals across the stack of pickups.
Mira: Okay, so moving into the actual experiment, what does the DMRadio-Core setup look like in practice? What are the concrete specifications they've designed?
Kai: They describe a specific setup: it uses a five tesla magnet with a thirty-seven centimeter inner bore, and the magnet is two point nine five meters tall and segmented.
Lev: And what kind of axions are they targeting with this specific configuration? Is it general, or tied to a specific model?
Paper summary: Kai: They state that this setup is sensitive to DFSZ axions in the three zero to two hundred mega hertz frequency range, which corresponds to masses between one hundred twenty and eight hundred neV over c2.
Mira: That's a specific mass range, which gives us a target for the physics we're trying to probe with this experiment. How does this compare to other existing experiments?
Kai: They say that the magnet for DMRadio-Core contains an order of magnitude less stored energy than the magnet used in the DMRadio-m3 experiment, yet both experiments can achieve comparable sensitivity to DFSZ axions.
Lev: So, they're trading a lot of magnetic energy for a new way to collect the signal, and still hitting that same sensitivity benchmark. That's interesting overhead management.
Mira: Then there's also the DMRadio-GUT experiment, which is presented as the ultimate low-mass axion search within their series. What are its parameters?
Kai: DMRadio-GUT is sensitive to axions in the one hundred kilohertz to thirty mega hertz range, which means masses between zero point four and one hundred twenty neV over c2 for DFSZ sensitivity.
Lev: And what's the scale of that setup compared to DMRadio-Core? Are we talking about similar magnet sizes or much smaller ones?
Kai: DMRadio-GUT uses a ninety centimeter bore, an eight point five meter long, and an eighteen tesla magnet to achieve a live scan time of two point two years.
Mira: So the comparison here is really about scaling down while maintaining sensitivity across different mass regimes. What's the biggest limitation they admit for these experiments?
Kai: They mention that some pickup sizes can lead to prohibitively slow scan rates or require tuning parameters that are just unphysical for those frequency ranges, so they omit those specific axion searches.
Lev: So it sounds like the method has practical boundaries based on what you can actually build and tune in real life. Where does the geometry stop working well?
Kai: The Core geometry itself is designed to decouple the size of the pickup from the physical extent of the magnet, which is a big step compared to traditional geometries.
Mira: So, when we look at this paper, what's their overall vision for where this kind of experiment goes next? What are they hoping this paves?
Kai: They present DMRadio-Core as a pathfinder towards a large-scale realization of an experiment that can achieve sensitivity to DFSZ axions in the one hundred kilohertz to thirty mega hertz range, which is what DMRadio-GUT aims for.
Lev: So they see this current setup as a stepping stone rather than the final destination, focusing on proving the concept first.
Kai: That's right. They're using this new geometry to explore unexplored parameter space while simultaneously building towards that larger experiment goal for lower mass axions.
Conclusion: Kai: So we've been looking at this new design called DMRadio-Core for searching for those low-mass QCD axions, and now we're coming to the conclusion of what that actually means for our search strategy.
Mira: Yeah, so they’ve just shown how you can optimize the physical setup to make it work better with less magnetic energy while still catching those signals.
Kai: It seems like the authors are really pushing this geometry because it opens up a whole new way to think about the required hardware for these experiments.
Lev: From my side, I’m looking at that reduced stored energy as a huge plus for building something feasible on current technology, which is what we need to move forward in error correction hardware.
Mira: Exactly, because the assumptions they're making about how you arrange those pickups outside the high field region are really critical to whether this actually translates into a working experiment.
Kai: And it's worth remembering that they’re aiming for a near-term test in the three zero to two hundred mega hertz range, which is right around those one hundred twenty to eight hundred neV over c2 masses we were talking about earlier.
Lev: That frequency window is tight, so if the geometry doesn't deliver that signal strength they need, then the entire experiment just stalls because tuning becomes too hard.
Mira: It seems like this paper isn't just about building one specific machine; it’s more of a blueprint for how to approach low-mass axion searches in general.
Kai: Right, it’s a pathfinder, which means they aren't claiming this is the final answer, but they are laying down the groundwork for what bigger experiments could look like down the road.
Lev: So we’re looking at DMRadio-Core as a proof of concept that might eventually lead to something much bigger with even lower mass sensitivity.
Mira: Right, so next time we talk about this, we'll take a closer look at how these different magnet configurations actually stack up in terms of the live scan times they mentioned earlier.
Department of Physics Stanford University · Kavli Institute for Particle Astrophysics and Cosmology Stanford University · SLAC National Accelerator Laboratory · Department of Physics Princeton University · Laboratory for Nuclear Science Massachusetts Institute of Technology · Department of Nuclear Engineering University of California Berkeley · Physics Division Lawrence Berkeley National Laboratory Applied Science and Technology College of Engineering University of California Berkeley Accelerator Technology and Applied Physics Division Lawrence Berkeley National Laboratory
hep-ex, hep-ph, physics.ins-det
Submitted: 2026-04-17
Updated: 2026-10-07
Comments: 10 pages, 6 figures. v3: Corrected a scan time for DMRadio-GUT. Updated version to match journal version
Journal ref: Phys. Rev. D 114, 092004 (2026)
DOI: 10.1103/rwv6-8hr3
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 98/100
The gist: Searches for QCD axions with masses in the neV/c2 mass range are strongly motivated by new physics at the GUT scale and by well-motivated pre-inflationary axion symmetry breaking scales.
Key concepts
- Scan Rate
- This is how fast an experiment tunes its resonance frequency to find axions. A higher scan rate allows the experiment to achieve a better signal-to-noise ratio (SNR) for a given axion coupling, which is crucial for sensitive searches.
- Core Geometry
- This geometry separates the location of the signal pickup from the physical size of the magnet bore. Instead of placing pickups inside a large magnet, this design allows them to be placed in regions with lower static magnetic fields while still capturing axion signals.
- Bulk Electron Shuttling (BES)
- BES is a technique used to drive physical electron currents on the walls of the pickup. This helps move these currents from areas of high magnetic field into regions where the DC magnetic field is low, improving signal collection in the new geometry.
Terminology
Summary
Searches for QCD axions with masses in the neV/c2 mass range are strongly motivated by new physics at the GUT scale and by well-motivated pre-inflationary axion symmetry breaking scales. p. 0 Yi,3, 2 and B. Zemenu1, 3, 2
1Department of Physics, Stanford University, Stanford, CA 94305
2Kavli Institute for Particle Astrophysics and Cosmology, Stanford University, Stanford, CA 94305
3SLAC National Accelerator Laboratory, Menlo Park, CA 94025
4Department of Physics, Princeton University, Princeton, NJ 08544
5Laboratory for Nuclear Science, Massachusetts Institute of Technology, Cambridge, MA 02139
6Department of Nuclear Engineering, University of California, Berkeley, Berkeley, CA 94720
7Physics Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720
8Department of Physics, University of California, Berkeley, Berkeley, CA 94720
9Applied Science and Technology, College of Engineering,
University of California, Berkeley, Berkeley, CA 94720
10Accelerator Technology and Applied Physics Division,
Lawrence Berkeley National Laboratory, Berkeley, CA 94720
(Dated: October 9, 2026)
Searches for QCD axions with masses in the neV/c2 mass range are strongly motivated by new
physics at the GUT scale and by well-motivated pre-inflationary axion symmetry breaking scales. This parameter space is challenging to probe due to the small axion–photon couplings, which typically require large, high-field magnets with substantial stored energy. In this paper, we propose
a new experimental geometry based on a narrow-bore, segmented solenoid that optimizes the collection of the axion-induced signal using LC resonators outside the high-field region of the magnet
bore. In this paper, we propose
a new experimental geometry based on a narrow-bore, segmented solenoid that optimizes the collection of the axion-induced signal using LC resonators outside the high-field region of the magnet
bore. This alternative optimization significantly reduces the required stored magnetic energy while
preserving sensitivity, thus enabling a near-term experiment in the 30–200 MHz (120–830 neV/c2
)
range, with a cost-effective, staged scaling to a GUT-scale experiment in the 100 kHz–30 MHz
(0.Where the paper says it
Where the paper says it
Where the paper says it
Foundational Concepts of the Core Geometry and DMRADIO
The primary figure of merit of resonant axion haloscopes is the scan rate: the rate at which the experiment tunes its resonance frequency such that it can achieve sensitivity to axions at a given axion-photon coupling gaγγ and a given signal-to-noise ratio (SNR). In conventional solenoidal geometries, the pickup is located within the magnet bore, and therefore the sensitivity is limited by the bore size and field. p. 0 In this paper, we propose
a new experimental concept, DMRadio-Core, designed
as a scalable, near-term implementation that explores an
alternative geometry optimization: reducing the magnet
bore size while employing an extended structure that couples to the ac signal induced outside the high-field region
(see Fig. In conventional solenoidal geometries, the
pickup is located within the magnet bore, and therefore
the sensitivity is limited by the bore size and field. If the external shield, and hence the pickup volume, can
be expanded such that the boundary conditions are enforced at larger distances, the experimental sensitivity to
the dark matter signals increases. The Core geometry decouples the size of the pickup
from the physical extent of the magnet.Where the paper says it
Where the paper says it
Optimization of the Core Geometry
The optimization process involves reformulating the equation of scan rate in terms of quantifiable experimental measurements. The FEM simulations provide numerical results for |V |, Q, and Leff for a given magnet and pickup design. Key aspects of the pickup design include:
-
Employing an enclosed structure which intersects the magnetic field lines produced by a solenoid. p. 2
Where the paper says it
Specifically, these simulations have shown that an optimized geometry involves a fully enclosed pickup that penetrates through the core of a solenoid. Pickup design The pickup shown in Figure 3a employs an enclosed structure which intersects the magnetic field lines produced by a solenoid.
-
Utilizing bulk electron shuttling (BES) as derived in Appendix D of [40] to drive physical electron currents on the walls of the pickup from the region of high magnetic field into the regions where the dc magnetic field is low. p. 2
Where the paper says it
Pickup design The pickup shown in Figure 3a employs an enclosed structure which intersects the magnetic field lines produced by a solenoid. The effective axion current drives physical electron currents on the walls of the pickup from the region of high magnetic field into the regions where the dc magnetic field is low.
-
Segmenting pickups by stacking them axially in a solenoidal magnet that is also segmented axially to add coherently induced signals. p. 3
Where the paper says it
However, to avoid cavity modes within the target frequencies of an experiment, it is instead beneficial to segment the pickups by stacking them axially in a solenoidal magnet that is also segmented axially. This allows for the axion induced signal to be added coherently across multiple structures.
The DMRadio-Core Experiment
The DMRadio-Core experiment utilizes a 5 T, 37 cm inner bore, 2.95 m tall segmented solenoidal magnet and is sensitive to DFSZ axions in the 30–200 MHz frequency range corresponding to masses of 120–830 neV/c2. p. 7 This experiment utilizes
a 5 T, 37 cm inner bore, 2. 95 m tall segmented solenoidal
magnet and is sensitive to DFSZ axions in the 30–200
MHz frequency range corresponding to masses of 120–
830 neV/c
2
. This experiment utilizes
a 5 T, 37 cm inner bore, 2. 95 m tall segmented solenoidal
magnet and is sensitive to DFSZ axions in the 30–200
MHz frequency range corresponding to masses of 120–
830 neV/c
2
. 35 m height. As such, the
magnet for DMRadio-Core contains an order of magnitude less stored energy than the magnet of DMRadio-m3
,
while allowing the two experiments to achieve comparable sensitivity to DFSZ axions. The magnitude of the induced current grows at larger
radii within the bore of the magnet and can be understood through the effect of BES as each magnetic field line contributes to
additional current that will flow radially outwards within the structure. The pickup tuning elements and readout components
are connected to a slit near the interface region between
the niobium pickup and the copper plates.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
The Core Geometry for DMRADIO-GUT
DMRadio-GUT is the ultimate low-mass axion experiment within the DMRadio series with sensitivity to axions in the 100 kHz–30 MHz frequency range, corresponding to 0.4–120 neV/c2 masses at DFSZ sensitivity. p. 6 THE CORE GEOMETRY FOR
DMRADIO-GUT
DMRadio-GUT is the ultimate low-mass axion experiment within the DMRadio series with sensitivity to axions in the 100 kHz–30 MHz frequency range, corresponding to 0. 4–120 neV/c2 masses at DFSZ sensitivity, as
shown in Figure 7. The
listed times do not take into account operations time or time
dedicated to rescans. the use of three room-temperature soft iron segments reduces the dc field to Bdc < 170 mT in the region outside
the coils such that the entire region is below the Hc1 of
niobium. For SNR = 3, the scan rate corresponds to a live
scan time of 0. THE CORE GEOMETRY FOR
DMRADIO-GUT
DMRadio-GUT is the ultimate low-mass axion experiment within the DMRadio series with sensitivity to axions in the 100 kHz–30 MHz frequency range, corresponding to 0.Where the paper says it
Where the paper says it
Where the paper says it
Conclusion
Using this geometry we have presented the design of the DMRadio-Core experiment. This experiment utilizes a 5 T, 37 cm inner bore, 2. p. 7 Using this geometry we have presented the design of
the DMRadio-Core experiment. This experiment utilizes
a 5 T, 37 cm inner bore, 2. 95 m tall segmented solenoidal
magnet and is sensitive to DFSZ axions in the 30–200
MHz frequency range corresponding to masses of 120–
830 neV/c
2
. In addition to its science goals covering unexplored parameter space, this experiment also acts as a
pathfinder towards a large-scale realization of an experiment which has sensitivity to DFSZ axions in the 0. In this paper, we propose
a new experimental geometry based on a narrow-bore, segmented solenoid that optimizes the collection of the axion-induced signal using LC resonators outside the high-field region of the magnet
bore. This alternative optimization significantly reduces the required stored magnetic energy while
preserving sensitivity, thus enabling a near-term experiment in the 30–200 MHz (120–830 neV/c2
)
range, with a cost-effective, staged scaling to a GUT-scale experiment in the 100 kHz–30 MHz
(0.Where the paper says it
Where the paper says it
Where the paper says it
Acknowledgments
Stanford University and UC Berkeley gratefully acknowledge support from the Gordon and Betty Moore Foundation, grant numbers 7941 and 61988545-138055 respectively. This work was supported by the U. p. 7 ACKNOWLEDGMENTS
Stanford University and UC Berkeley gratefully acknowledge support from the Gordon and Betty Moore
Foundation, grant numbers 7941 and 61988545-138055
respectively. This work was supported by the U. This work was supported by the U. Department of Energy, Office of High Energy Physics
program under the QuantISED program, FWP 101260,
and by the National Science Foundation under award
2411650.Where the paper says it
Where the paper says it
Appendix A: Lumped element–cavity formalism correspondence in DMRadio-Core
In the magneto-quasistatic limit relevant for DMRadio, the system is more naturally described in circuit language. The pickup can reside in a region of near-zero static magnetic field, and still be sensitive. p. 8 Thus, the backreaction mode-overlap formulation and
the circuit description are equivalent statements of the
same underlying Maxwell dynamics. The pickup can reside in a region of near-zero static magnetic field, and
still be sensitive.Where the paper says it
Figure 7. The sensitivity of the DMRadio-GUT experiment alongside DMRadio-Core. p. 8 From PoyntFIG. The sensitivity of the DMRadio-GUT experiment
alongside DMRadio-Core.Where the paper says it
DMRadio-GUT is sensitive to axions at DFSZ sensitivity in the 100 kHz–30 MHz range using a 90 cm bore, 8. p. 8 The sensitivity of the DMRadio-GUT experiment
alongside DMRadio-Core. DMRadio-GUT is sensitive to axions at DFSZ sensitivity in the 100 kHz–30 MHz range using
a 90 cm bore, 8. DMRadio-GUT is sensitive to axions at DFSZ sensitivity in the 100 kHz–30 MHz range using
a 90 cm bore, 8. 5 m long, 18 T magnet to achieve a scan time
of 2. 5 m long, 18 T magnet to achieve a scan time
of 2. DMRadio-Core relies on a 37 cm bore, 3 m long,
5 T magnet.Where the paper says it
Where the paper says it
Where the paper says it
Figure 6. Motivated by the magnet design presented in the previous section. p. 8 9
2 0 2
r [m]
−4
−2
0
2
4
z [m]
0
6
12
18
Magnetic Field [T]
FIG. Motivated by the magnet design presented in the previous section, an experimental setup that achieves the goals
of DMRadio-GUT employs an 18 T peak field segmented
solenoidal magnet with a 90 cm bore.Where the paper says it
The pickups (red) traverse the high-field region and each have a radius of 2. p. 8 Motivated by the magnet design presented in the previous section, an experimental setup that achieves the goals
of DMRadio-GUT employs an 18 T peak field segmented
solenoidal magnet with a 90 cm bore. The pickups (red) traverse the high-field region and each have a radius of 2. The pickups (red) traverse the high-field region and each have a radius of 2. Soft iron is also included in this design to reduce the dc magnetic field within the pickup region.Where the paper says it
Where the paper says it
Figure 5. (a) DMRadio-Core is sensitive to DFSZ axions over the 70–140 MHz frequency range (290–580 neV/c2) for Ops 1. p. 7 (a) DMRadio-Core is sensitive to DFSZ axions over the 70–140 MHz frequency range (290–580 neV/c
2
) for Ops 1. The
sensitivity assumes SNR=3 with a local dark matter density of ρDM = 0.Where the paper says it
The scan rate of DMRadio-Core over the Ops 1 range of 70–140 MHz. p. 7 7 × 10−11 GeV−1
[46] with the
ADMX-SLIC experiment having constrained dark matter axions around 42 MHz at gaγγ ≲ 4 × 10−13 GeV−1
[31]. (b) The scan
rate of DMRadio-Core over the Ops 1 range of 70–140 MHz. The calculated scan rates for DMRadio-Core are shown
in Figure 5b over the range of 70–140 MHz. Since some
pickup sizes can result in prohibitively slow scan rates
or require unphysical tuning parameters, they are omitted from the axion search within those frequency ranges. Specifically the 80–86 MHz range only utilizes the
do = 120 cm pickup and the 94–100 MHz only utilizes the do = 146 cm pickup. The data for the scan rates is available at [47].Where the paper says it
Where the paper says it
Where the paper says it
**Table I. Live scan times for different magnet configurations. p. 6 24 yr
TABLE I. Live scan times for different magnet configurations
to achieve SNR= 3.Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improved AI system can perform parameter space exploration for QCD axions by using DMRadio-Core's geometry to significantly reduce required stored magnetic energy while preserving sensitivity in the 30–200 MHz range.
-
Improved AI system can optimize experimental design by using finite element modeling simulations with COMSOL to determine an optimal pickup geometry that involves a fully enclosed structure penetrating the core of a solenoid.
-
Improved AI system can accurately calculate the scan rate for resonant axion experiments by using Equation 3, which incorporates parameters like SNR2, Q, Leff, and Gaagga to extract dν/dt.
-
Improved AI system can design future GUT-scale experiments by using preliminary designs from DMRadio-GUT as a pathfinder to estimate the required magnet specifications and live scan times for reaching sensitivity in the 100 kHz–30 MHz range.
Abstract
Searches for QCD axions with masses in the neV/ c squared mass range are strongly motivated by new physics at the GUT scale and by well-motivated pre-inflationary axion symmetry breaking scales. This parameter space is challenging to probe due to the small axion-photon couplings, which typically require large, high-field magnets with substantial stored energy. In this paper, we propose a new experimental geometry based on a narrow-bore, segmented solenoid that optimizes the collection of the axion-induced signal using LC resonators outside the high-field region of the magnet bore. This alternative optimization significantly reduces the required stored magnetic energy while preserving sensitivity, enabling a near-term experiment in the 30-200 MHz (120-830 neV/ c squared) range, with a cost-effective, staged scaling to a GUT-scale experiment in the 100 kHz-30 MHz (0.4-120 neV/ c squared) range.
Sources
- Axions In String Theory
- The QCD Axion and Moduli Stabilisation
- An M Theory Solution to the Strong CP Problem and Constraints on the Axiverse
- The type IIB string axiverse and its low-energy phenomenology
- On String Theory Expectations for Photon Couplings to Axion-Like Particles
- String Theory and Grand Unification Suggest a Sub-Microelectronvolt QCD Axion
- Standard Model-Axion-Seesaw-Higgs Portal Inflation. Five problems of particle physics and cosmology solved in one stroke
- Axion Predictions in $SO(10)\times U(1)_{\rm PQ}$ Models
- Axion mass prediction from minimal grand unification
- Axion properties in GUTs
- The QCD Axion and Unification
- Axion Dark Matter, Proton Decay and Unification
- Axion Couplings in Grand Unified Theories
- Axion Dark Matter Detection using an LC Circuit
- Broadband and Resonant Approaches to Axion Dark Matter Detection
- The search for low-mass axion dark matter with ABRACADABRA-10cm
- First Results from ABRACADABRA-10 cm: A Search for Sub-$\mu$eV Axion Dark Matter
- ADMX SLIC: Results from a Superconducting LC Circuit Investigating Cold Axions
- Search for axion-like dark matter with ferromagnets
- The Cryogenic System of DMRadio-50L
Related papers
- DrSai: an agentic AI system toward autonomous physics analysis workflows at BESIII
- CANTON- mu Proposal: A Next-Generation Muon g-2 Measurement at Sub-0.1 ppm Precision
- Search for dark photon dark matter near 1,meV using a superconducting mixer
- ANUBIS detector and its sensitivity to neutral long-lived particles
- How to scale your HEP ML models: A recipe for robust architecture comparisons at scale
- Search for the 16 O(ppp) to 13 C pi+ pi+ e+ Decay Mode in Super-Kamiokande Using Machine Learning Techniques