TiSR in Multi-Domain Research

• TiSR is a multifaceted acronym representing distinct research objects across biomedical vision, equations discovery, image reconstruction, and wireless methods.
• It encompasses varied methodologies such as textual inversion with self-supervised refinement, genetic-programming-driven symbolic regression, and convex optimization in super-resolution tasks.
• Experimental results across applications show improved metrics—ranging from better image quality and reduced CPU time to enhanced communication error rates—highlighting the need for field-specific interpretation.

Searching arXiv for the cited TiSR-related papers to ground the article in current records. TiSR, also written as TISR in some papers, is not a single canonical concept in the arXiv literature. The acronym is used for at least five distinct research objects across biomedical vision, symbolic regression, super-resolution, and wireless communications: Textual Inversion and Self-supervised Refinement for radiology report generation (Luo et al., 2024); thermodynamics-informed symbolic regression as a Julia-based framework for equations of state (Martinek et al., 2023); the TiSR symbolic-regression package as benchmarked under FastSRB (Martinek, 20 Aug 2025); Thermal Image Super-Resolution as a reconstruction task (Yan et al., 2022); twin-image super-resolution in multi-image satellite super-resolution (Lin et al., 25 Feb 2026); and Time Index Modulation-Driven Standalone RIS for symbiotic radio (Pihtili et al., 2024). This suggests that domain-level disambiguation is necessary whenever the acronym appears without expansion.

1. Scope and nomenclature

The capitalization pattern correlates with field usage. In radiology report generation, the acronym appears as TISR, expanding to Textual Inversion and Self-supervised Refinement (Luo et al., 2024). In symbolic regression, it appears as TiSR, expanding to thermodynamics-informed symbolic regression and denoting both a framework and a Julia package (Martinek et al., 2023). In computer vision, TISR may denote either Thermal Image Super-Resolution as a problem setting (Yan et al., 2022) or twin-image super-resolution as a specific multi-image super-resolution scenario (Lin et al., 25 Feb 2026). In wireless communications, TiSR denotes Time Index Modulation-Driven Standalone RIS mechanism for symbiotic radio (Pihtili et al., 2024).

These usages are methodologically unrelated. Some refer to a trainable model wrapped around an encoder–decoder pipeline, some to a genetic-programming engine, some to an inverse problem, and some to a communications mechanism. A plausible implication is that bibliographic retrieval based only on the acronym is inherently noisy unless constrained by field, title expansion, or arXiv identifier.

2. Textual Inversion and Self-supervised Refinement in radiology report generation

In radiology report generation, TISR is built on top of the standard encoder–decoder paradigm and is intended to address two shortcomings attributed to mainstream approaches: overlooking the modality gap and ignoring report content constraints (Luo et al., 2024). Let $\mathcal{I}\in\mathbb{R}^{B\times M\times D}$ denote visual features extracted by an image encoder and $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$ text embeddings produced by the decoder’s autoregressive loop, with $B$ the batch size, $M$ image patches, $N$ maximum report length, and $D$ embedding dimension. The decoder predicts each next word via

$$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$

The first component, textual inversion, learns a lightweight MLP that maps image features into pseudo-words in the same embedding space as the text decoder:

$$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$

$$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$

These pseudo-words are then processed by attention in the same space as ordinary word tokens. The stated purpose is to eliminate the cross-modal feature gap by representing images as pseudo words.

The second component, self-supervised refinement, applies cross-modal interaction and contrastive alignment to refine the pseudo-words without extra annotations. The interaction stage is

$$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$

followed by fusion

$\mathcal{T}\in\mathbb{R}^{B\times N\times D}$0

A score matrix $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$1 is formed over the batch, and the refinement objective is a symmetric cross-entropy loss $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$2. The full training objective is

$\mathcal{T}\in\mathbb{R}^{B\times N\times D}$3

Architecturally, TISR is described as orthogonal to most existing methods and plug-and-play. No change is required to the base encoder or decoder; the framework adds the inversion MLP and the contrastive head. The authors apply it to R2Gen, R2GenCMN, and VLCI, with preprocessing following each baseline’s existing pipeline. Experiments are conducted on IU X-ray with 7 470 images, 3 955 reports, split 7:1:2 [official], and MIMIC-CXR with 377 110 images, 227 835 reports, using official train/val/test splits. Metrics include BLEU-1…4, METEOR, ROUGE-L, and Clinical Efficacy: precision, recall, F1 via CheXbert label extraction.

The reported gains are consistent across baselines and datasets. On IU-Xray with R2Gen, BLEU-1 improves from 0.443 to 0.470, BLEU-4 from 0.168 to 0.187, METEOR from 0.175 to 0.194, and ROUGE-L from 0.355 to 0.369. On MIMIC-CXR with R2Gen, BLEU-1 improves from 0.350 to 0.358, BLEU-4 from 0.103 to 0.106, CE-Precision from 0.424 to 0.467, CE-Recall from 0.254 to 0.302, and CE-F1 from 0.317 to 0.367. The paper reports no formal $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$4-values. Qualitative attention maps indicate improved focus on clinically relevant lung regions and fewer hallucinated findings, including an example where the baseline misses atelectasis and hallucinates consolidation, while the TISR-enhanced model highlights and mentions the true finding.

3. Thermodynamics-informed symbolic regression for equations of state

In symbolic regression, TiSR denotes a Julia-based, genetic-programming-driven framework specifically engineered to accelerate and partly automate the discovery of analytical equations of state from often noisy, scattered thermodynamic data (Martinek et al., 2023). Its core search loop combines a modified NSGA-II multi-objective selector, an island-model population structure on a static ring topology, and a grammar-checking layer to forbid physically meaningless subexpressions such as $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$5 or parameters in exponents. The per-generation workflow includes random initialization or seed injection, mutation and subtree crossover, individual instantiation, Levenberg–Marquardt parameter identification, singularity checks, complexity and residual evaluation, periodic migration, and hybrid selection by Pareto sorting followed by tournament filling.

The objective layer is explicitly multi-objective. For data $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$6 and model $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$7, TiSR defines raw residuals $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$8 and optionally processed residuals $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$9. The default parameter-estimation objective is the weighted mean square error

$B$0

while selectable optimization targets include $B$1, $B$2, $B$3, relative-error metrics, total node count $B$4, $B$5, and $B$6. Parameter fitting uses LM with early stopping by splitting data into fit and validation sets and aborting when validation residual increases for $B$7 consecutive LM iterations.

TiSR incorporates pre- and post-processing machinery intended for EOS development. Before regression, inputs may be normalized or Z-scaled, outputs transformed, and a pre_residual_processing! hook supplied to enforce known functional structure. The paper gives the example of fitting $B$8 when one knows $B$9. Post-processing includes a drastic_simplify! mutation, symbolic simplification with SymbolicUtils.jl, and a planned $M$0-regularized fitting step. Built-in thermodynamic supports include protected evaluation of $M$1, $M$2, and division, factor variables, and support for Helmholtz energy formulation; full thermodynamic consistency enforcement is explicitly described as not yet built in.

The default function set is

$M$3

and arbitrary Julia-level functions may be registered with protected-domain guards. In the case-study discussion, TiSR is reported to have been used in Frotscher et al. (2023) to rediscover simplified virial expansions and polynomial fits to experimental PVT data with population size 200–300 per island, 3–5 islands, 500–2000 generations, and run time on 8 cores of approximately 1–3 h for moderate data sets $M$4. One reported model is

$M$5

with RMSE $M$6 and max relative error $M$7, using 5–7 parameters and recursive complexity $M$8. The framework is described as not ready for end users yet, and the roadmap includes constrained fitting, singularity prevention in two-phase regions, direct Helmholtz EOS discovery, regularization, tighter coupling to REFPROP/CoolProp, and a GUI front-end.

4. TiSR as benchmarked under FastSRB

The paper “Fast Symbolic Regression Benchmarking” does not introduce TiSR as a new algorithm; instead, it uses the existing TiSR Julia package (“\tisr,” Martinek et al. 2023) as one of two off-the-shelf symbolic-regression toolkits in the FastSRB framework (Martinek, 20 Aug 2025). For this benchmark, TiSR is configured as a classic tree-based genetic-programming engine with the common function set

$M$9

Tree-structure restrictions prohibit nesting of $N$0 with each other or themselves and likewise for $N$1; in TiSR, the pow operator may only raise to a constant exponent. Other GP parameters remain at package defaults.

FastSRB’s central intervention is an early-termination callback. TiSR natively supports a user callback, which the authors attach by editing commit 5b541b3. Internally, approximately every 15 s the search loop invokes the callback on all members of the current hall of fame. A run is eligible for early termination only if the relative-error measure

$N$2

satisfies $N$3 and the candidate’s tree size $N$4 does not exceed $N$5. If these filters pass, the expression is simplified with Sympy, constants are rounded to five significant digits, and the result is compared against a curated set of functionally equivalent ground-truth forms. A match signals immediate termination; near-misses are logged for future inclusion.

The benchmark comprises 120 Feynman-derived SRSD tasks, each run 5× for 30 min max, with 200 train / 200 test points, no multiprocessing, on an 8-core AMD Ryzen 7 PRO machine. TiSR’s rediscovery rates under FastSRB are reported as 100.0 % on the easy category, 78.0 % on medium, 42.6 % on hard, and 69.4 % overall. Time savings from early termination are reported as 63 % less CPU time (approximately 184 h saved). For comparison, SymbolicRegression.jl rises from 26.7 % to 44.7 % rediscovery and saves 41.2 % CPU time. The paper attributes TiSR’s stronger rediscovery rate primarily to the new benchmarking methodology—acceptance of equivalent forms plus early stopping—rather than to any intrinsic superiority of TiSR’s core GP operators. Caveats include three dropped problems (B4, B11, III.9.52) due to recurrent Sympy failures, a fairness assumption that discovered expressions would survive to normal termination, variable maximum expression complexity, and the evolving nature of the curated list of acceptable forms.

5. Thermal Image Super-Resolution

In computer vision, TISR also denotes Thermal Image Super-Resolution, a task motivated by the lower native resolution, higher noise, and wider temperature-to-radiance nonlinearity of thermal sensors (Yan et al., 2022). In the PBVS-2022 Challenge, the goal is to reconstruct either a $N$6 or $N$7 higher-resolution thermal image from a given low-resolution input while maintaining both PSNR/SSIM and perceptual or clinical utility. Yan et al. address this problem with BN-CSNT, a Bilateral Network with Channel Splitting Network and Transformer.

BN-CSNT uses two parallel branches. The context branch applies channel splitting and $N$8 Swin-Based Blocks to capture long-range contextual cues; the spatial branch uses a shallow transformer with two Swin Basic Layers to preserve fine local detail. The context branch is coupled with an Attention Refinement Module (ARM), implemented in squeeze-and-excitation style:

$N$9

The two branches are merged by a Feature Fusion Module (FFM), followed by pixel-shuffle upsampling and a final convolution. Training uses an $D$0 loss for Track-1 ($D$1 SR on Flir data) and for the first phase of Track-2 ($D$2 Axis$D$3Axis); the second phase of Track-2 adds an LSGAN objective and an SSIM loss, yielding

$D$4

The reported training hyperparameters are Adam with $D$5, $D$6, initial learning rate $D$7 decayed by half every 200 epochs, batch size 16 with $D$8 crops, 1000 total epochs, and augmentation by random horizontal/vertical flips and rotation. Final single-model challenge results are PSNR = 33.64, SSIM = 0.9263 for Track-1 ($D$9) and PSNR = 21.08, SSIM = 0.7803 for Track-2 ($$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$0). Compared to prior TiSR methods on the same challenge benchmark, including CSN with approximately 32.5 dB / 0.918 and SwinIR-based SR with approximately 33.2 dB / 0.923 on $$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$1, BN-CSNT yields a +0.4 dB PSNR gain and +0.003 SSIM gain. The paper attributes this to the bilateral design, channel splitting, ARM-based reweighting, and lightweight fusion of long-range and local features.

6. Twin-image super-resolution in remote sensing

A different super-resolution usage appears in remote sensing, where TISR denotes twin-image super-resolution, described as the most challenging multi-image super-resolution (MISR) scenario because the high-resolution image must be recovered from a low-resolution image and its subpixel-shift counterpart (Lin et al., 25 Feb 2026). The observation model is

$$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$2

where $$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$3 is the unknown HR image, $$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$4 is a blur-plus-downsampling operator, and $$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$5 is a fixed half-pixel shift on the HR grid. The reconstruction objective is the convex problem

$$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$6

with convex self-similarity prior

$$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$7

The paper solves this formulation with COSUP, a convex self-similarity unfolding supermode super-resolution algorithm. The derivation proceeds through ADMM by introducing an auxiliary variable $$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$8 and alternating between a proximal step for $$\mathcal{O}_n = f_d(\mathcal{I},\,\mathcal{T}_{n-1}),\qquad \mathcal{R}_n = \mathrm{softmax}(f_l(\mathcal{O}_n)).$$9 and a closed-form $$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$0-update involving $$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$1. In the network realization, the shift operator is embedded as a parameter-free pixel shift layer; the proximal map is approximated by Model 1, a residual-in-residual CNN whose inner blocks are Swin-Transformer layers and which is pretrained under the self-similarity loss

$$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$2

The data-fidelity step is implemented as Model 2, which uses the Woodbury identity to reduce the inverse to a $$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$3 matrix $$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$4 realized as a tiny symmetric fully-connected layer. The unfolding uses $$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$5 stages; Model 1 has 4 Swin blocks with embedding dimension 96 and window 8; the parameter $$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$6 is learned.

Training uses HR panchromatic patches from GeoEye-1 (0.58 m), CBERS-4A (2 m), and PlanetScope (3 m), converted to panchromatic via NTSC weights and sliced into $$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$7 patches. Twin LR pairs are synthesized by Gaussian blur, $$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$8 downsampling, and half-pixel shift. The prox network is pretrained with Adam, learning rate $$P_1 = \mathrm{Dropout}(\mathrm{GELU}(f_l(\mathcal{I}))),$$9 halved on plateau, 110 epochs, batch size 6; then the entire unfolding model is trained with $$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$0 loss on the final stage output, with $$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$1 initialized to 2. Inference speed is approximately 50 ms per $$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$2 LR pair on RTX 3090. On 100 LR-pair samples from SPOT-5 THR mode, the official CNES supermode product gives average NIQE 4.85 and PI 3.49, whereas COSUP gives NIQE 3.99 and PI 2.76. The reported qualitative outcome is better preservation of sharp edges and fine textures with less over-smoothing and ringing.

7. Time Index Modulation-Driven Standalone RIS for symbiotic radio

In wireless communications, TiSR denotes a Time Index Modulation-Driven Standalone RIS mechanism for symbiotic radio (Pihtili et al., 2024). The system comprises a single-antenna primary transmitter (PTx), an $$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$3-antenna primary receiver (PRx), a standalone RIS with $$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$4 unit cells, and a passive energy harvester (EH). Group 1 unit cells assist PTx transmission by phase shifts, Group 2 are perfect absorbers for wireless energy harvesting to power the RIS, and Group 3 reflect while conveying RIS information using $$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$5 discrete phase shifts per RIS symbol. All links follow a Rician fading model with path loss.

The PTx uses time index modulation (TIM) across a block of $$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$6 slots with $$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$7 information slots. Information slots transmit symbols $$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$8 with average power $$\mathcal{P} = f_l\bigl(\mathrm{Dropout}(\mathrm{GELU}(f_l(P_1)))\bigr)\in\mathbb{R}^{B\times M\times D}.$$9, while the remaining slots transmit a deterministic power symbol $$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$0 with $$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$1:

$$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$2

The total bits per block are $$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$3, where $$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$4 are carried by the PTx symbols and $$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$5 by the slot pattern. At the RIS, the RF input power is

$$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$6

and harvested DC power follows the constant–linear–constant (CLC) rectifier model

$$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$7

Standalone operation requires

$$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$8

At the receiver, the paper derives an LLR-based detector for the TIM structure. The received signal includes the PTx direct path and the RIS-reflected contribution from Groups 1 and 3, and the slotwise LLR compares the hypotheses of an information-bearing slot versus a power slot. With equal priors, the LLR reduces to a log-sum-exp expression over the two RIS phase states and the $$\mathcal{P}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{P}\,\mathcal{T}^\top}{\sqrt D}\Bigr)\,\mathcal{T},\qquad \mathcal{T}'=\mathrm{softmax}\!\Bigl(\tfrac{\mathcal{T}\,\mathcal{P}^\top}{\sqrt D}\Bigr)\,\mathcal{P},$$9 PTx symbols, with numerical stabilization via

$\mathcal{T}\in\mathbb{R}^{B\times N\times D}$00

A mapping-aided detector then sums slotwise LLRs over each legitimate pattern $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$01 and selects the maximizing pattern. The stated complexity reduction is from

$\mathcal{T}\in\mathbb{R}^{B\times N\times D}$02

for brute-force ML to

$\mathcal{T}\in\mathbb{R}^{B\times N\times D}$03

complex multiplications.

The paper also gives BER expressions. For PTx information, conditioned on the instantaneous equivalent channel $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$04 and assuming M-PSK, the symbol-error probability is approximated by an angular integral in $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$05. For RIS information, the phase choice is repeated over $\mathcal{T}\in\mathbb{R}^{B\times N\times D}$06 slots, giving effective SNR

$\mathcal{T}\in\mathbb{R}^{B\times N\times D}$07

and for BPSK on RIS phases,

$\mathcal{T}\in\mathbb{R}^{B\times N\times D}$08

Conceptually, the mechanism is framed as mutualistic symbiosis: the PTx simultaneously delivers information and energy, the RIS self-harvests through Group 2 absorbers to sustain battery-free operation, Group 1 improves the PTx-to-PRx link, and Group 3 piggybacks RIS information with negligible extra power.