---
title: 'PhysDrive: Physics-Driven Control & Dataset'
url: https://www.emergentmind.com/topics/physdrive
type: topic
---

# PhysDrive: Physics-Driven Control & Dataset

PhysDrive refers to a family of physics-driven strategies in which a deliberately chosen drive variable is used to regulate transport, relaxation, memory, or sensing, and it also names a recent multimodal dataset for in-vehicle physiological monitoring. In the cited literature, the driving variable may be a bulk stochastic kick in a granular fluid, a body force on oppositely driven disks, an oscillating linear potential in a disordered quantum chain, strain in a spin-mechanical device, a global angular-velocity cycle for bistable material bits, or an energy-routed latent transition in a world model. Taken together, these works suggest that PhysDrive is best understood not as a single formalism but as a recurring design principle: embed the drive directly into the governing dynamics, power balance, or control objective so that nonequilibrium structure becomes a controllable resource rather than a disturbance [1002.2150] [1707.09438] [1702.06208] [1503.06793] [2605.18303] [2507.19172].

## 1. Conceptual scope

A unifying feature of PhysDrive is that the drive appears explicitly in the evolution law. In the oppositely driven disk system, the overdamped dynamics are
$$
\eta \frac{d\mathbf{R}_i}{dt}=\sum_{j\neq i}\mathbf{F}_{ij}^{\rm int}+\mathbf{F}_i^{\rm drive},
$$
with species-dependent body forces along $\pm x$; in PH-Dreamer, the canonical Port-Hamiltonian template is
$$
\dot{x}=[J(x)-R(x)]\,\nabla_x H(x)+G(x)\,u,
$$
so that energy routing, dissipation, and port injection are built into the transition model itself [1707.09438] [2605.18303].

The same logic appears in classical nonequilibrium matter. In driven granular fluids, bulk kicks compensate collisional dissipation and maintain a homogeneous steady state with a well-defined granular temperature; in localized-drive diffusion, a single biased bond generates a Poisson problem with a source-sink structure; in dynamic material memory, the drive is encoded as an orbit in the $(\Omega,\dot{\Omega})$ plane, and switching depends on how that orbit crosses threshold curves [1002.2150] [1106.1838] [2508.16257].

This breadth has an immediate consequence. PhysDrive does not imply that all driven systems share a universal reduced description. The granular-fluid analysis explicitly states that driven granular dynamics is not dynamically universal in the mode-coupling sense, because long-time behavior depends explicitly on the restitution coefficient through the memory prefactor $A_k(e)$, and PH-Dreamer states that its PH constraints act as auxiliary biases rather than formal closed-loop guarantees [1002.2150] [2605.18303].

## 2. Driven nonequilibrium media

In "Glass Transition for Driven Granular Fluids" [1002.2150], PhysDrive is realized as homogeneous bulk driving of dissipative hard spheres. The system consists of $N$ identical hard spheres with binary collisions characterized by a normal restitution coefficient $e$, while neighboring pairs receive opposite random kicks to conserve momentum locally. Within a mode-coupling-theory construction, the normalized density correlator $\phi_k(t)$ obeys a Mori-Zwanzig-type equation with a positive memory kernel, and the long-time limit $f_k=\lim_{t\to\infty}\phi_k(t)$ satisfies the standard nonergodicity self-consistency equation. The central prediction is an ideal glass transition at a finite packing fraction for all $0\le e\le 1$, with the critical packing fraction $\phi_c(e)$ increasing as $e$ decreases. The approach also recovers the canonical two-step relaxation scenario, the critical law $\phi_k(t)-f_k^c\propto t^{-a(e)}$, the von Schweidler law in the early $\alpha$ regime, and the divergence
$$
\tau_\alpha\sim [\phi_c(e)-\phi]^{-\gamma(e)}.
$$
At the same time, the paper emphasizes non-equilibrium specificity: detailed balance is broken, fluctuation-dissipation relations are violated, and equilibrium Newtonian or Brownian rescalings do not collapse the driven granular data [1002.2150].

A distinct but closely related PhysDrive mechanism appears in "Long-range steady state density profiles induced by localized drive" [1106.1838]. There, a single biased bond in an otherwise diffusive lattice acts as a dipole source for the steady-state density field. Away from the driven bond, the density satisfies a Laplace equation; at the bond endpoints, the continuity equation produces localized source and sink terms. The density deviation therefore maps onto the electrostatic potential of a dipole, giving algebraic rather than exponential spatial decay in two or higher dimensions. In $d=2$,
$$
\delta \rho(r)\simeq \frac{P\cdot \hat r}{2\pi r},
$$
while in $d>2$ the decay is $r^{-(d-1)}$. The associated stationary current has the structure of the dipole electric field. The same mapping survives exclusion interactions, with amplitudes renormalized by local correlations [1106.1838].

These two lines of work share a common interpretation. In the granular case, bulk drive stabilizes a nonequilibrium steady state in which glassy arrest remains sharply defined. In the localized-diffusion case, a microscopically local drive produces long-range steady-state organization. This suggests that PhysDrive can be either extensive, as in bulk temperature maintenance, or highly localized, as in a single driving bond, without losing predictive control over far-from-equilibrium structure.

## 3. Collective transport and dynamical phases

In "Velocity Force Curves, Laning, and Jamming for Oppositely Driven Disk Systems" [1707.09438], PhysDrive appears as a species-dependent body force in an athermal overdamped disk mixture. For equal populations driven oppositely and for $\phi\ge 0.55$, four nonequilibrium phases emerge as the drive $F_D$ increases: a jammed crystalline cluster, a completely phase-separated collisionless state, a continuously mixing fluctuating phase, and a laning smectic state. At $\phi=0.848$, the reported boundaries are $0<F_D<0.3$ for phase I, $0.3<F_D<1.15$ for phase II, $1.15<F_D<5.45$ for phase III, and $F_D>5.45$ for phase IV. The transitions are encoded directly in the velocity-force curves. The I$\to$II transition produces a sharp jump in $\langle V_1\rangle$, the II$\to$III transition exhibits negative differential mobility, and the III$\to$IV transition restores nearly free flow. Structure-factor signatures evolve correspondingly from sixfold Bragg peaks to a liquid-like ring and then to smectic peaks. The collisionless phase-separated and laned states are absorbing states in the sense that interspecies contacts vanish and velocity fluctuations disappear; the jammed phase is also absorbing, although it is contact-rich [1707.09438].

In "Dynamics and Nonmonotonic Drag for Individually Driven Skyrmions" [2106.06093], the drive acts on a single skyrmion moving through a bath of other skyrmions. The modified Thiele equation,
$$
\alpha_d\,\mathbf v_i+\alpha_m(\hat z\times \mathbf v_i)=\mathbf F_i^{ss}+\mathbf F_i^D,
$$
contains both dissipative and Magnus terms. This immediately changes the rheology. In the damping-dominated regime, the effective viscosity increases monotonically with density, and the driven skyrmion always slows relative to the single-particle limit. In the Magnus-dominated regime, the velocity becomes nonmonotonic in density and can exceed the single-particle value. For $\alpha_m/\alpha_d=9.95$ at fixed $F_D=0.5$, $\langle V_\parallel\rangle$ peaks near $\rho\approx 1.25$ at approximately $0.78$, while the total speed reaches approximately $0.9$ and remains above the single-particle value across a broad density window. The measured skyrmion Hall angle starts near zero at threshold, increases with drive, and saturates toward the intrinsic Hall angle; at fixed drive, increasing density can simultaneously reduce the Hall angle and boost the speed. The proposed mechanism is a density imbalance $\Delta \rho$ created transverse to the drive, which the Magnus term converts into an additional longitudinal component, summarized by
$$
V_{\rm boost}\propto (\alpha_m/\alpha_d)\,\Delta \rho.
$$
Negative differential conductivity appears in $\langle V_\parallel\rangle(F_D)$ for sufficiently large $\alpha_m/\alpha_d$ [2106.06093].

A common misconception is that stronger crowding must always increase drag. The skyrmion study shows that this is correct in overdamped or damping-dominated settings but fails once gyroscopic transport channels become relevant. The disk study shows an analogous point from a different angle: increasing drive can first destabilize ordered low-collision motion into a collisional phase before larger drive restores collisionless laning.

## 4. Derivative-space control and programmable material memory

"Dynamic driving enables independent control of material bits for targeted memory" [2508.16257] formulates PhysDrive as control in derivative space. The material bit is a pre-buckled, double-clamped elastic beam with binary states defined by buckling direction. The global control input is the disk angular velocity profile $\Omega(t)$, whose derivatives generate inertial forces with different dependencies: the centrifugal contribution scales as $f_\Omega\sim \Omega^2$, while the Euler contribution scales as $f_E\sim \dot{\Omega}$. Because different beams have different geometric sensitivities, they can be selectively switched by designing drive cycles whose orbits in the $(\Omega,\dot{\Omega})$ plane cross the appropriate switching curves $u_i(\Omega,\dot{\Omega})$ and $d_i(\Omega,\dot{\Omega})$ in the intended order and orientation [2508.16257].

The model used to rationalize these thresholds is a reduced-order von Mises-truss analogue with a central mass connected to two linear springs. In the rotating frame, the equilibrium conditions include the elastic spring forces and the inertial forces
$$
\mathbf F_{\Omega}=-m\,\boldsymbol{\Omega}\times\left(\boldsymbol{\Omega}\times(\mathbf r+R\,\mathbf e_y)\right),\qquad
\mathbf F_E=-m\,\dot{\boldsymbol{\Omega}}\times(\mathbf r+R\,\mathbf e_y).
$$
A realizable target orbit must satisfy the directionality constraint
$$
\dot{\Omega}(\tau)=\frac{d\Omega(\tau)}{d\tau}\frac{d\tau}{dt},
$$
so only crossings oriented away from the origin can produce switching. The experiments show that two harmonic protocols with identical amplitude $a_1=a_2=36.5\ {\rm rad/s}$ but different frequency can drive the same initial state $(00)$ to $(10)\to(11)$ or to $(01)\to(11)$, depending on which side of the $u_1$-$u_2$ intersection the orbit traverses [2508.16257].

The most elaborate demonstration is a five-bit architecture designed so that a single clockwise orbit beginning and ending at $(\Omega,\dot{\Omega})=(0,0)$ executes an erase-write sequence capable of producing an arbitrary target $(t_1 t_2 t_3 t_4 t_5)$. The paper reports a library of orbits that writes all 26 five-bit strings used for uppercase-letter encoding and experimentally demonstrates the sequential writing of “ORBIT.” Each selected orbit was repeated five times with consistent switching behavior. A notable implication is that local one-by-one actuation is not required for independent control: global drive, if shaped in derivative space, can realize any-to-any transitions within a single cycle [2508.16257].

## 5. Quantum and strongly driven coherent systems

In "Driving induced many-body localization" [1702.06208], PhysDrive is implemented as an oscillating linear potential applied to a weakly disordered one-dimensional interacting hardcore-boson chain. The static undriven model with $U=1.5J$ and $W=2J$ is ergodic at half filling. After a time-dependent gauge transformation, the drive dresses the hopping by a Peierls phase, and high-frequency averaging yields the effective hopping
$$
J_{\rm eff}=J\,\mathcal J_0\!\left(\frac{A}{\omega}\right).
$$
Tuning $A/\omega$ near a zero of $\mathcal J_0$ suppresses tunneling and enhances the relative strength of disorder and interactions. Numerically, the system enters a Floquet-MBL phase above a critical frequency and within a finite amplitude window around the first Bessel zero. The quasienergy level-spacing ratio crosses from the Wigner-Dyson value $r_{\rm WD}\approx 0.53$ toward the Poisson value $r_{\rm POI}\approx 0.39$, and the long-time imbalance saturates near $0.6$ at $\omega=5J$ tuned to $J_{\rm eff}=0$. The reported critical frequency is $\omega_c\approx 4J$ at the first root and $\omega_c\approx 3.25J$ near the second root. The result is important because periodic drive more often destabilizes localization, whereas here a specific coherent suppression of tunneling induces it [1702.06208].

In "Strong mechanical driving of a single electron spin" [1503.06793], PhysDrive is realized without electromagnetic control fields. The system is a single NV center in a diamond cantilever, with electronic spin $S=1$, zero-field splitting $D_0=2.87$ GHz, and hyperfine splitting $A_{\rm HF}=2.18$ MHz. The relevant transverse strain coupling is
$$
H_{\rm str,\perp}=-h g_0^\perp(\hat a+\hat a^\dag)(S_+^2+S_-^2),
$$
which drives the otherwise forbidden $\Delta m_s=\pm 2$ transition between $\lvert -1,m_I\rangle$ and $\lvert +1,m_I\rangle$. For a classical coherent phonon field,
$$
H_{\rm drive}(t)=h(\Omega_m/2\pi)\cos(\omega_m t)(S_+^2+S_-^2).
$$
The experiment observes strain-driven Rabi oscillations with $\Omega_m/2\pi=1.14\pm 0.01$ MHz at a tip displacement of approximately $100$ nm, mechanically induced Autler-Townes splittings, and beyond-RWA crossings and anticrossings as $\Omega_m$ approaches and exceeds $\omega_m$. The maximum reported drive reaches $\Omega_m^{\max}/2\pi\approx 10.75$ MHz. Continuous dynamical decoupling in the dressed basis extends the Ramsey coherence time from $T_2^*=3.6\pm 0.1\ \mu{\rm s}$ to $14.0\pm 0.6\ \mu{\rm s}$. Residual dephasing is attributed to electric-field or strain noise, temperature drift of $D_0$, and second-order magnetic coupling [1503.06793].

These two studies delimit complementary quantum meanings of PhysDrive. One uses periodic forcing to renormalize an effective static Hamiltonian and access a Floquet-MBL phase; the other uses intrinsic strain to generate coherent strong driving and dressed-state protection. In both cases, the drive is not merely a perturbation but the central engineering handle.

## 6. Physics-structured learning and engineering control

"PH-Dreamer: A Physics-Driven World Model via Port-Hamiltonian Generative Dynamics" [2605.18303] brings PhysDrive into model-based reinforcement learning. The framework combines an RSSM backbone with a PH shadow transition on projected physical latents, an explicit kinematics-aware energy world model, and an energy-guided Actor-Critic regularized by Lagrangian multipliers. The explicit branch parameterizes
$$
\mathcal H(q_t,p_t)=V_{\theta_V}(q_t)+\frac12 p_t^\top M_{\theta_L}^{-1}(q_t)p_t,
$$
computes work and dissipation through $G_{\theta_G}$ and $D_{\theta_D}$, and supplies $\nabla_a \mathcal H_{t+1}$ and $\nabla_a^2 \mathcal H$ to the policy objective. Across DeepMind Control Suite tasks, the reported asymptotic average return is $789.2$ for PH-Dreamer versus $762.5$ for R2Dreamer; imagined reward averages are $738.9$ versus $702.5$; latent log phase volume is reduced by $4.18\%-8.41\%$; energy consumption is reduced by up to $7.80\%$; and mean squared jerk is reduced by up to $9.38\%$. The paper explicitly notes, however, that the PH constraints are auxiliary and do not provide closed-loop guarantees, and that the explicit energy branch relies on proprioceptive inputs rather than pure pixel-based discovery [2605.18303].

A more classical control-theoretic PhysDrive appears in "Data Set Description: Identifying the Physics Behind an Electric Motor -- Data-Driven Learning of the Electrical Behavior (Part II)" [2003.06268]. The dataset contains approximately $40$ million samples from an IPMSM fed by a two-level IGBT inverter. Each row stores measured $(i_d,i_q,\epsilon,n_k,n_{k-1})$ at controller-cycle instant $k$ together with $(i_d,i_q)$ at $k+1$, but successive rows do not form a continuous time series. The controller cycle is $T_s=50\ \mu{\rm s}$, the prediction horizon is $n_p=1$, and the dataset spans $478$ current setpoints under $U_{DC}=300$ V and $n_{\rm me}=1000\ {\rm min}^{-1}$. The paper advocates per-vector linear or nonlinear discrete-time models, such as
$$
\hat x_{k+1}=K_n x_k \quad\text{or}\quad \hat x_{k+1}=f_n(x_k),
$$
and recommends class balancing over a grid in $(i_d,i_q,\epsilon)$ that yields $16{,}956$ valid classes per subset and approximately $99\%$ coverage at $48$ samples per class. The stated rationale is that learned models capture saturation, flux harmonics, inverter nonlinearity, dead-time, interlocking, and measurement offsets that nominal white-box models omit [2003.06268].

"Driveability Constrained Models for Optimal Control of Hybrid Electric Vehicles" [2303.12603] makes PhysDrive explicitly vehicular. For a P2 parallel hybrid, the stage cost augments fuel consumption with penalties on gear changes, engine starts, and used torque fraction:
$$
L=\dot m_{\rm fuel}\Delta t+L_\gamma+L_\epsilon+L_{T_{\rm res}}.
$$
Dynamic programming on Artemis cycles shows clear trade-offs. The baseline fuel-optimal strategy achieves $4.58$ L/100 km, $18$ shifts per minute, $3.1$ engine starts per minute, and average torque reserve $58.3\%$. Penalizing gear shifts reduces shifts to $0.93$ per minute at a fuel penalty of $+1.6\%$; penalizing engine starts reduces starts to $0.67$ per minute at $+3.2\%$ fuel; penalizing torque usage increases average reserve to $65.8\%$ at $+2.5\%$ fuel. The paper’s practical conclusion is that driveability objectives can be encoded as explicit physics-grounded penalties and then distilled into real-time rules such as dwell-time constraints, gear hysteresis, and reserve targets [2303.12603].

## 7. PhysDrive as a multimodal in-vehicle sensing dataset

In its most specific current usage, "PhysDrive: A Multimodal Remote Physiological Measurement Dataset for In-vehicle Driver Monitoring" [2507.19172] names a dataset for contactless physiological sensing in real driving. It includes synchronized RGB video, near-infrared video, and raw mmWave radar together with six physiological ground truths: ECG, BVP, respiration, HR, RR, and SpO$_2$. The dataset comprises $48$ licensed drivers, evenly split by sex, aged $18$-$41$ years with mean $24.9$ and standard deviation $4.1$. Each driver contributes six segments of about five minutes, for a total of about $24$ hours and about $16.5$ km per driver. The design spans three vehicle classes, four illumination conditions, two driver-action conditions, and three road conditions. RGB is recorded at $30$ fps and $1024\times 576$ resolution, NIR at $30$ fps and $1920\times 1080$, and radar at $20$ fps with effective bandwidth $2.6$ GHz, a $12$-antenna virtual array, range resolution of about $6$ cm, and angular resolution $14^\circ$ [2507.19172].

The benchmark protocol uses an $80/10/10$ cross-subject train-validation-test split over the $48$ drivers, with five independent runs. Evaluation employs MAE, RMSE, Pearson correlation, and SNR. The reported cross-subject HR results show a clear modality hierarchy in this dataset: for RGB, PhysNet reaches HR MAE $6.29$, RMSE $8.93$, and $P=0.61$; for NIR, PhysNet reaches HR MAE $10.69$, RMSE $13.21$, and $P=0.12$; for mmWave, mmFormer reaches HR MAE $3.65$, RMSE $5.09$, and $P=0.97$, with RR MAE $1.49$ and $P=0.83$. The paper also notes that radar waveform recovery lags behind direct HR and RR regression, plausibly because waveform recovery is more sensitive to residual temporal misalignment. SpO$_2$ is included but not benchmarked because the distribution is narrow and no hypoxia induction was performed for safety [2507.19172].

The dataset is also explicit about its limitations. The participant pool is predominantly East Asian; passenger monitoring is outside scope; software-level synchronization leaves up to $1$ s drift; and the SpO$_2$ channel lacks the variability needed for strong estimation benchmarks. Even so, the dataset makes PhysDrive a concrete shared benchmark as well as an abstract methodological theme: it operationalizes physics-aware multimodal sensing in a domain where illumination changes, motion, vibration, and vehicle-specific multipath are intrinsic rather than artificially suppressed [2507.19172].

Source: https://www.emergentmind.com/topics/physdrive