---
title: Diffusion-based Schrödinger Bridge
url: https://www.emergentmind.com/topics/diffusion-based-schrodinger-bridge-sb
type: topic
---

# Diffusion-based Schrödinger Bridge

Searching arXiv for recent and foundational papers on diffusion-based Schrödinger bridges to ground the article in the current literature.
Searching arXiv for exact papers referenced in the source material.
Diffusion-based Schrödinger Bridge (SB) denotes a class of finite-horizon stochastic transport models in which one seeks a diffusion process whose path measure is closest, in Kullback–Leibler divergence, to a reference diffusion while matching prescribed endpoint distributions. In modern machine learning, this formulation is used both as a stochastic-control problem and as a path-space version of entropy-regularized optimal transport, and it has become a central organizing framework for generative diffusion, bridge matching, and finite-time sampling between arbitrary distributions [2106.01357][2303.16852].

## 1. Variational and stochastic-control formulation

In its standard dynamic form, diffusion-based SB solves
$$
\Pi^\star = \arg\min_{\Pi} \mathrm{KL}(\Pi\|P)
\quad \text{s.t.} \quad
\Pi_0=p_0,\;\Pi_T=p_T,
$$
where \(P\) is a reference path measure on continuous trajectories, typically induced by a diffusion process, and \(p_0,p_T\) are the prescribed endpoint marginals [2106.01357]. For a Wiener prior, the reference dynamics can be written as \(dW_t^\epsilon=\sqrt{\epsilon}\,dW_t\), while in more general formulations the controlled diffusion has the form
$$
dX_t = \big(f_t(X_t)+\sigma_t u_t(X_t)\big)\,dt + \sigma_t\,dW_t,
$$
with the control chosen to minimize kinetic energy subject to the endpoint constraints [2506.22565].

This path-space problem admits several equivalent interpretations. One is the stochastic-control view, in which the objective is the expected running energy
$$
\mathbb{E}_{p^u}\!\left[\int_0^1 \frac12\,u_t(X_t)^2\,dt\right],
$$
or its Brownian special case, under the constraint that the terminal marginal equals the target [2506.22565]. Another is the static entropic optimal transport view, in which the optimal endpoint coupling solves an entropy-regularized transport problem induced by the transition kernel of the reference diffusion [2106.01357][2303.16852]. The literature repeatedly emphasizes that SB interpolates between optimal transport and stochastic transport: as the diffusion amplitude tends to zero, the Schrödinger cost tends to \(W_2^2\) [2106.01357][2303.16852].

A central structural fact is that the optimal drift can be written in terms of Schrödinger potentials. In one standard form,
$$
u_t^\star(x)=\sigma_t\nabla \log \varphi_t(x),
$$
with forward and backward potentials coupled through the endpoint conditions [2506.22565]. The optimal bridge also factorizes through the reference transition kernel:
$$
p^\star_{s,t}(y,x)=p_{t|s}(x|y)\,\psi_s(y)\,\varphi_t(x),
$$
and, in the Brownian setting, it can be represented as a mixture of Brownian bridges weighted by the optimal endpoint coupling [2402.03207]. This factorization is what makes diffusion-based SB simultaneously a control problem, a bridge problem, and an entropic transport problem.

## 2. Relation to diffusion models, score models, and flow matching

A major theme in the literature is that diffusion-based SB both generalizes and reinterprets score-based generative modeling. The foundational DSB work shows that the first DSB iteration recovers the methodology of reverse-time score-based diffusion, with the flexibility of using shorter time intervals, while subsequent iterations reduce the discrepancy between the forward and backward endpoint marginals [2106.01357]. In this sense, classical score-based generation appears as a first iterate of an SB solver rather than as a distinct construction.

This relation is most transparent through time reversal. For a forward diffusion
$$
dX_t = f(X_t)\,dt + \sqrt{2}\,dB_t,
$$
the reverse-time process has drift \(-f + 2\nabla \log p_{T-t}\), which is exactly the score-corrected reverse drift used in score-based diffusion models [2106.01357]. Several later works make this connection more explicit by showing that diffusion models can be understood as special cases of SB under additional restrictions. One paper states that diffusion is a special case of SB under a memoryless base process, where the endpoint pairing becomes independent and the bridge matching objective collapses to standard score matching [2602.15396]. Another presents a unified bridge framework in which ODE flow matching corresponds to the \(\sigma=0\) case, while SB methods correspond to the \(\sigma>0\) stochastic case [2503.21756].

This unification also clarifies a recurring misconception. Matching SB marginals is not the same as recovering the full SB path law. The unified bridge analysis explicitly notes that mini-batch Schrödinger bridge flow matching can reproduce SB marginals, but the full path measure remains an ODE bridge solution rather than the true diffusion bridge [2503.21756]. Conversely, DSBM and related iterative SB methods are designed to target the stochastic path measure itself [2303.16852].

A second misconception concerns finite-time transport. Standard denoising diffusion requires a sufficiently long horizon so that the final distribution becomes approximately Gaussian. SB does not require the reference forward process to reach the target prior exactly; instead, iterative fitting adjusts the path measure so that the terminal marginal becomes the desired prior within a prescribed finite horizon [2106.01357]. This finite-time feature is one of the principal reasons SB is repeatedly proposed for accelerated generation and sampling.

## 3. Numerical schemes and training paradigms

The canonical numerical backbone of diffusion-based SB is Iterative Proportional Fitting (IPF), which alternates KL projections enforcing the final and initial marginals [2106.01357]. DSB approximates these IPF steps by learning Gaussian conditional transitions with neural drifts, turning the continuous-state SB problem into alternating forward and backward regression problems [2106.01357]. This construction was the first to place diffusion-based SB squarely within the generative modeling toolkit.

A more geometric reformulation appears in Diffusion Schrödinger Bridge Matching (DSBM), which introduces Iterative Markovian Fitting (IMF). IMF alternates between a Markovian projection and a reciprocal projection, and the corresponding KL Pythagorean identities imply monotonic convergence to the SB fixed point [2303.16852]. Unlike classical IPF, IMF preserves the endpoint marginals at every iterate, and DSBM implements the Markovian projection by regression against the bridge score induced by the reference process [2303.16852]. The paper explicitly recovers flow matching, rectified flow, and aligned bridge learning as special or limiting cases, thereby placing several later transport methods inside an SB geometry [2303.16852].

A distinct line of work replaces iterative reciprocal/Markovian alternation with direct matching. “Optimal Schrödinger bridge matching” proves that the optimal projection of any reciprocal process \(T_\pi\), built from any accessible transport plan \(\pi\), is exactly the SB \(T^\star\). The practical objective reduces to a bridge-matching mean-squared error, and the paper further shows that this objective is equivalent, up to a constant, to the LightSB energy-based objective [2402.03207]. The corresponding practical solver, LightSB-M, uses a Gaussian mixture parameterization of the adjusted Schrödinger potential, which makes the normalizing integral and the induced drift tractable in closed form [2402.03207].

Several papers simplify the original DSB objectives by reparameterizing the prediction target. “Simplified Diffusion Schrödinger Bridge” replaces the original DSB target by direct next-state prediction and argues that score-based generative models are natural initial solutions for DSB; a pretrained SGM can then be refined by alternating SB training [2403.14623]. A later paper extends this logic through three reparameterizations—Iterative Proportional Mean-Matching, Iterative Proportional Terminus-Matching, and Iterative Proportional Flow-Matching—and uses pre-trained SGMs as initialization for SB-based models [2508.18095]. The stated effect is to accelerate and stabilize training by aligning SB targets with diffusion-model denoising or flow targets [2508.18095].

At the opposite extreme from neural iterative training, one work derives a one-step data-driven SB diffusion process whose drift can be estimated directly from samples using a closed-form formula based on the reference transition kernel. This avoids neural network training altogether and produces a simulation-free, computationally light sampler [2405.12453]. Another efficiency-oriented development, Quantized Diffusion Schrödinger Bridges, observes that simulation-free SB methods still require paired endpoints through entropic OT; it therefore computes the endpoint coupling on anchor-quantized endpoint distributions, proves stability of the regularized coupling under quantization, and then lifts the plan back to raw data points through cell-wise sampling [2605.11983].

## 4. Structural variants and generalized bridge models

A substantial portion of the literature generalizes diffusion-based SB beyond the balanced Brownian setting. One important extension adds a quadratic state cost-to-go,
$$
q(x)=\tfrac12 x^\top Qx,
$$
to the control objective. This regularized SB induces forward and backward reaction-diffusion equations rather than pure heat equations, and the state penalty acts like a state-dependent killing or creation rate of probability mass [2406.00503]. The key result is an explicit closed-form Markov kernel for the corresponding reaction-diffusion operator, which makes the regularized bridge exactly solvable, even for non-Gaussian endpoints, and recovers the heat kernel in the limit \(Q\downarrow 0\) [2406.00503].

A second extension softens the terminal marginal constraint. Soft-constrained Schrödinger Bridge replaces the hard condition \(L(X_T^u)=\mu_T\) by a KL penalty between the controlled terminal law and the target. The solution remains an \(h\)-transform diffusion, but the optimal terminal law becomes a geometric mixture of the target law and a reference-induced terminal law [2403.01717]. This construction is explicitly proposed as a route toward robust generative diffusion when the hard SB constraint is either too brittle or not feasible [2403.01717].

A third extension removes mass conservation. Unbalanced Diffusion Schrödinger Bridge augments the state space with a coffin state and derives time reversal for diffusions with killing and birth. In this formulation, forward killing becomes backward birth, and finite endpoint measures of different masses can be connected within an SB framework [2306.09099]. The resulting model is designed for populations with genuine birth, death, proliferation, or extinction events, where balanced probability-preserving transport is structurally inappropriate [2306.09099].

Other structural variants adapt SB to specific data geometries or supervision regimes. Riemannian Diffusion Schrödinger Bridge generalizes DSB to compact manifolds, replacing Euclidean Gaussian perturbations by geodesic random walks and manifold score/divergence operators [2207.03024]. Aligned Diffusion Schrödinger Bridges assumes access to paired samples \((x_0^i,x_1^i)\), interprets them as samples from the optimal coupling, and uses Doob’s \(h\)-transform to derive an aligned bridge-matching loss that bypasses IPF [2302.11419].

| Variant | Key modification | Representative paper |
|---|---|---|
| Regularized SB | Quadratic state cost, reaction-diffusion kernel | [2406.00503] |
| Soft-constrained SB | Terminal KL penalty, geometric-mixture terminal law | [2403.01717] |
| Unbalanced DSB | Killing/birth with coffin state | [2306.09099] |
| Riemannian DSB | Compact-manifold bridge with geodesic random walk | [2207.03024] |
| Aligned DSB | Uses paired endpoint supervision | [2302.11419] |

## 5. Energy-based sampling, Bayesian computation, and non-memoryless bridges

A major recent direction uses diffusion-based SB for targets known only through energies or unnormalized densities. For Bayesian computation and posterior sampling, one paper proposes replacing long-horizon denoising by a finite-time SB between a Gaussian reference and the target or posterior. The resulting methods, DDPS, DSB-PS, DDGS, and DSB-GS, reinterpret posterior and general sampling as reverse-time SDE learning plus SB-based finite-time refinement [2308.14106].

For Boltzmann sampling, Adjoint Schrödinger Bridge Sampler formulates the target as \(\nu(x)\propto e^{-E(x)}\), learns the SB drift by Adjoint Matching, and learns a corrector term by Corrector Matching, without requiring target samples during training [2506.22565]. The paper states that this relaxes the memoryless condition that restricted earlier matching-based diffusion samplers and allows arbitrary source distributions, including Gaussian and harmonic priors [2506.22565]. It also proves global convergence to the SB solution under the condition that the matching steps reach critical points [2506.22565].

Adjoint Schrödinger Bridge Matching pushes the non-memoryless viewpoint into high-dimensional generative modeling. It first learns a forward controlled process from data to an energy-defined prior, thereby inducing an informative endpoint coupling, and only then learns the backward generative process by bridge matching [2602.15396]. The paper explicitly contrasts this with standard diffusion models, in which the forward process is memoryless and induces independent data-noise coupling. It argues that the non-memoryless regime yields more localized couplings, less transport cost, straighter trajectories, and fewer function evaluations, and reports improved fidelity with fewer sampling steps on image generation benchmarks [2602.15396].

These works suggest a broader reinterpretation of SB: not merely as an interpolation device between two empirical marginals, but as a scalable sampler for energy-based, posterior, and domain-specific prior distributions. A plausible implication is that the boundary between diffusion samplers and SB solvers is now largely methodological rather than conceptual.

## 6. Applications, empirical scope, and recurring technical issues

Diffusion-based SB has been applied well beyond unconditional image generation. In time series, SBTS defines a path-space bridge whose discrete-time marginal matches the full joint law of the observed sequence, not merely an endpoint pair. The optimal drift is generally path-dependent, and the paper estimates it either by kernel regression or by LSTM-based approximation before sampling new trajectories with an Euler scheme [2304.05093]. Reported experiments cover autoregressive processes, GARCH, fractional Brownian motion, deep hedging, and sequential image generation [2304.05093].

In medical imaging, guided MRI reconstruction is formulated as a bridge between a guiding image distribution and a target contrast distribution. The method combines SB sampling with conjugate-gradient data consistency and introduces an \(I^2\)SB-inversion procedure to better address structural differences between contrasts [2411.14269]. On paired T1 and T2-FLAIR data, the paper reports acceleration up to 14.4 and better reconstruction accuracy and stability than the compared methods [2411.14269].

Biological and molecular applications appear repeatedly. SBalign reports sizeable improvements on protein conformational change prediction and lineage-traced cell differentiation by using aligned supervision [2302.11419]. Unbalanced DSB is evaluated on heterogeneous molecular single-cell responses to cancer drugs and on the emergence and spread of viral variants, precisely because these settings involve population-size changes that balanced SB cannot represent [2306.09099]. LightSB-M is tested on continuous SB benchmarks, single-cell trajectory inference, and unpaired image-to-image translation [2402.03207]. ASBS reports results on synthetic energy functions, alanine dipeptide, and amortized conformer generation [2506.22565].

Across this literature, several technical issues recur. First, exactness is often ambiguous: some methods recover the full SB path law, whereas others recover only the marginal path or a projection of it [2503.21756]. Second, simulation-free methods shift the computational bottleneck from trajectory simulation to endpoint coupling estimation, especially entropic OT on unpaired data [2605.11983]. Third, iterative forward-backward training can accumulate approximation error, which motivates single-step matching, reciprocal/Markovian projections, exact kernels, or pretrained initialization [2303.16852][2402.03207][2406.00503][2508.18095]. Fourth, the role of the base process remains contested: memoryless bases simplify training but induce independent endpoint pairing, whereas non-memoryless bases can yield straighter and more efficient trajectories at the cost of a more complex learning problem [2602.15396][2506.22565].

Taken together, these developments define diffusion-based Schrödinger Bridge as a broad research program rather than a single algorithm. Its core invariant is the finite-horizon KL projection of path measures under endpoint constraints; its diversity lies in how those constraints are relaxed, parameterized, matched, or solved in practice.

Source: https://www.emergentmind.com/topics/diffusion-based-schrodinger-bridge-sb