Pathwise Conditioning Methods

• Pathwise conditioning is a method that systematically rewrites expectations or dynamic programs at the trajectory level to bypass discontinuities in marginal distributions.
• It is applied across domains such as financial Monte Carlo, Gaussian process regression, and stochastic control to smooth functions, reduce variance, and stabilize derivative estimators.
• By converting ill-posed, noisy pathwise problems into tractable representations, it facilitates efficient computation and improved accuracy in complex modeling tasks.

Pathwise conditioning is a domain-dependent term for rewriting an expectation, control problem, posterior law, or dynamic program at the level of trajectories, sampled paths, or pathwise linear systems rather than only at the level of marginal distributions. Across quantitative finance, Gaussian processes, stochastic control, backward dynamic programming, and related Monte Carlo sensitivity methods, the common pattern is to condition on enough path structure to obtain a smoother, lower-variance, better-conditioned, or more directly evaluable representation. The term is therefore polysemous rather than uniform: in some settings it means conditional Monte Carlo smoothing of discontinuous path functionals, in others a pathwise random-function representation of a conditional Gaussian law, and in others a quenched formulation obtained by freezing one source of noise and only then randomizing it (Gerstner et al., 2018, Wilson et al., 2020, Friz et al., 2024).

1. Conceptual scope and recurring structure

The broadest shared idea is that a problematic object is replaced by a pathwise representation in which the dependence on randomness or parameters is handled more tractably. In discretely monitored barrier-option Monte Carlo, the discontinuous barrier indicator is replaced by conditional survival probabilities, so that differentiation becomes possible through a smooth recursion (Gerstner et al., 2018). In Gaussian-process regression, a posterior sample is represented as a prior sample plus a data-dependent correction, so that posterior sampling is performed at the level of random functions rather than only finite-dimensional conditional covariances (Wilson et al., 2020). In rough stochastic control, one first fixes the realization of the noise to be conditioned on, solves the control problem pathwise for that deterministic rough path, and only afterward randomizes that path (Friz et al., 2024).

A useful way to compare usages is to distinguish the object being conditioned on, the representation produced, and the intended benefit.

Domain	Conditioned object	Main benefit
Barrier-option Monte Carlo	one-step barrier survival	smooth pathwise Greeks
Gaussian processes	prior sample path and observations	solve-once, evaluate-many-times posterior sampling
Rough stochastic control	fixed realization of conditioned-on noise	quenched formulation of conditional control
Hedge-ratio reduction	empirical averages across paths	lower-dimensional, more stable linear systems
BSDE dynamic programs	pathwise martingale/control corrections	iterative upper and lower bounds

These uses share a family resemblance but not a single formal definition. This suggests that “pathwise conditioning” is best understood as an umbrella term for methods that move conditioning from abstract distributions to explicit pathwise constructions. A plausible implication is that the term is most informative when accompanied by the specific representation being conditioned—survival probabilities, prior sample paths, fixed rough paths, or empirical residual moments—rather than used in isolation.

2. Conditional smoothing of discontinuous path functionals

In Monte Carlo Greeks for discretely monitored barrier options, pathwise conditioning is a concrete smoothing device for discontinuous payoffs. The one-dimensional Black–Scholes setting considered in "Monte Carlo pathwise sensitivities for barrier options" (Gerstner et al., 2018) uses the discrete-time evolution

$$S_{j+1} = S_j \exp\!\left(\left(\mu-\frac{\sigma^2}{2}\right)\Delta t+\sigma\sqrt{\Delta t}\,Z_j\right), \qquad Z_j\sim N(0,1)\ \text{i.i.d.}$$

and the main example is a discretely monitored single-asset knock-up-out call with payoff

$$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$

Standard pathwise differentiation fails because the indicator

$\mathbf 1_{\{\max_j S_j\le B\}}$

makes the payoff discontinuous as a functional of the simulated path (Gerstner et al., 2018).

The paper’s remedy is to condition each step on survival below the barrier. For the up-and-out case, the one-step survival probability is

$$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$

and the truncated normal is mapped to a parameter-independent domain by

$z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$

Iterating this yields the conditional representation

$$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$

The barrier discontinuity is therefore absorbed into smooth functions $p_t$ and the transformed recursion (Gerstner et al., 2018).

This construction is explicitly described as a combination of the one-step survival idea of Glasserman–Staum with “stable differentiation.” After transformation, differentiation proceeds through the smooth recursion

$$\frac{\partial p_t}{\partial \Theta_i} = \frac{\partial f}{\partial s}\frac{\partial S_t}{\partial \Theta_i} +\frac{\partial f}{\partial \vartheta_i}, \qquad \frac{\partial S_{t+1}}{\partial \Theta_i} = \frac{\partial g}{\partial s}\frac{\partial S_t}{\partial \Theta_i} +\frac{\partial g}{\partial \pi}\frac{\partial p_t}{\partial \Theta_i} +\frac{\partial g}{\partial \vartheta_i},$$

rather than through a discontinuous payoff map (Gerstner et al., 2018). The resulting Greek estimator is pathwise after conditioning and introduces no likelihood-ratio term. The paper emphasizes that this avoids bump-size tuning and removes discretization error in the Greek itself. It also reports stable Greeks where standard Monte Carlo finite-difference Delta is unstable, and lower CPU times than finite differences in the reported experiments (Gerstner et al., 2018).

This is perhaps the clearest canonical use of pathwise conditioning: before differentiating, one conditions on enough of the path structure to remove the discontinuity that blocks pathwise differentiation. The conceptual sequence given there is explicit: raw barrier payoff is discontinuous; condition on one-step survival at each monitoring date; rewrite the price as expectation of a smooth weighted payoff on $[0,1]^T$; apply recursive pathwise differentiation (Gerstner et al., 2018).

3. Conditioning as reduction of pathwise linear solves

A distinct use appears in "Faster Forward Sensitivities: Reduced stochastic hedge ratios from pathwise algorithmic differentiation" (Fries, 13 May 2026). There, Monte Carlo and pathwise AD have already produced sensitivities with respect to model primitives,

$$b_{\ell i}:=\frac{dV(t,\omega_\ell)}{dM_i(t,\omega_\ell)},$$

and the remaining task is to convert them into hedge ratios with respect to market instruments,

$$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$0

through the pathwise systems

$$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$1

The paper identifies these per-path solves as potentially “expensive, unstable, and unnecessarily high-dimensional” (Fries, 13 May 2026).

The proposed “reduced stochastic hedge ratios” take the form

$$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$2

with $$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$3. Conditioning here is not Rao–Blackwellization over latent variables but a replacement of many pathwise, often ill-posed solves by lower-dimensional systems built from empirical averages across paths (Fries, 13 May 2026). Two coefficient criteria are then distinguished.

The first minimizes the “full empirical pathwise residual”

$$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$4

leading to normal equations

$$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$5

The second enforces a “projected moment equation”

$$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$6

yielding a reduced system

$$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$7

The paper stresses that the projected formulation “avoids products $$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$8” and can be cheaper to assemble, while the least-squares route can worsen conditioning because it “involves products $$V(S_1,\dots,S_T)= \begin{cases} (S_T-K)^+ =: q(S_T), & \text{if } \max_{j=1,\dots,T} S_j \le B,\ 0, & \text{otherwise.} \end{cases}$$9, may square condition numbers if solved through normal equations, and depends on the scaling of the primitive components” (Fries, 13 May 2026).

In this setting, pathwise conditioning has several layers. At the original level, each $\mathbf 1_{\{\max_j S_j\le B\}}$0 may be singular, nearly singular, noisy, or non-unique. At the reduced level, the matrices $\mathbf 1_{\{\max_j S_j\le B\}}$1 and $\mathbf 1_{\{\max_j S_j\le B\}}$2 can themselves be ill-conditioned or rank deficient, but they are built from empirical averages and therefore pool information across paths (Fries, 13 May 2026). The basis $\mathbf 1_{\{\max_j S_j\le B\}}$3 is explicitly treated as a conditioning tool: empirical orthonormality is “mainly a notational and conditioning convenience,” while too-rich bases can “overfit Monte-Carlo noise and lead to ill-conditioned reduced systems” (Fries, 13 May 2026).

This usage is narrower than the barrier-option one but structurally similar: an unstable pathwise problem is stabilized by aggregating across paths before solving. A plausible implication is that here “conditioning” names a move from local pathwise inversion to global empirical regression or projection. The paper is explicit that the reduced equations can “remain well-defined when the pathwise systems are singular, non-unique, or noisy” (Fries, 13 May 2026).

4. Pathwise conditioning as quenched formulation and pathwise geometry

In rough stochastic control, pathwise conditioning is formulated as a quenched problem. "Controlled rough SDEs, pathwise stochastic control and dynamic programming principles" (Friz et al., 2024) studies a partially controlled diffusion

$\mathbf 1_{\{\max_j S_j\le B\}}$4

with $\mathbf 1_{\{\max_j S_j\le B\}}$5 the unconditioned Brownian noise and $\mathbf 1_{\{\max_j S_j\le B\}}$6 the noise one conditions on. The classical value is written as

$\mathbf 1_{\{\max_j S_j\le B\}}$7

The paper’s central move is to replace $\mathbf 1_{\{\max_j S_j\le B\}}$8 by a deterministic rough path $\mathbf 1_{\{\max_j S_j\le B\}}$9 and solve, for fixed $$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$0, the controlled rough SDE

$$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$1

with rough value function

$$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$2

Only afterward is $$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$3 randomized to $$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$4 (Friz et al., 2024).

The bridge to the conditional problem is the identity

$$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$5

and the paper’s main value-function identification is

$$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$6

(Friz et al., 2024). The abstract states the conceptual reason succinctly: Brownian statistics for the to-be-conditioned-on noise are not required, aligned with the “pathwise” intuition that these should not matter upon conditioning (Friz et al., 2024).

This meaning of pathwise conditioning is closely related to earlier deterministic pathwise analysis. "Pathwise integration with respect to paths of finite quadratic variation" (Ananova et al., 2016) does not define a pathwise conditional expectation operator, but it develops a pathwise integral

$$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$7

for gradient-type non-anticipative integrands, proves the pathwise isometry

$$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$8

and derives the decomposition

$$p_0 = \mathbb P(S_1<B\mid S_0) = \Phi\!\left( \frac{\log(B/S_0)-\left(\mu-\frac{\sigma^2}{2}\right)\Delta t}{\sigma\sqrt{\Delta t}} \right),$$9

The paper explicitly interprets this as a “signal plus noise decomposition” (Ananova et al., 2016). Although no conditioning operator is defined, the induced quadratic-variation geometry supplies a projection-like structure: regular path functionals are split into a rough component generated by the pathwise integral and a residual with zero quadratic variation (Ananova et al., 2016). This suggests a deterministic analogue of conditional decomposition without introducing probabilistic averaging.

A counterpoint is provided by "Sharp pathwise nonuniqueness for additive SDEs" (Hess-Childs et al., 26 Apr 2026). For additive-noise equations

$z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$0

the paper asks whether conditioning on the whole noise path $z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$1 determines the trajectory $z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$2. Its answer is negative below the classical threshold: for every $z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$3, in the Brownian theorem realized as $z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$4, $z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$5, there exists a drift for which there is a unique weak solution but pathwise uniqueness fails, hence no strong solution (Hess-Childs et al., 26 Apr 2026). In the paper’s own pathwise-conditioning interpretation, even after conditioning on the entire Brownian trajectory, multiple adapted solution trajectories may coexist (Hess-Childs et al., 26 Apr 2026). This is a useful corrective to any overly general slogan that conditioning on the full noise path must determine the state path.

5. Pathwise conditioning of Gaussian-process posteriors

In Gaussian-process regression, pathwise conditioning denotes a random-function representation of the posterior. "Pathwise Conditioning of Gaussian Processes" (Wilson et al., 2020) and the later dissertation "Scalable Gaussian Processes: Advances in Iterative Methods and Pathwise Conditioning" (Lin, 9 Jul 2025) present the central identity as a sample-path form of Gaussian conditioning. For noisy observations $z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$6, $z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$7, the classical finite-dimensional posterior at test inputs $z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$8 is

$z=\Phi^{-1}(u\,p_0), \qquad u\in(0,1).$9

with

$$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$0

$$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$1

Pathwise conditioning replaces the distribution-first route with the identity

$$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$2

where $$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$3 and $$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$4 are evaluations of the same prior sample path and $$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$5 (Lin, 9 Jul 2025). In function notation,

$$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$6

(Wilson et al., 2020).

The decomposition is explicitly “prior sample + data-dependent update term” (Lin, 9 Jul 2025). Its practical advantage is that the expensive computation is concentrated in the training-side linear solve

$$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$7

after which the same posterior sample can be evaluated at arbitrary locations via

$$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$8

The dissertation emphasizes that this turns posterior sampling into “one linear solve per sample,” after which that sample can be evaluated at many test points cheaply, which is especially useful in Bayesian optimization and Thompson sampling (Lin, 9 Jul 2025).

A central methodological consequence is the decoupling of prior-path approximation from posterior updating. Approximate prior paths may be generated by random Fourier features, Karhunen–Loève expansions, or SPDE/finite element constructions, while the update is kept in the canonical kernel basis or an inducing-point basis (Wilson et al., 2020). The paper explicitly advocates decoupling because unified feature-space conditioning can produce “variance starvation,” whereas using a global basis for the prior and a local basis for the update better matches the geometry of the problem (Wilson et al., 2020). The later dissertation integrates this representation with iterative linear system solvers such as CG, SGD, SDD, and AP, recasting GP posterior sampling as matrix-vector-product-based linear algebra and extending the same idea to inducing-point and latent-Kronecker settings (Lin, 9 Jul 2025).

This GP usage is closely tied to Matheron’s rule. The novelty claimed in the 2020 paper is not the Gaussian identity itself, but articulating Gaussian conditionals at the level of random variables and deriving a general family of approximations for efficiently sampling GP posteriors (Wilson et al., 2020). In this area, pathwise conditioning is therefore a function-level realization of Gaussian conditioning rather than a smoothing or variance-reduction device in the narrow Monte Carlo sense.

6. Adjacent usages, non-usages, and limits of the term

Several nearby literatures use “pathwise” centrally without always using “conditioning” in the strict Rao–Blackwell sense. In "Pathwise Iteration for Backward SDEs" (Bender et al., 2016), nested conditional expectations in convex stochastic dynamic programs

$$PV_{t_0}(S_0) = e^{-r(t_T-t_0)} \int_0^1\cdots\int_0^1 p_0\cdots p_{T-1}\, q\!\bigl(S_T(u^{(T)},\dots,u^{(1)})\bigr)\, du^{(T)}\cdots du^{(1)}.$$9

are replaced by pathwise upper and lower recursions,

$p_t$0

whose conditional expectations yield improved supersolutions and subsolutions (Bender et al., 2016). The method is “pathwise” because it constructs non-adapted pathwise quantities by backward recursion along each simulated trajectory, with martingales acting as control variates. This is close in spirit to pathwise conditioning because deeply nested conditioning is reorganized into local pathwise corrections, but the paper’s own emphasis is on pathwise iteration and bound improvement rather than a named conditioning operator (Bender et al., 2016).

In reinforcement learning, "Relative Entropy Pathwise Policy Optimization" (Voelcker et al., 15 Jul 2025) uses pathwise policy gradients with a reparameterizable stochastic actor,

$p_t$1

so that the actor update backpropagates through

$p_t$2

The paper itself describes the operational stack as action-conditioned critic, policy update conditioned on sampled actions, on-policy targets for action-conditioned value learning, and KL-constrained policy movement (Voelcker et al., 15 Jul 2025). This is a sense of “pathwise conditioning” tied to action-conditioned critics and local validity of $p_t$3, not to conditional expectation identities. The authors argue that stable pathwise updates require an accurate action-conditioned value function and a relative-entropy trust region that keeps the actor near the distribution on which $p_t$4 was trained (Voelcker et al., 15 Jul 2025).

By contrast, "Pathwise Gradient Variance Reduction with Control Variates in Variational Inference" (Ng et al., 2024) is explicitly not a paper on pathwise conditioning in the strict Rao–Blackwell or conditional-expectation sense. It studies variance reduction for reparameterization gradients

$p_t$5

by control variates, including a proposed Stein-based zero-variance control variate construction, but does not explicitly discuss conditioning on part of the reparameterization noise or conditional Monte Carlo (Ng et al., 2024). This distinction matters because the term “pathwise conditioning” is sometimes loosely extended to any variance reduction on pathwise estimators, whereas the paper argues for a stricter separation between conditioning and additive mean-zero corrections (Ng et al., 2024).

A final, older example of genuine path-statistic conditioning is provided by the Anderson polymer model. "Overlaps and Pathwise Localization in the Anderson Polymer Model" (Comets et al., 2011) conditions the polymer measure on the number of jumps $p_t$6, producing a canonical measure and the conditioned free-energy density

$p_t$7

This decomposition yields

$p_t$8

with $p_t$9, separating the canonical energetic gain from the Poisson cost of realizing jump density $$\frac{\partial p_t}{\partial \Theta_i} = \frac{\partial f}{\partial s}\frac{\partial S_t}{\partial \Theta_i} +\frac{\partial f}{\partial \vartheta_i}, \qquad \frac{\partial S_{t+1}}{\partial \Theta_i} = \frac{\partial g}{\partial s}\frac{\partial S_t}{\partial \Theta_i} +\frac{\partial g}{\partial \pi}\frac{\partial p_t}{\partial \Theta_i} +\frac{\partial g}{\partial \vartheta_i},$$0 under the reference walk (Comets et al., 2011). Here the conditioning is neither on observations nor on a source of noise but on a global path statistic.

Taken together, these examples show both the reach and the limits of the term. Pathwise conditioning can denote conditional Monte Carlo smoothing, quenched noise freezing, random-function Gaussian updates, empirical stabilization of pathwise hedge systems, or canonical conditioning on jump counts. It does not, however, automatically include every pathwise variance-reduction method. The most consistent encyclopedia-level characterization is therefore that pathwise conditioning refers to a class of techniques that impose conditioning at the level of trajectories, sampled paths, or pathwise representations in order to replace an ill-posed, discontinuous, expensive, or nonlocal problem by a more tractable one. The exact object being conditioned, and the mathematics of the resulting representation, depend strongly on the field (Gerstner et al., 2018, Friz et al., 2024, Wilson et al., 2020).