diffqcp: Differentiating through quadratic cone programs

diffqcp is a JAX library to form the derivative of the solution map to a quadratic cone program (QCP) with respect to the QCP problem data as an abstract linear operator and to compute Jacobian-vector products (JVPs) and vector-Jacobian products (VJPs) with this operator. The implementation is based on the derivations in our paper (see below) and computes these products implicitly via projections onto cones and sparse linear system solves. Our approach therefore differs from libraries that compute JVPs and VJPs by unrolling algorithm iterates. We directly exploit the underlying structure of QCPs.

Features include: - Hardware acclerated: JVPs and VJPs can be computed on CPUs, GPUs, and (theoretically) TPUs. - Support for many canonical classes of convex optimization problems including - linear programs (LPs), - quadratic programs (QPs), - second-order cone programs (SOCPs), - and semidefinite programs (SDPs).

Quadratic cone programs

A quadratic cone program is given by the primal and dual problems

math \begin{equation*} \begin{array}{lll} \text{(P)} \quad &\text{minimize} \; & (1/2)x^T P x + q^T x \\ &\text{subject to} & Ax + s = b \\ & & s \in \mathcal{K}, \end{array} \qquad \begin{array}{lll} \text{(D)} \quad &\text{maximize} \; & -(1/2)x^T P x -b^T y \\ &\text{subject to} & Px + A^T y = -q \\ & & y \in \mathcal{K}^*, \end{array} \end{equation*} where $x \in \mathbf{R}^n$ is the primal variable, $y \in \mathbf{R}^m$ is the dual variable, and $s \in \mathbf{R}^m$ is the primal slack variable. The problem data are $P\in \mathbf{S}_+^{n}$, $A \in \mathbf{R}^{m \times n}$, $q \in \mathbf{R}^n$, and $b \in \mathbf{R}^m$. We assume that $\mathcal K \subseteq \mathbf{R}^m$ is a nonempty, closed, convex cone with dual cone $\mathcal{K}^*$.

diffqcp currently supports QCPs whose cone is the Cartesian product of the zero cone, the positive orthant, second-order cones, and positive semidefinite cones. Support for exponential and power cones (and their dual cones) is in development. For more information about these cones, see the appendix of our paper.

Usage

diffqcp is meant to be used as a CVXPYlayers backend --- it is not designed to be a stand-alone library. Nonetheless, here is how it use it. (Note that while we'll specify different CPU and a GPU configurations, all modules are CPU and GPU compatible--we just recommend the following as JAX's BCSR arrays do have CUDA backends for their mv operations while the BCOO arrays do not.)

For both of the following problems, we'll use the following objects:

```python import cvxpy as cvx

problem = cvx.Problem(...) probdata, _, _ = problem.getproblemdata(cvx.CLARABEL, solveropts={'usequadobj': True}) scscones = cvx.reductions.solvers.conicsolvers.scsconif.dimstosolverdict(prob_data["dims"])

x, y, s = ... # canonicalized solutions to problem ```

Optimal CPU approach

If computing JVPs and VJPs on a CPU, we recommend using the equinox.Modules HostQCP and QCPStructureCPU as demonstrated in the following pseudo-example.

```python from diffqcp import HostQCP, QCPStructureCPU from jax.experimental.sparse import BCOO from jaxtyping import Array

P: BCOO = ... # Only the upper triangular part of the QCP matrix P A: BCOO = ... q: Array = ... b: Array = ...

problemstructure = QCPStructureCPU(P, A, scscones) qcp = HostQCP(P, A, q, b, x, y, s, problem_structure)

Compute JVPs

dP: BCOO ... # Same sparsity pattern as P dA: BCOO = ... # Same sparsity pattern as A db: Array = ... dq: Array = ...

dx, dy, ds = qcp.jvp(dP, dA, dq, db)

Compute VJPs

dP, dA will be BCOO arrays, dq, db just Arrays

dP, dA, dq, db = qcp.vjp(f1(x), f2(y), f3(s)) ```

Optimal GPU approach

If computing JVPs and VJPs on a GPU, we recommend using the equinox.Modules QCPStructureGPU and DeviceQCP.

```python from diffqcp import DeviceQCP, QCPStructureGPU from jax.experimental.sparse import BCSR from jaxtyping import Array

P: BCSR = ... # The entirety of the QCP matrix P A: BCSR = ... q: Array = ... b: Array = ...

problemstructure = QCPStructureGPU(P, A, scscones) qcp = DeviceQCP(P, A, q, b, x, y, s, problem_structure)

Compute JVPs

dP: BCSR ... # Same sparsity pattern as P dA: BCSR = ... # Same sparsity pattern as A db: Array = ... dq: Array = ...

dx, dy, ds = qcp.jvp(dP, dA, dq, db)

Compute VJPs

dP, dA will be BCSR arrays, dq, db just Arrays

dP, dA, dq, db = qcp.vjp(f1(x), f2(y), f3(s)) ```

Citation

arXiv:2508.17522 [math.OC] @misc{healey2025differentiatingquadraticconeprogram, title={Differentiating Through a Quadratic Cone Program}, author={Quill Healey and Parth Nobel and Stephen Boyd}, year={2025}, eprint={2508.17522}, archivePrefix={arXiv}, primaryClass={math.OC}, url={https://arxiv.org/abs/2508.17522}, }

Next steps

diffqcp is still in development! WIP features and improvements include: - Support for the exponential cone, the power cone, and their dual cones. - Batched problem computations. - Migration of tests from our torch branch. - Heuristic JVP and VJP computations when the solution map of a QCP is non-differentiable.

See also

Core dependencies (diffqcp makes essential use of the following libraries) - Equinox: Neural networks and everything not already in core JAX (via callable PyTrees). - Lineax: Linear solvers.

Related - CVXPYlayers: Construct differentiable convex optimization layers using CVXPY. (WIP: diffqcp is being added as a backend for CVXPYlayers.) - CuClarabel: The GPU implemenation of the second-order QCP solver, Clarabel. - SCS: A first-order QCP solver that has an optional GPU-accelerated backend. - diffcp: A (Python with C-bindings) library for differentiating through (linear) cone programs.