Total: 1
Sample-based quantum subspace methods construct a small subspace from measured basis states and solve the target problem classically within it. We study these methods for heat conduction in magnetized plasmas, where the anisotropy $χ_\parallel/χ_\perp$ reaches $10^8$ to $10^{10}$. We prove two results for the energy-norm Galerkin solution. First, if the measurement basis is aligned with the flux surfaces and the subspace contains the poloidal-mode-zero block, the error is bounded by $\|r_0\|_{L^+}(χ_\parallel-χ_\perp)^{-1/2}$, where the numerator does not depend on $χ_\parallel$. The bound is sharp, so the accuracy improves as the anisotropy grows. Second, in the computational basis, a subspace that misses a single node of a flux surface suppresses the flux-function component there as $χ_\parallel^{-1/2}$, which gives an error floor when the limiting solution is nonzero on that surface. We verify both results on a shaped cross-section and on a KSTAR-like Miller equilibrium with ten qubits. At $χ_\parallel/χ_\perp=10^9$, the computational basis gives an error of order unity for every shot budget, whereas a quantum Fourier transform on the poloidal register gives $1.3\times10^{-5}$. The same barrier limits sample-based quantum diagonalization. The aligned measurement needs no time evolution, so the circuit is only state preparation and a five-qubit quantum Fourier transform. On a ten-qubit superconducting processor, the aligned basis gives a median error of $7.0\times10^{-6}$ with 768 shots, whereas the computational basis gives errors above 0.94 up to $1.2\times10^4$ shots. A trapped-ion processor shows the same behavior.