Theory of optical long-baseline interferometry on polarized sources

This paper presents a comprehensive theory for optical long-baseline interferometry on polarized sources by introducing a generalized Mueller matrix formalism that relates observed and object Stokes visibilities, demonstrating the necessity of correcting for polarization crosstalk to accurately recover complex visibilities even for unpolarized sources.

The Gist

Imagine you are trying to take a photograph of a distant, twinkling star using two giant telescopes working together as one super-telescope. This is called interferometry. By combining the light from these two telescopes, astronomers can see details so fine that they could read a license plate on a car from hundreds of miles away.

However, light isn't just a simple wave; it has a "twist" to it called polarization. Think of light waves like ropes. You can shake a rope up and down (vertical polarization), side to side (horizontal polarization), or in a circle (circular polarization). Most stars emit light that is a messy mix of all these directions (unpolarized), but some cosmic objects, like the super-hot gas swirling around black holes, emit light that is very neatly organized in one specific direction (highly polarized).

The Problem: The "Twisty" Hallway

In the past, building these optical super-telescopes was like trying to run a race through a hallway filled with mirrors, prisms, and glass windows. Every time the light beam hit a mirror or passed through glass, it got a little "twisted" or "stretched."

If the two telescopes had slightly different hallways (one mirror was slightly tilted compared to the other), the light waves would arrive at the meeting point out of sync. It's like two runners trying to high-five, but one is wearing shoes that make them run sideways while the other runs straight. They miss the high-five, and the image gets blurry or disappears.

For a long time, astronomers tried to solve this by building perfectly symmetrical hallways so the twists canceled out. This worked for simple stars. But now, we are looking at fainter, more exotic objects that have strong "twists" (polarization). Even with symmetrical hallways, the instrument itself can create "ghost" signals—fake polarization patterns that look real but are just artifacts of the mirrors and glass.

The Solution: A New "Translation Dictionary"

This paper, written by Guy Perrin, introduces a new mathematical "dictionary" to translate what the telescope sees into what the star is actually doing.

Here is the core idea broken down with analogies:

1. The "Stokes Visibility" (The Four-Part Message)
Traditionally, astronomers measured just one thing: the brightness of the interference pattern (like the volume of a sound).
Perrin suggests we need to measure four things at once, like a four-channel radio broadcast:

• Channel I: The total brightness (Volume).
• Channel Q: How much the light is shaking up/down vs. side-to-side.
• Channel U: How much the light is shaking at a 45-degree angle.
• Channel V: How much the light is spinning in a circle.

These four channels together tell the full story of the star's polarization.

2. The "Generalized Mueller Matrix" (The Translator)
The telescope acts like a noisy translator. If you whisper a message (the star's light) into one end, the machine twists it before it comes out the other end.

• Old way: We assumed the machine was perfect or just slightly broken, so we tried to ignore the noise.
• New way: Perrin created a Generalized Mueller Matrix. Think of this as a specific set of instructions or a "decoder ring" that knows exactly how your specific telescope twists the light.

By applying this decoder ring to the messy data coming out of the telescope, we can mathematically "un-twist" the signal. We can separate the real signal from the "ghost" signals created by the instrument.

3. The "Ghost Polarization" (The Illusion)
The paper makes a crucial, surprising point: Even if a star is completely un-polarized (a messy mix of all directions), the telescope can trick you into thinking it is polarized.

Imagine you are looking at a white light bulb through a pair of sunglasses that are slightly crooked. Even though the bulb is white, the sunglasses might make it look slightly blue or green.

• In the past, astronomers might have thought, "Oh, the star is blue!"
• Perrin's math says: "No, the star is white. Your sunglasses (the telescope) are just adding a blue tint. Let's subtract that tint to see the truth."

Why This Matters

This new theory is like upgrading from a black-and-white camera to a high-definition, 3D camera with noise-canceling headphones.

• For Radio Astronomers: They have been doing this for decades because radio waves are easier to measure directly.
• For Optical Astronomers: This paper finally brings that same level of precision to visible light.

Now, when we look at the center of our galaxy or the flares from black holes, we won't just see where the light is coming from; we will be able to map exactly how the light is vibrating. This helps us understand the magnetic fields and the extreme physics happening in the most violent parts of the universe, free from the "ghosts" and "twists" introduced by our own instruments.

In short: This paper gives astronomers a new mathematical tool to clean up the "static" in their telescopes, allowing them to see the true, twisted nature of light from the most extreme objects in the universe.

Technical Summary

1. Problem Statement

Optical long-baseline interferometry (OLBI) faces significant challenges when observing sources with high degrees of polarization (e.g., synchrotron sources like Galactic Center flares). While radio interferometry has long-established formalisms for handling polarization, optical interferometry has historically relied on designing symmetrical optical trains to minimize polarization effects (differential birefringence, rotation, and coating differences) to preserve fringe contrast.

However, with the advent of large telescopes and sensitive instruments (e.g., VLTI/GRAVITY, CHARA/MIRC-X), it is now possible to observe faint, highly polarized sources. The existing optical formalisms are often insufficient because:

• They do not fully account for the complex polarization characteristics of non-ideal beam trains.
• They fail to provide a unified framework for deriving the spatial coherence of polarized sources.
• They often neglect "polarization crosstalk," where instrumental effects create "ghost" polarized visibilities even when observing unpolarized sources, leading to biased measurements of complex visibilities (moduli, phases, and closure phases).

2. Methodology

The paper establishes a comprehensive analytical formalism by adapting concepts from radio aperture synthesis (specifically the Radio Interferometer Measurement Equation, or RIME) to the optical domain.

• Jones Vector Formalism: The author describes monochromatic polarized waves as complex Jones vectors. The spatial distribution of the source is integrated over the field of view.
• Coherence Matrix: The spatial and polarization coherence is characterized using the $2\times2$ coherence matrix ($C$), defined as the ensemble average of the outer product of the Jones vector and its conjugate transpose ($C = \langle \mathbf{E}\mathbf{E}^\dagger \rangle$).
• Stokes Visibility Generalization: The paper defines "Stokes visibilities" ($V_I, V_Q, V_U, V_V$) as the normalized Fourier transforms of the Stokes parameters ($I, Q, U, V$). Unlike radio conventions where coherence matrices are sometimes called visibility matrices, this paper strictly separates coherent flux (correlations) from normalized visibilities to account for atmospheric turbulence and single-mode fiber fluctuations common in optics.
• Propagation via Jones Matrices: The propagation of waves through the interferometer's optical train is modeled using $2\times2$ complex Jones matrices ($J_n$ and $J_m$) for each beam path.
• Generalized Mueller Matrix: A key theoretical step is the derivation of a "Generalized Mueller Matrix" ($M_{nm}$). This $4\times4$ matrix relates the observed Stokes visibilities to the intrinsic object Stokes visibilities, accounting for the specific baseline geometry and the polarization properties of the optical system.

3. Key Contributions

The paper introduces several novel theoretical constructs and relationships:

• Unified Matrix Relationship: The derivation of a linear matrix equation (Eq. 21 & 22) linking the observed coherence vector ($\tilde{C}_{nm}$) to the object's Stokes visibility vector ($V_{nm}$) via a transfer matrix ($T_{nm}$).
• Generalized Mueller Matrix ($M_{nm}$): Defined as $M_{nm} = T_{I2}^{-1} T_{nm}$, this matrix describes how the instrument modifies the Stokes visibilities. It generalizes the standard single-beam Mueller matrix to the multi-aperture interferometric case.
• Polarization Crosstalk Mechanism: The formalism explicitly demonstrates that even for unpolarized sources ($Q=U=V=0$), non-ideal optical trains (with differential birefringence or rotation) generate non-zero off-diagonal terms in the coherence matrix. This creates "ghost" polarized visibilities that bias the measurement of the classical intensity visibility ($V_I$).
• Calibration Framework for Single-Mode Interferometers: The paper provides a specific calibration protocol for single-mode interferometers (where phase fluctuations are traded for intensity fluctuations). It derives equations to:
  • Correctly estimate gain factors ($g_n$) in the presence of polarization.
  • De-bias the squared moduli of visibilities.
  • Recover unbiased differential phases and closure phases by applying the inverse of the Generalized Mueller Matrix to the observed data.

4. Results and Examples

The author applies the formalism to several specific scenarios to illustrate its utility:

• Symmetric Interferometers: Even in perfectly symmetric interferometers without diattenuation, the optical beam properties (phase retardance) do not cancel out for the polarized part of the visibilities. Calibration is still required unless the source is strictly unpolarized.
• Polarized Point Sources: The model shows that the visibility of an unpolarized source is offset by the visibility of a polarized point source, plus an additional term proportional to the source's Stokes parameters.
• Extended Sources with Uniform Polarization: For extended sources with uniform polarization properties, the Stokes visibilities are proportional to the Fourier transform of the object's shape. The paper notes that at high spatial frequencies (beyond the object's size), the cross-coherence is less biased by Stokes visibilities, but photometric calibration remains sensitive to source polarization.
• Quasi-Unitary Matrices: The paper references previous work showing that VLTI beam trains can be approximated by quasi-unitary matrices, allowing for the practical calculation of the transfer matrix $T_{nm}$ using Pauli matrices.

5. Significance

This paper is significant for the future of high-angular-resolution astronomy for several reasons:

• Enabling Polarimetric Imaging: It provides the necessary theoretical foundation to perform accurate polarimetric imaging with optical interferometers, which is crucial for studying magnetic fields in accretion disks, jets, and stellar surfaces.
• Correction of Systematic Errors: It highlights that classical complex visibilities (amplitude and phase) are inherently biased by instrumental polarization crosstalk, even for unpolarized targets. Ignoring this leads to incorrect image reconstruction and closure phase measurements.
• Standardization: By adopting and adapting the rigorous formalism used in radio astronomy (Jones/Mueller matrices), the paper offers a consistent language and mathematical framework for the optical community, bridging the gap between the two domains.
• Instrument Design and Calibration: The derived equations allow for the development of robust calibration pipelines for current and future instruments (e.g., upgrades to VLTI/GRAVITY or CHARA), ensuring that the full potential of polarized source observations is realized.

In conclusion, Perrin establishes that to fully exploit the capabilities of modern optical interferometers, one must move beyond scalar visibility treatments and adopt a vectorial approach that simultaneously solves for the spatial structure and the polarization state of the source, correcting for the instrument's specific polarization transfer function.