XQ100: High-Resolution Quasar Survey

• XQ100 is a high-resolution, blind survey of 100 quasars (3.5 < z < 4.5) that serves as a benchmark for measuring the Lyman-α forest flux power spectrum.
• It utilizes VLT/X-shooter spectroscopy covering redshifts 3.4–4.2 and velocity scales k = 0.003–0.064 s/km, enabling precise constraints on thermal and cosmological parameters.
• The survey’s strength lies in its re-analysis with improved covariance treatment (Wilson, Iršič & McQuinn 2022), ensuring minimal contamination from high-density absorbers.

XQ100 denotes the X-Shooter Quasar Legacy Survey of 100 quasars, originally a blind survey of 100 quasars with $3.5 < z_{\rm qso} < 4.5$, observed with the VLT/X-shooter spectrograph at resolving power $R \sim 4000$–7000. In contemporary Lyman-$\alpha$ forest cosmology, the name also refers to the high-resolution quasar dataset used to measure the small-scale one-dimensional flux power spectrum, $P_{1\mathrm{D}}$, in a regime that is inaccessible to larger but lower-resolution surveys. In the PRIYA analysis, the dataset is represented not by the older XQ100 power-spectrum measurement from Iršič et al. (2017), but by the re-analysis of Wilson, Iršič & McQuinn (2022), which improved the covariance treatment using Ly$\beta$ regions. Within that framework, XQ100 functions as a clean, high-resolution benchmark for constraining the thermal and cosmological structure encoded in the small-scale Lyman-$\alpha$ forest (Ho et al., 22 Sep 2025).

1. Definition and observational basis

XQ100 is a quasar survey designed around high-redshift sightlines and high-resolution spectroscopy. In the analysis under discussion, its practical role is to provide measured values of the one-dimensional Lyman-$\alpha$ forest flux power spectrum in redshift bins together with a covariance matrix. The adopted redshift interval is

$z = 3.4 - 4.2,$

and the velocity-space wavenumber range is

$k = 0.003 - 0.064~{\rm s/km},$

often described in the text as $0.003-0.065~{\rm s/km}$ (Ho et al., 22 Sep 2025).

The survey’s defining observational characteristic is that it is blind-selected rather than assembled from absorber-targeted archival sightlines. That distinction is methodologically important because residual contamination from high-column density absorbers can bias the inferred flux power spectrum, especially at high $R \sim 4000$0. A central conclusion of the PRIYA study is that XQ100’s survey design makes it substantially cleaner than some alternative high-resolution compilations.

A concise summary of the XQ100 dataset as used in the PRIYA study is given below.

Aspect	XQ100 characteristic	Role in analysis
Survey identity	X-Shooter Quasar Legacy Survey of 100 quasars	High-resolution Lyman-$R \sim 4000$1 forest dataset
Quasar redshift selection	$R \sim 4000$2	Defines original survey sample
Spectrograph	VLT/X-shooter, $R \sim 4000$3–7000	Enables small-scale $R \sim 4000$4 measurement
Power-spectrum dataset used	Wilson, Iršič & McQuinn (2022) re-analysis	Improved covariance treatment using Ly$R \sim 4000$5 regions
Fitted redshift range	$R \sim 4000$6	Likelihood domain
Fitted scale range	$R \sim 4000$7	Access to small/non-linear modes

The survey should therefore be understood not simply as a catalog of quasars, but as a calibrated statistical input to flux-power-spectrum inference. Its significance derives less from sample size than from the combination of spectral resolution, redshift coverage, and survey cleanliness.

2. Physical regime probed by XQ100

The main physical contribution of XQ100 is access to the small-scale, non-linear 1D flux power spectrum of the Lyman-$R \sim 4000$8 forest. In the analysis, the authors emphasize that high-resolution spectra such as XQ100 probe scales down to $R \sim 4000$9, corresponding roughly to a few $\alpha$0 in comoving terms, and summarize the reach as “probing down to $\alpha$1 at $\alpha$2” (Ho et al., 22 Sep 2025).

This scale coverage is what distinguishes XQ100 from larger surveys such as eBOSS. The PRIYA eBOSS analysis used only

$\alpha$3

whereas XQ100 extends substantially deeper into the high-$\alpha$4 regime. DESI is described as projected to reach $\alpha$5, still short of the XQ100 upper limit. The resulting complementarity is central: eBOSS constrains cosmology well on larger scales, while XQ100 probes the small-scale sector in which thermal and absorber-related effects become much more prominent.

The relevant small-scale physics includes thermal broadening, pressure smoothing, signatures of inhomogeneous He II reionization, and contamination from Lyman limit systems (LLSs) and other high-column density absorbers (HCDs). The study repeatedly stresses that these effects are not well covered by eBOSS and that this is precisely why XQ100 remains informative despite containing far fewer quasars. A common misconception is therefore that a smaller high-resolution sample must be statistically secondary to a much larger low-resolution survey. In the context of $\alpha$6, that is not correct: XQ100 accesses a different regime of scale sensitivity, particularly for the thermal nuisance sector.

3. Statistical representation and emulator-based interpretation

In the PRIYA framework, XQ100 enters as a likelihood over observed $\alpha$7 bins in redshift and velocity-space wavenumber. The theoretical prediction is supplied by hydrodynamical simulations and a multi-fidelity emulator. The analysis states that the simulated power is evaluated directly on the observed bins: “We chose to compute the $\alpha$8 on the bins of the observed data” (Ho et al., 22 Sep 2025).

The cosmological parameterization most directly associated with XQ100 is the primordial power spectrum written as

$\alpha$9

where $P_{1\mathrm{D}}$0 is the amplitude, $P_{1\mathrm{D}}$1 is the scalar spectral index, and the pivot $P_{1\mathrm{D}}$2 is chosen to align with Lyman-$P_{1\mathrm{D}}$3 sensitivity.

The mean transmitted flux is modeled through an effective optical depth relation,

$P_{1\mathrm{D}}$4

with redshift evolution expressed as a variation around Kim et al. (2007):

$P_{1\mathrm{D}}$5

$P_{1\mathrm{D}}$6

The paper explicitly notes that $P_{1\mathrm{D}}$7 and $P_{1\mathrm{D}}$8 here are mean-flux nuisance parameters, not the gas temperature $P_{1\mathrm{D}}$9 used elsewhere in parts of the IGM literature.

The covariance model is

$\beta$0

where $\beta$1 is the observational covariance matrix from XQ100 and $\beta$2 represents added sample-variance uncertainty from finite simulation volume and emulator effects. The log-likelihood is constructed from a $\beta$3 of the form

$\beta$4

with $\beta$5, $\beta$6, and $\beta$7.

The interpretation of XQ100 is entirely built on the PRIYA emulator suite. PRIYA consists of MP-Gadget hydrodynamical simulations in a $\beta$8 box, including 60 low-fidelity simulations at $\beta$9 and 3 high-fidelity simulations at $\alpha$0. The emulator is extended to

$\alpha$1

specifically to cover XQ100 and KODIAQ-SQUAD. Reported interpolation errors are $\alpha$2 median for $\alpha$3, rising to $\alpha$4 at $\alpha$5; the resolution-convergence error is $\alpha$6 for $\alpha$7 and $\alpha$8 for $\alpha$9. Since the XQ100 statistical errors are around $\alpha$0, the emulator is described as sufficiently accurate for this dataset.

4. Parameter space and nuisance modeling

The emulator varies 11 parameters spanning cosmology, thermal history, reionization, AGN feedback, and mean flux. The XQ100-relevant cosmological parameters are $\alpha$1, $\alpha$2, $\alpha$3, and $\alpha$4. The He II and H I reionization sector is represented by $\alpha$5, $\alpha$6, $\alpha$7, and $\alpha$8. The remaining astrophysical parameters include $\alpha$9, together with the mean-flux nuisance parameters $z = 3.4 - 4.2,$0 and $z = 3.4 - 4.2,$1 (Ho et al., 22 Sep 2025).

Posterior sampling is performed with Cobaya MCMC. For XQ100, the analysis adopts uniform priors on most emulator parameters within the ranges of Table 2, a Gaussian prior

$z = 3.4 - 4.2,$2

and a Gaussian prior

$z = 3.4 - 4.2,$3

A technically important component of the XQ100 likelihood is marginalization over residual HCD contamination. Four nuisance amplitudes are fitted:

$z = 3.4 - 4.2,$4

The template is taken from Rogers et al. (2018), based on Illustris absorber populations. The study emphasizes that PRIYA itself already contains an HCD population and that simulated spectra are masked for DLAs with $z = 3.4 - 4.2,$5. The nuisance parameters therefore quantify residual or additional contamination in the data relative to PRIYA rather than the total absorber abundance. This distinction is central to the interpretation of XQ100, because the preferred HCD amplitudes are all consistent with zero.

5. Cosmological and thermal constraints from XQ100 alone

The main XQ100-only cosmological constraints are reported at the pivot scale $z = 3.4 - 4.2,$6. The scalar spectral index is

$z = 3.4 - 4.2,$7

and the primordial amplitude satisfies

$z = 3.4 - 4.2,$8

with the latter quoted as a 95% upper limit rather than a two-sided constraint (Ho et al., 22 Sep 2025).

For the mean-flux nuisance sector, the study reports

$z = 3.4 - 4.2,$9

For He II reionization, XQ100 prefers an extended history with

$k = 0.003 - 0.064~{\rm s/km},$0

while $k = 0.003 - 0.064~{\rm s/km},$1 is described as only weakly constrained by XQ100 alone because $k = 0.003 - 0.064~{\rm s/km},$2 lies below the XQ100 redshift range. The H I reionization midpoint is quoted as

$k = 0.003 - 0.064~{\rm s/km},$3

but is explicitly characterized as poorly constrained and essentially prior-dominated.

The HCD nuisance limits are

$k = 0.003 - 0.064~{\rm s/km},$4

These bounds are interpreted as evidence that XQ100 is consistent with no significant excess HCD contamination beyond what is already present in PRIYA.

A particularly important result is that XQ100 alone constrains the thermal history without external IGM temperature data. The analysis argues that the small-scale modes, especially

$k = 0.003 - 0.064~{\rm s/km},$5

and more specifically

$k = 0.003 - 0.064~{\rm s/km},$6

carry the strongest sensitivity to He II heating and thermal smoothing. The inferred mean IGM temperature from XQ100 alone is reported to be consistent with the eBOSS+$k = 0.003 - 0.064~{\rm s/km},$7 baseline at the 68% level. This suggests that the small-scale $k = 0.003 - 0.064~{\rm s/km},$8 by itself encodes substantial information about the thermal history.

6. Relation to eBOSS, KODIAQ-SQUAD, and future joint analyses

In the PRIYA study, XQ100 is evaluated against two comparators: the large-volume eBOSS DR14 forest sample and the high-resolution KODIAQ-SQUAD compilation. Relative to eBOSS, XQ100 yields broader cosmological constraints but covers much smaller scales. The eBOSS+external-$k = 0.003 - 0.064~{\rm s/km},$9 baseline gives

$0.003-0.065~{\rm s/km}$0

so XQ100 is fully consistent but less constraining in the cosmological sector (Ho et al., 22 Sep 2025). Its distinctive value lies instead in the thermal nuisance sector, where it provides information not available from the eBOSS baseline alone.

The contrast with KODIAQ-SQUAD is sharper. XQ100 gives $0.003-0.065~{\rm s/km}$1 and $0.003-0.065~{\rm s/km}$2 consistent with eBOSS and Planck, and its HCD nuisance amplitudes are consistent with zero. KODIAQ-SQUAD, by contrast, favors a significantly higher $0.003-0.065~{\rm s/km}$3 and a strong excess $0.003-0.065~{\rm s/km}$4, which the paper interprets as evidence of selection bias toward high-column density absorbers and overdense regions. When eBOSS priors are imposed on $0.003-0.065~{\rm s/km}$5, KODIAQ-SQUAD shifts to even higher $0.003-0.065~{\rm s/km}$6 while the goodness of fit changes only marginally, from

$0.003-0.065~{\rm s/km}$7

That behavior is presented as evidence that the KODIAQ-SQUAD anomaly reflects an $0.003-0.065~{\rm s/km}$8-$0.003-0.065~{\rm s/km}$9 degeneracy coupled to sample bias rather than robust cosmological information.

No analogous behavior is found for XQ100. The study accordingly treats XQ100 as the more robust high-resolution dataset for small-scale Lyman-$R \sim 4000$00 forest inference. The reasons given are its blind survey design, HCD nuisance amplitudes consistent with zero, cosmological consistency with eBOSS and Planck, thermal history consistent with the eBOSS+external-temperature baseline, and the absence of extreme redshift-dependent LLS contamination.

For future analyses, XQ100 is described as complementary to eBOSS and DESI because cosmological and thermal information are “largely scale-separated”: larger scales constrain the primordial amplitude and tilt, while high-$R \sim 4000$01 modes constrain thermal broadening, pressure smoothing, and He II heating. A plausible implication is that XQ100’s enduring importance will be methodological rather than merely archival. In joint fits, it supplies the small-scale thermal information needed to self-consistently fit thermal nuisance parameters, thereby reducing reliance on external temperature data and strengthening the cosmological interpretation of larger-survey $R \sim 4000$02 measurements (Ho et al., 22 Sep 2025).