---
title: 'Jiutian-300: High-Res Cosmological Simulation'
url: https://www.emergentmind.com/topics/jiutian-300
type: topic
---

# Jiutian-300: High-Res Cosmological Simulation

Searching arXiv for Jiutian-300 and closely related Jiutian simulation papers.
Jiutian-300 is a large cosmological $N$-body simulation within the broader Jiutian simulation suite developed for the China Space Survey Telescope (CSST) extragalactic surveys. In the primary-run taxonomy of the suite, it is the $300\,h^{-1}\mathrm{Mpc}$, $6144^3$-particle realization under the fiducial Planck 2018 $\Lambda$CDM cosmology, designed to combine high mass resolution with sufficient volume for halo, subhalo, galaxy-formation, and reionization studies [2503.21368]. Across recent work, Jiutian-300 functions both as a standalone dark-matter backbone for structure-formation analyses and as an input to downstream pipelines including merger-tree construction, semi-analytic galaxy modeling, mock light-cone generation, and post-processed radiative-transfer calculations for Epoch of Reionization observables [2511.03281][2603.09805].

## 1. Position within the Jiutian simulation suite

The Jiutian simulations are organized as a hybrid suite with four complementary modules: primary runs, emulator runs, reconstruction runs, and extension runs [2503.21368]. Jiutian-300 belongs to the primary runs, whose role is to provide high-resolution, large-volume $N$-body predictions under the fiducial concordance cosmology and to serve as the basis for halo/subhalo catalogs, merger trees, and mock galaxy catalogs.

Within the primary runs, three box sizes are specified: Jiutian-300, Jiutian-1G, and Jiutian-2G, each with $6144^3$ particles [2503.21368]. Jiutian-300 is the smallest-volume and highest-mass-resolution member of this trio, with mass resolution up to $1.0 \times 10^7\,h^{-1} M_\odot$ [2503.21368]. This places it in a distinct niche relative to Jiutian-1G and Jiutian-2G: it sacrifices survey-scale volume in exchange for improved access to low-mass halo and subhalo structure.

A plausible implication is that Jiutian-300 is the preferred primary run when the scientific priority is substructure fidelity or high-redshift source modeling rather than maximal cosmological volume. That inference is consistent with its use in convergence studies of subhalo populations [2508.07678] and in reionization calculations that require resolved low-mass halos [2603.09805].

## 2. Numerical specification and cosmological setup

Jiutian-300 is described as a dark matter-only cosmological $N$-body simulation with box size $L = 300\,h^{-1}\mathrm{Mpc}$ and particle number $N_p = 6144^3 = 231{,}928{,}233{,}984$ [2508.07678]. Its particle mass is given as $1.005 \times 10^7\,h^{-1}M_\odot$ [2508.07678], consistent with the suite-level statement that the primary runs reach mass resolution up to $1.0 \times 10^7\,h^{-1} M_\odot$ [2503.21368].

The simulation adopts Planck 2018 cosmological parameters,
$$
\Omega_{\Lambda} = 0.6889,\quad
\Omega_{m} = 0.3111,\quad
\Omega_{b} = 0.049,\quad
h = 0.6766,\quad
\sigma_{8} = 0.8102,\quad
n_{s} = 0.9665,
$$
as reported in reionization and suite-overview work [2603.09805][2503.21368]. It starts at $z=127$, ends at $z=0$, and has 128 snapshot outputs; in the reionization application, 39 snapshots with $z \geq 6$ are used [2603.09805]. The primary runs are also described as having high time-resolution snapshots and on-the-fly particle lightcones [2503.21368].

Two different code attributions appear in the supplied literature. The reionization study specifies **Gadget-4** as the simulation code for Jiutian-300 [2603.09805], whereas the subhalo-convergence study describes Jiutian-300 as run with **GADGET-3/4** [2508.07678]. Since both formulations are present in the source material, the safest characterization is that Jiutian-300 is a GADGET-family dark-matter-only simulation.

## 3. Halo finding, subhalo tracking, and merger trees

A central feature of the Jiutian primary runs is the use of two independent subhalo and merger-tree pipelines: **SubFind+LHaloTrees** and **HBT+** [2503.21368]. For Jiutian-300 specifically, the reionization study reports halo identification with the Friend-of-Friend (FoF) algorithm and subhalo identification with **SUBFIND** [2603.09805], while the subhalo-distribution analysis uses **HBT+** catalogs [2508.07678]. This dual-pipeline strategy is an explicit design feature of the broader suite [2503.21368].

In the reionization application, the minimum halo mass resolved is $2 \times 10^{8}\,h^{-1}\,{\rm M}_\odot$, corresponding to 20 particles [2603.09805]. In the HBT+-based analyses across the Jiutian suite, subhalos with more than 20 dark matter particles are treated as resolved objects, and the time-domain tracking of HBT+ is emphasized for its handling of subhalo hierarchy, mergers, and persistent identity [2511.03281][2503.21368]. The mock-catalog work based on Jiutian-1G further notes that HBT+ resolves ambiguities such as central/satellite flip-flopping and fragmentations and tracks orphan galaxies when subhalos are disrupted [2511.03281]. This suggests analogous methodological advantages whenever HBT+ products from Jiutian-300 are employed.

The suite overview highlights one specific scientific result enabled by these tree constructions: the subhalo peak mass functions of different levels are approximately universal [2503.21368]. For the level-1 subhalo peak mass function,
$$
g_1(\mu) \equiv \frac{dN}{d\ln\mu} = (a_1\mu^{\alpha_1} + a_2\mu^{\alpha_2}) e^{-c\mu^d},
$$
with $\mu = m_{\rm peak}/M_{\rm host}$, and higher-level peak mass functions can be built through self-convolutions,
$$
g_\ell(\mu) = \int^{\infty}_0 g_{\ell-1}(\mu') g_1(\beta\mu/\mu') d\ln\mu' .
$$
The paper states that this universality extends down to subhalo-to-host mass ratios of approximately $10^{-5}$ [2503.21368].

## 4. Role in galaxy-population and mock-catalog construction

Jiutian-300 is part of a simulation environment whose high-level products include mock galaxy catalogs, lensing maps and catalogs, mock images, and emission-line galaxy catalogs [2503.21368]. The suite overview states that, on top of the primary runs, four sets of mock galaxy light-cone catalogs are produced from semi-analytical models and subhalo abundance matching, with observational properties including galaxy SED, emission lines, lensing distortions, and mock images [2503.21368].

Although the detailed mock-catalog construction in the supplied material is centered on Jiutian-1G rather than Jiutian-300, the relevant methodology clarifies how Jiutian primary runs are generally used. Merger trees extracted from the simulations are coupled to the **GAEA** semi-analytical model of galaxy formation, and spectral energy distributions are generated with **StarDuster**, a neural-network-based stellar population synthesizer trained on radiative-transfer simulations [2511.03281]. Galaxy light-cones up to $z=5$ are then generated with **BLiC**, which interpolates galaxy properties over time using an optimized interpolation scheme [2511.03281].

In this pipeline, the comoving-distance condition for placing galaxies on the light cone is
$$
D_{\mathrm{comoving}}(z)\equiv \int_{0}^{z}\frac{c\, dz'}{H_0 \sqrt{\Omega_{m}(1+z')^3 + \Omega_{\Lambda}}} = \left \| \vec{X}(z) \right \| .
$$
For the SED modeling, StarDuster uses FSPS with a Chabrier (2003) IMF, includes geometry-aware dust attenuation, and adopts
$$
M_{\rm dust} = 0.33\, M_Z
$$
for the dust mass based on cold-gas metal mass [2511.03281].

The supplied sources do not state that Jiutian-300 itself is the run used for the CSST extragalactic light-cone product described in [2511.03281]; that paper instead identifies Jiutian-1G as the flagship run for balancing resolution and volume. However, because Jiutian-300 is one of the primary runs that supply halo, subhalo, and merger-tree products [2503.21368], a plausible implication is that it serves as a higher-resolution complement for problems where lower-mass structure is especially important.

## 5. Jiutian-300 in reionization and 21-cm signal modeling

A major scientific use of Jiutian-300 is in modeling the Epoch of Reionization (EoR) and the associated 21-cm signal [2603.09805]. In that workflow, Jiutian-300 supplies the underlying density field and halo catalogs, which are coupled to the semi-analytic model **L-Galaxies 2020** to generate galaxy catalogs with star-formation histories, stellar masses, metallicities, and related properties [2603.09805].

The resulting galaxy catalog is then post-processed with the one-dimensional radiative-transfer code **Grizzly**, which models the ionization and heating of the intergalactic medium and computes the 21-cm observables [2603.09805]. The dark-matter density field from Jiutian-300 is mapped onto a $256^3$ grid with cell width $1.17\,h^{-1}\,{\rm cMpc}$, and the dark-matter density is used as a proxy for the IGM gas density in each cell [2603.09805].

The differential 21-cm brightness temperature is given in the study as
$$
T_{\rm 21cm} = 27\,\mathrm{mK} \frac{\Omega_{\rm b}h^2}{0.023} \left(\frac{0.14}{\Omega_{\rm m}h^2} \frac{1+z}{10}\right)^{0.5} (1+\delta_{\rm m})\,x_{\rm HI}\left(1-\frac{T_{\rm CMB}}{T_{\rm S}}\right),
$$
where $\delta_{\rm m}$ comes from the Jiutian-300 density field, $x_{\rm HI}$ from Grizzly, and $T_{\rm S}$ is assumed equal to the kinetic temperature $T_{\rm k}$ for $z<12$ [2603.09805]. The 21-cm power spectrum is
$$
\Delta^2_{\rm 21cm}(k) = \frac{k^3}{2\pi^2} P_{21}(k).
$$

The reionization study reports that ionized regions produced by galaxies with star-formation histories derived from L-Galaxies 2020 are slightly larger and warmer than those obtained with a constant star-formation rate, and that, for a fixed stellar mass, galaxies produce smaller ionized regions with increasing stellar-mass-weighted stellar age $\tau_{\rm age}$ [2603.09805]. The paper concludes that different models of galactic star-formation history affect the gas heating and ionizing processes during the EoR and consequently the 21-cm global signal and power spectrum [2603.09805]. In this context, Jiutian-300 is not merely a background density realization; it is the spatial framework that fixes the topology of recombination, bubble growth, and heating.

## 6. Numerical convergence, orphan modeling, and limits of interpretation

Jiutian-300 is also used as a high-resolution benchmark for subhalo abundance and phase-space studies [2508.07678]. In that analysis, Jiutian-300 is compared with Jiutian-1G, which has the same particle count but lower mass resolution, in order to determine which subhalo statistics are numerically converged and how far orphan-based corrections can extend their reliability.

The main result is that the surviving subhalo peak mass function converges only for subhalos with peak mass $m_{\mathrm{peak}}$ above 5000 particles [2508.07678]. For Jiutian-300, this corresponds to $m_{\mathrm{peak}} \gtrsim 5\times10^{10}\,h^{-1}M_\odot$ [2508.07678]. Below that threshold, numerical disruption becomes significant. The study further reports that including orphan subhalos, with survival determined using the merger-timescale model of Jiang et al., accurately recovers the peak-mass function and outperforms other tested models [2508.07678].

The merger-timescale model is written as
$$
\frac{T_{\rm merger}}{T_{\rm dyn}} = A \frac{f(\epsilon)}{\ln\Lambda} \left(\frac{r_c(E)}{r_{\rm host}}\right)^{\gamma_t} \left(\frac{M_{\rm host}}{M_{\rm sat}}\right)^{\beta_t},
$$
and the empirically fitted surviving subhalo peak mass function takes a double-Schechter form,
$$
\frac{dN}{d\ln\mu} = \left(a \mu^{-\alpha} + b \mu^{-\beta}\right) \exp\left(-c \mu^{\gamma}\right),
$$
with $\mu = m_\mathrm{peak}/M_\mathrm{vir}$ [2508.07678].

Including orphans allows recovery of the real-space spatial and velocity distributions to $5$--$10\%$ accuracy down to scales of $0.1$--$0.2\,h^{-1}\mathrm{Mpc}$ [2508.07678]. However, convergence below $0.1\,h^{-1}\mathrm{Mpc}$ remains difficult, and the paper attributes residual discrepancies partly to cosmic variance and finite-box effects in the smaller Jiutian-300 simulation [2508.07678]. The study also emphasizes that redshift-space multipoles are more difficult to recover than real-space statistics because poorly resolved close pairs in real space contaminate much larger scales through Fingers-of-God distortions [2508.07678]. It therefore recommends modified or alternative redshift-space measures that reduce sensitivity to small projected separations.

This evidence is important for interpreting Jiutian-300 outputs. The simulation is high resolution by the standards of large cosmological boxes, but the literature explicitly cautions against uncritical use of its inner-halo subhalo phase space and small-scale redshift-space clustering [2508.07678].

## 7. Scientific significance, accessibility, and nomenclature

Within the Jiutian program, Jiutian-300 occupies the intersection of three methodological priorities: high-resolution dark-matter dynamics, compatibility with multiple subhalo/tree pipelines, and downstream interoperability with semi-analytic and radiative-transfer frameworks [2503.21368][2603.09805]. Its scientific use cases therefore span galaxy-halo connection modeling, subhalo statistics, mock-data generation, and EoR signal prediction.

The suite as a whole provides low-level products such as particle snapshots, particle lightcones, halo/subhalo catalogs, and merger trees, as well as high-level products including multiple mock galaxy catalogs, lensing products, mock images, and constrained-realization datasets [2503.21368]. The total data volume is reported as approximately $8$ PB, and public access is organized through the Jiutian collaboration website, with gradual release of high-level data and contact-based access for full or low-level data [2503.21368].

A possible source of confusion is the term “JiuTian,” which also appears in unrelated work on the **JiuTian Intelligent Network Simulation Platform** for wireless communications [2310.06858] and **JiuTian Chuanliu**, a large spatiotemporal urban-sensing model [2510.23662]. Those systems share the name but are distinct from Jiutian-300. In the cosmological literature, Jiutian-300 specifically denotes the $300\,h^{-1}\mathrm{Mpc}$ primary-run $N$-body simulation within the Jiutian simulations for CSST extragalactic surveys [2503.21368].

Taken together, the cited works portray Jiutian-300 as a high-resolution cosmological backbone rather than a single-purpose simulation. Its role is defined as much by the derivative pipelines built on top of it as by its raw particle realization: halo and subhalo catalogs, merger trees, semi-analytic galaxies, mock light cones, and post-processed reionization fields all depend on the same underlying density evolution [2503.21368][2603.09805]. This suggests that Jiutian-300 is best understood as infrastructure for a family of precision cosmology and galaxy-evolution workflows, with well-documented strengths in resolution and equally explicit caveats in volume-limited and inner-halo regimes [2508.07678].

Source: https://www.emergentmind.com/topics/jiutian-300