Dialogue Concerning the Two Shock Codes Yifei Jin (金刈非), John Raymond Abstract In this paper, we summarize the shock physics and the treatment of radiative transfer in two well-established shock codes – the MAPPINGS code ( Dopita 1976 ; Binette et al. 1985 ; Sutherland & Dopita 1993 ) and the Cox/Raymond code (hereafter CR code ) ( Cox 1972 ; Raymond 1976 ; Raymond 1979 ) . We compare the ionization states, temperatures, electron densities, and the energy transportation of the shock models with shock velocities of 50, 110, 150 and 300 km ​ s − 1 \rm km~s^{-1} . In summary, both codes adopt the Rankine-Hugoniot flow equation to describe the shock flows, giving the same shock physical properties at the immediate area behind shock fronts. The different treatments of radiative transfer in these two codes leads to somewhat different computation of the ionization and thermal structures of shocks, as well as the emission-line fluxes. This work highlights the importance of the delicate treatment of photoionization in shock models, providing insight of the future development of shock codes, such as the 3D shock codes. I Introduction Detailed models of the emission spectra of radiative shock waves date back to the work of Cox 1972 . As more powerful computers became available, models by Dopita 1976 , Shull & McKee 1979 and Raymond 1979 added more elements, more atomic rates and more detailed radiative transfer. Over the years, the code of Dopita 1976 evolved into the MAPPINGS code, and the Raymond 1979 code was modified to deal with high densities and other special cases. A model for shocks in supernova (SN) ejecta was developed by Itoh 1981 , and subsequent authors added more capabilities ( Borkowski & Shull 1990 ; Sutherland & Dopita 1995 ; Blair et al. 2000 ; Koo et al. 2013 ) . As with all models, there are some approximations. Most of the models assume that the thermal conduction term in the energy equation can be ignored, because the transverse B field will be compressed in the shock so that the component of the temperature gradient along the B field will be small. If that is not the case, thermal conduction can be significant ( Borkowski et al. 1989 ) . Most models assume that the elemental abundances are constant throughout the postshock cooling region, but the gradual liberation of refractory elements such as C, Si, Ca and Fe as grains are sputtered affects both the flow structure and the emission spectrum ( Slavin & Raymond 2025 ) . Most tabulated models assume equal electron and ion temperatures, though most models are capable of dealing with unequal electron and ion temperatures with energy transfer between the species by Coulomb collisions, as in Borkowski et al. 1989 and models of shocks in SN ejecta ( Itoh 1981 ; Blair et al. 2000 ) . The models we discuss here assume a 1D, steady flow, but thermal instabilities affect shocks faster than about 150 \rm km~s^{-1} ( Binette et al. 1985 ; Innes 1992 ; Sutherland et al. 2003 ) . The models of radiative shocks also ignore the pressure of cosmic rays and their contribution to the ionization rate. This is probably a good assumption for radiative shocks because the particle acceleration efficiency of the relatively slow shocks that are typically radiative is small. However, reacceleration and compression of energetic particles in the postshock flow might make this component important. The applications of shock models are broad. The theoretical emission-line spectra created from the shock codes provide a reference to interpret the observational spectra by identifying shocks ( Yan 2018 ; Flury et al. 2024 ; O’Sullivan et al. 2025 ) , calibrating the chemical abundance ( Dopita et al. 2016 ; Dopita et al. 2019 ; D’Eugenio et al. 2025 ) , and measuring the shock speeds and densities ( Long et al. 2018 ; Caldwell et al. 2025 ) . The theoretical shock emission-line spectra libraries are created by using different parameters for many models. Dopita & Sutherland 1995 run the original MAPPINGS code to explore the spectral signatures of the low-density fast radiative shocks. This grid is later developed by Allen et al. 2008 and is widely applied to Herbig-Haro objects (HH objects), supernova remnants, AGN jets and outflows, and galactic merger systems, because of its large coverage of the range of shock velocities, preshock densities, chemical abundance and magnetic field strengths. Hartigan et al. 1987 presented a grid of models from the CR code code that has been widely used for interpreting HH object and SNR spectra. The models can also be used to interpret absorption line spectra ( Raymond et al. 1991 ) . Unfortunately, validation of shock models is difficult despite the broad application of shock models. Comparison of different shock codes is one approach to validation. The philosophy is to use different codes to run a series of models (benchmarks) with identical set-ups to test different aspects of the numerical codes. Ferland 1995 gathered the most well-known photoionization and shock codes developers to undertake this benchmark testing 30 years ago, and concluded that the scatter of predicted shock emission-lines is large among the shock codes because of the range of the collisional rates and radiative transfer implemented in the shock codes. The advent of the James Webb Telescope and the large spatially-resolved database of the Local Volume Mapper project in the fifth Sloan Digital Sky Survey also open the new window to undertake innovative shock physics by providing a huge amount of emission-line spectra. Also, the shock codes have evolved in the past 30 years by improving the calculation of the preionization in preshock gas and updating the atomic data. It is therefore time to create a new comparison of shock codes to align the fundament physics in different codes. This is also the preparation for the next-generation shock models dealing with 3D time-dependent dynamic flows instead of the 1D steady flow. In this paper, we run a set of benchmarks by using the MAPPINGS V ( Sutherland & Dopita 2017 ) and the CR code ( Raymond et al. 2020 ) . In Sections II to IV , we summarize the flow physics, atomic physics and the radiative transfer in the two codes. We describe the benchmark models in Section V and the comparison of the products from the two codes from Section VI to VIII . We briefly summarize the comparison of the models with the observational data in Section IX and the conclusion is in Section X II Shock Models II.1 Shock Flow Dynamics In both the MAPPINGS and CR code codes, the shock flow is described by the Rankine-Hugoniot flow equations and the entropy-producing solution is selected. Following Cox 1972 , both the steady shock flow and the shock jump are solved in the same manner by deriving the solution to the compression factor, x\equiv\rho_{1}/\rho_{0} , between any two points in the flow. The entire shocked flow can be consistently computed by using a single shock flow solver. The flow maintains the conversation of mass flux: \rho_{0}v_{0}=\rho_{1}v_{1}, (1) where \rho is the gas density and v is the gas velocity. The subscript numbers refer to any two points, f_{0} and f_{1} , in the flow. The compression factor, x , is defined as x\equiv\rho_{1}/\rho_{0}=v_{0}/v_{1}. (2) The frozen-in magnetic field component perpendicular to the flow velocity is constrained by: B_{0}/\rho_{0}=B_{1}/\rho_{1}. (3) There are three pressures, gas pressure P_{gas} , ram pressure P_{ram} , and magnetic pressure P_{B} in shocks: P_{gas}=nkT=\frac{\rho kT}{\mu m}\equiv(\gamma-1)E, (4) P_{ram}=\rho v^{2}, (5) P_{B}=B^{2}/8\pi, (6) where k is the Boltzmann constant, T is the temperature, n is the total gas number density, \mu is the molecular weight per gas particle, m is the atomic mass unit, and E is the internal specific energy. The conservation of momentum flux is expressed in terms of the pressures: P_{gas,0}+P_{ram,0}+P_{B,0}=P_{gas,1}+P_{ram,1}+P_{B,1}. (7) Finally, the energy conservation is expressed as: \begin{split}\frac{1}{\rho_{0}}\left(gP_{gas,0}+\frac{1}{2}P_{ram,0}+2P_{B,0}-\langle L\rangle\Delta t\right)\\ =\frac{1}{\rho_{1}}\left(gP_{gas,1}+\frac{1}{2}P_{ram,1}+2P_{B,1}\right),\end{split} (8) where g=\gamma/(\gamma-1) , and \langle L\rangle\Delta t is the energy loss between f_{0} and f_{1} through the energy radiated from shocks. In CR code , the ionization state of plasma is explicitly tracked as an additional energy term of the mean ionization energy per particle ( Cox 1972 ) , while MAPPINGS couples the cooling with the ionization states of plasma without tracking the ionization energy separately. The competition among the gas pressure P_{gas} , shock ram pressure P_{ram} and magnetic pressure P_{B} is the key to determine whether the flow is shocked. As summarized in Sutherland & Dopita 2017 , shocks only exist when the flow is supersonic according to the following criteria. \gamma P_{gas}/P_{ram}<1, (9) 2P_{B}/P_{ram}<1. (10) In the CR code , the ionization energy is kept as a separate term in the energy equation. Therefore, changes in the ionization energy are not included in the cooling rate, while they are included in MAPPINGS . This difference in bookkeeping only affects the cooling rates that are printed out by the codes. The energetics and temperatures are not affected. Neither the CR code nor MAPPINGS includes thermal conduction or the energy and pressure terms corresponding to cosmic rays. II.2 The Precursor Calculation MAPPINGS: For shocks with velocity v_{shock}>40\rm~km~s^{-1} ( Dopita & Sutherland 2017 ) , the EUV photons formed in the cooling zone create a photoionization precursor upstream. The photoionization in the precursor changes from fully ionized gas to partially ionized gas as the shock velocity declines from fast shocks ( v_{shock}> 100 km s -1 ) to slow shocks ( v_{shock}\sim 50 km s -1 ). In the MAPPINGS code before the forth version ( Allen et al. 2008 , such as MAPPINGS III used in) , precursors are assumed to be in equilibrium with constant density, and they are computed separately from the shocked flow. Since MAPPINGS V ( Sutherland & Dopita 2017 ) , the full structure of precursor is computed consistently at all velocities by solving the time-dependent ionization and temperature evolution over the entire preshock region by taking a multizone scheme instead of a single constant density approximation. CR code: In the CR code, the radiation field coming out through the shock front is used to compute the ionization state at the input parameter values of n_{0} and T_{0} at the shock. The ionization state and ionizing radiation field can be used for input for another shock run, and the result iterated. The resulting equilibrum is not entirely self-consistent, in that the flux of ionizing photons is partly balanced by the flow of low ionization state material into the precursor. On the other hand, the assumption that the precursor ionization state is in equilibrium with the radiation field produced by the shock is seldom met. For instance, when the shock first begins to cool effectively and produce strong EUV radiation, the precursor will be underionized. Later, as the shock slows down, the recombination time in the preshock gas, around 3\times 10^{5} years for a density of 1 \rm cm^{-3} , is likely to exceed the dynamical time scale of the SNR, and the precursor will be overionized compared to the ionizing flux. Therefore, the CR code treats the ionization fractions of H and He as free parameters, though they are often set to the equilibrium values to minimize the number of free parameters. Cox & Raymond 1985 gives a simple method to approximate a shock in partially neutral gas as the sum of emission from a thin zone of cooling and ionization just behind the shock plus the emission from a slower shock in fully ionized gas. III Atomic Data MAPPINGS: The atomic data in MAPPINGS has been synchronized with an updated database since its second version ( Sutherland & Dopita 1993 ) . MAPPINGS V is based on the CHIANTI 8 database ( Del Zanna et al. 2015 ) and includes 80,000 cooling and recombination lines, and temperature-dependent spline-based interpolation of collisional transitions. The charge exchange and low temperature dielectronic recombination are also included as described in ( Sutherland & Dopita 1993 ) . CR code: Most of the atomic data were assembled in the 1970’s and 1980’s, with some updates in more recent years. Many important rates have not changed much in the intervening time, for instance collisional excitation rates of strong permitted lines and excitation rates and Einstein A values for the strong forbidden lines. The main systematic difference is that the CR code uses ionization rates from Summers ( Summers 1974 ) . They agree with the rates used by CHIANTI and by MAPPINGS at temperatures below the temperature at which the abundance of an ion peaks, and they overestimate the ionization rates at temperatures significantly above that. Another difference, which only becomes important at density on the order of stellar coronal densities, is that the CR code includes the density dependence of the dielectronic recombination rates based on fits to Summers 1974 . The code includes charge transfer between H and heavier ions, along with low temperature dielectronic recombination. Collisional excitation by protons is included for many fine structure transitions. Detailed treatment of [Fe II] emission was added by Hartigan et al. 2004 and Koo et al. 2016 . IV Radiative Transfer The treatment of radiative transfer, especially the scattering and absorption of hydrogen resonance lines, is different in the two codes. MAPPINGS: In MAPPINGS , the cooling spectrum is tracked from 10 -6 eV to 10 5 eV with the fixed logarithmic bin size. Both continuum and line radiation fields are considered and the absorption is computed by the cross-section of each ion. The transfer of hydrogen and helium resonance lines is assumed to be the linear combination of Case A and Case B based on the local optical depth of Ly \gamma , using the method of Brocklehurst 1971 and Johnson 1972 . This is because the high temperature and large velocity gradients in shock flows make the hydrogen lines neither Case A or Case B. The scattering of the other resonance lines is included by increasing their effective path length. The resonance line absorption coefficient across each slab is increased by a factor of 1/P_{esc} , where P_{esc} is the escape probability formulism given by ( Capriotti 1965 ) . The effects of thermal broadening and the velocity gradient within each slab are included. CR code: The CR code bins the emission spectrum into energy bins, typically on the order of 1 eV, and it computes the absorption by the continuum photoabsorption of the atoms and ions. The exception is atomic hydrogen, for which it assumes Case B. That is, all the higher Lyman lines are absorbed locally, and they scatter enough that they are converted to Ly \alpha , the two-photon continuum, and lines of the Balmer and higher series. In addition, the Lyman continuum is treated in the on-the-spot approximation. That is, photons emitted by recombination to the ground state are absorbed locally, so in effect only recombination to n=2 and above counts toward the recombination rate. Both these approximations are very good in the photoionization/recombination zone below 10,000 K. They are less accurate in the narrow region behind the shock where neutral H is excited and ionized, because the optical depth for both lines and continuum is lower there. An escape probability formalism is available in the CR code for use in high density shocks, where the escape probability effectively reduces the Einstein A values of permitted lines and allows collisional quenching at densities below those that would apply to optically thin gas. This option was used for shocks in O star winds ( Krolik & Raymond 1985 ) . V benchmark models The shock structure is sensitive to the shock velocity. In slow shocks, the ionization state of the preshock gas effectively changes the postshock structures by increasing the ionization states of H, He and He + , and altering the cooling efficiency in the postshock region. In fast shocks, the atoms are ionized to high ionization states, increasing the strength and hardness of the ionization fields radiated from the postshock cooling zones. The role of this radiation field becomes significant as the ionizing photons radiated upstream and downstream form the precursors and the recombination tails, where many of the optical emission lines are created. The intermediate speed shocks are in the transition regime, where the subtle differences between the pre-ionized preshock and the ionized postshock strongly change the shock structures and emission line fluxes. We choose four shock models with velocities of 50 \rm km~s^{-1} , 110 \rm km~s^{-1} , 150 \rm km~s^{-1} , and 300 \rm km~s^{-1} , covering the slow, intermediate speed and fast shocks. We do not include the shocks with v_{shock} > 300 \rm km~s^{-1} because the thermal instability of these ultra-fast shocks breaks the steady shock flow assumption, making the 1D shock calculation invalid. The magnetic field strength, gas density and chemical abundance are the other pivotal parameters that influence shock models. The typical magnetic field strength of ISM is around 10 \mu G in the Milky Way ( Han 2017 ) and is 1-10 \mu G in the nebulae ( Harvey-Smith et al. 2011 ) . We set a magnetic field of 5 \mu G as the representative magnetic field strength of the ISM. The gas density of the models is set to be 5 \rm cm^{-3} to represent the broad range of gas densities in the Milky Way ( Kennicutt & Evans 2012 ) . We only consider the most abundant 12 coolants following the Asplund et al. 2009 solar abundance set. VI Shock Structures The ionization and temperature of the preshock gas are determined from the photoionization equilibrium and ionization balance calculation using the upstream ionization field. The extended precursor region is established if the flux of ionizing photons passing upstream through the shock front exceeds the flux of neutrals approaching the shock. Once the gas passes through the shock front, the ion and electron temperatures are discontinuously enhanced. The postshock gas cools down strongly through the collisional cooling due to the high collisional rates. The population of each ionization state is calculated from the ionization and recombination rate of each ion. In these models, the electron-ion temperatures are locked together and a Maxwellian electron energy distribution is assumed. The shock structure is identified as four major regions/zones as shown in Figure 1 . These regions are also described in Cox 1972 ; Cox & Raymond 1985 ; Dopita & Sutherland 1996 and Sutherland & Dopita 2017 . a) The Immediate Shock Region – This is the region next to the shock front. The gas ionization states are those of the preionized preshocked gas but the temperature discontinuously increases on the plasma scale given by the proton Larmor radius or the ion skin depth. b) The Relaxing Zone – Here the ionization states of gas approach collisional ionization balance and the gas temperature remains hot, though significant cooling can occur is if the preshock gas is partly neutral ( Cox & Raymond 1985 ) . The plasma is optically thin in this zone. If the shock is much younger than the radiative cooling time, this is the only zone present, and it would be a nonradiative shock or Balmer line filament ( Chevalier & Raymond 1978 ) . c) The Cooling Zone – The temperature drops off from 10 5-6 K to 10 4 K in this zone as the majority of the cooling and emission of the internal ionizing radiation field occurs. The upstream radiation field preionizes the precursor, and the downstream radiation field further ionizes the cooler gas, forming the photoinization/recombination zone. In the cooling zone, the ionic populations recombine from their high-ionization states to the intermediate- and low-ionization species. For instance, the [O iii] emissions arise when the \rm O^{+3} population recombines to \rm O^{+2} . d) The Recombination Zone – As the plasma recombines and cools down, the downstream EUV photons re-ionize the gas, approaching photoionization-recombination equilibrium. This region behaves similar to an H ii region ionized by a hard ionization field, producing the majority of the optical emission-lines. For the shock flow with magnetic fields, the recombination tail is supported by the magnetic field pressure, which determines its extent, density, ionization state and temperature. VI.1 The Shock Front We first compare the solution of the shock front of the two codes by investigating the physical properties before and after the jump. In the preshock region, we compare the ionization state of H and He because it indicates the strength of the upstream ionization field, which is a reflection of the internally-generated ionizing photons of shocks. The H and He ionic populations are also the critical initial condition which determines the cooling in postshock regions. In the immediate postshock region, we compare the electron temperature, density and the hydrogen density. These parameters describe the gas properties given by the Rankine-Hugoniot jump conditions, indicating the accuracy of the equation solver implemented in each code. These solutions are not affected by any processes related to atomic physics, like cooling or ionization, because the energy term in equation 8 , \Delta L=0 , across the shock front, therefore, the properties of the immediate postshock regions are purely dominated by the flow dynamics. The ionization states of H and He of the preshock gas are shown in Table 2 . In the 50 \rm km~s^{-1} shock flow, the preshock hydrogen and helium are almost neutral with only around 1 percent ionization. This is because very few ionizing EUV photons are produced in the postshock region. The residual EUV arriving at the shock front is not sufficient to ionize the preshock hydrogen and helium. As the shock velocity increases, the postshock collisional ionization increases, and more ionizing photons are created. Therefore, in MAPPINGS V and CR code , the ionization state of hydrogen increases at the same pace from the neutral state in the 50 \rm km~s^{-1} shock to the full ionization in the 150 \rm km~s^{-1} and 300 \rm km~s^{-1} shocks through the half ionization state in the 110 \rm km~s^{-1} shock. The helium ionization in these two codes follows the same trend that fraction of ionized helium increases with increasing shock velocity. However, we notice that, in the fast shocks ( v_{shock} =150 \rm km~s^{-1} and 300 \rm km~s^{-1} ), the He-ionization field in MAPPINGS V is softer than that in CR code because the He +2 /He +1 ratio in MAPPINGS V is smaller than the ratio given by CR code . Table 3 shows the physical conditions of the immediate postshock gas. Both MAPPINGS V and CR code give similar electron temperatures; the 50 \rm km~s^{-1} shock temperature is around 7.5 \times 10 4 K, the 110 \rm km~s^{-1} shock temperature is 2.5 \times 10 5 K, the 150 \rm km~s^{-1} shock temperature is 3.2 \times 10 5 K, and the 300 \rm km~s^{-1} shock has the hottest temperature of 1.2 \times 10 6 K. The hydrogen gas densities given by the two codes are consistent to within 1.6% accuracy. The electron densities given by the two codes are also consistent within a uncertainty of 7 percent apart from the 50 \rm km~s^{-1} shock. The 13.6 percent of difference in n_{e} in the slow shock may caused by the different treatments of radiative transfer of the upstream ionizing radiation fields. VI.2 The Relaxing Zone In the relaxing zone, the gas approaches collisional equilibrium and the temperature is as hot as the immediate postshock region due to the lack of efficient cooling. In the models presented here, the electrons and the ions share the same temperature, as inferred from observations for shocks in this velocity range ( Ghavamian et al. 2013 ; Raymond et al. 2023 ) . In this region, the hydrogen is highly ionized so the gas is optical thin for Lyman lines. The gas in this region is close to the “Case A” assumption. At the transition region from the relaxing zone to the cooling zones, the hydrogen ionization state and the electron density are further elevated with the increasing collisional ionization due to the rising density of the shock flow. VI.3 The Cooling Zone The shock temperature starts to drop after the shocked gas reaches ionization balance and the collisional cooling becomes efficient. The cooling spectrum radiated from this region going upstream and downstream is the major contributor to the internal ionization field of shock flows. Therefore, the calculation of cooling processes determines the structures and emission-line spectra of the precursor and the recombination zone. Following Sutherland & Dopita 2017 , we define the cooling column, \lambda_{k} = \Delta r_{k}\times n_{H} , to account for the speed of cooling in the shock gas. The \Delta r_{k} is the distance from the shock front to the point at which the shock temperature drops to a specific temperature T_{e} , where k=3,4,5 corresponds to T_{e}=10^{3},10^{4} , and 10^{5} K. Figure 4 shows the changes of \lambda_{k} as a function of shock velocity. As the shock velocity increases, the temperature drops quickly from 10 5 K to 10 3 K. In the fast shock regime of v_{shock} > 110 \rm km~s^{-1} , the collisional ionization dominates the postshock structure until the very end of the shock flow. Therefore, the cooling length principally reveals the cooling time multiplied by 1/4 the shock velocity, where the cooling time is the thermal energy divided by n_{e}\times\Lambda(T) . The flatness of \lambda_{k} in Figure 4 at low velocity is because there are very few free electrons at 50 \rm km~s^{-1} . At high velocities, the slope is closer to v_{shock} 4 , because there is a v_{shock} 2 from the thermal energy and a v_{shock} from the shock speed and roughly v_{shock} from the lower \Lambda_{cool} at T_{e} above 100,000 K. We notice that the MAPPINGS V shock models cool down faster than the CR code models as shown in Figure 2 . For the 50 \rm km~s^{-1} shock model, the CR code model and the MAPPINGS V model both start cooling down at 10 14 cm from 8 \times 10 4 K to 10 4 K but the CR code model maintains its temperature at 10 4 K longer than the MAPPINGS V model does. For the 110 \rm km~s^{-1} and 150 \rm km~s^{-1} shock models, the CR code model cools down more efficiently than the MAPPINGS V model at 10 13 -10 14 cm, so that the temperature of the CR code shock model is cooler than the MAPPINGS V model between 10 13 cm and 10 16 cm. For the 300 \rm km~s^{-1} shock model, the MAPPINGS V model starts to cool down at 7 \times 10 17 cm, which is earlier than the CR code model does at 10^{18} cm. Therefore, the 300 \rm km~s^{-1} CR code shock has larger cooling zone than the MAPPINGS V shock. VI.4 The Recombination Zone The recombination zone is the end of shock flows where the typical temperature is around 10^{4} K. This zone is heated by the EUV ionizing photons produced in the cooling zone. The magnetic pressure supports the gas and lowers the density, so the thickness of the recombination zone is larger in the faster shocks with stronger magnetic fields. The gas becomes neutral in the recombination zone, contributing the majority of the emission-lines in shock spectra. Both MAPPINGS V and CR code create the recombination zones in 110 \rm km~s^{-1} , 150 \rm km~s^{-1} and 300 \rm km~s^{-1} shocks. However, the CR code models have the hotter and thicker recombination zone than the MAPPINGS V models, especially in the 300 \rm km~s^{-1} shocks. As shown in Figure 4 , the \lambda_{3} of the CR code models are systematically larger than the MAPPINGS V by 0.5 dex over the entire shock velocity range. This is because that different atomic data bases give different ionization and recombination rates, leading to the difference in the populations of the high and low ionization states as well as the follow-up cooling speed. VII Cooling spectra The cooling spectra providing the ionization radiation in postshock regions are composed of the free-free continuum, resonance lines and the two-photon continua produced by the metastable levels of H 0 , He 0 and He + . Figure 5 shows the cooling spectra at the immediate postshock region, the position of \rm T_{e} =1e4 K and the position of \rm T_{e} =1e3 K. As the temperature cools down, the ionizing spectrum softens. As with other shock properties, the ionizing spectrum is a function of shock velocity such that the ionizing radiation field is harder in faster shock flows. The 300 \rm km~s^{-1} shock has more ionizing emission-lines than the slower shock models. VIII Shock spectra The shock emission-line spectra produced by MAPPINGS V and CR code are the crucial theoretical reference for observers to obtain the properties of the ISM and SNRs. As shown in Figure 6 , we investigate the distribution of the emissivities of six diagnostic emission-lines, H \alpha , [O iii] , [N ii] , [S ii] , [O i] and [Ar iii] . Most of the optical emission-lines are radiated from the recombination zone and the end of the cooling zone of the shock flow. Therefore, the calculation of photoionization in the recombination zone affects these optical emission-line fluxes. Part of the [O i] and [O iii] are radiated before the cooling zone starts. The [O i] emissivity close to the shock front is caused by the residual neutral oxygen, O 0 , after the shock front. The [O iii] emissivity in the relaxing zone rises after the recombination of O +3 to O +2 in conjunction with the increasing density. In Table 4 , we compare the integrated shock spectra given by MAPPINGS V and CR code . The H \beta fluxes are similar given by both codes, which indicates that both MAPPINGS V and CR code produce the similar total ionization energy. The MAPPINGS V models give smaller fluxes than the CR code models in the emission-lines which are generated in the recombination zone, like the [S ii] \lambda\lambda 6717,32, [N ii] \lambda 6584, [N ii] \lambda 5677, [O i] \lambda 6300 and [Ar iii] \lambda 7751 . The difference in fluxes of these emission-lines recalls the fact that the MAPPINGS V and CR code codes have different treatments of photoionization, giving different structures in the recombination zone. The recombination tails in the MAPPINGS V models are shorter than the CR code models, leading to the smaller fluxes of the emission-lines generated in this region. The [O iii] \lambda 5007 emission predicted by these two codes have different dependences on the shock velocity. The [O iii] emission turns on from 110 \rm km~s^{-1} shock model. In the CR code models, the [O iii] /H \beta ratio remains relatively flat in the 150 \rm km~s^{-1} and 300 \rm km~s^{-1} shock models. In the MAPPINGS V models, the [O iii] /H \beta ratio declines by a factor of 2 from the 110 \rm km~s^{-1} shock model to the 300 \rm km~s^{-1} shock model. VIII.1 The Dependence on Magnetic Field We create two sets of models with the magnetic strength of 0.1 \mu G and 10 \mu G as the low- and high-magnetic strength comparison models. As shown in Figure 7 , the stronger transverse magnetic pressure reduces the density, and therefore collisional rates as well as the cooling rate, extending the size of the recombination zone. However, the CR code models have a stronger response to the magnetic field than the MAPPINGS V models. The recombination regions extend farther in the CR code models than those in the MAPPINGS V models as the magnetic strength increases. This is probably because the MAPPINGS code has more accurate treatment of the cooling and ionization, such as the refined frequency sampling of the ionizing spectra, than the CR code . In Figure 8 , we compare the ratios of [O iii] /H \beta , [N ii] /H \alpha , [S ii] /H \alpha and [O i] /H \alpha , which are the so-called BPT diagnostic lines ( Baldwin et al. 1981 , Baldwin, Phillips, & Terlevich diagnostic lines) . The [O iii] /H \beta ratios are mainly determined by shock velocities and the influence of the magnetic strength is only a secondary factor. For the other BPT-diagnostic line ratios, the CR code models and the MAPPINGS V models give the same dependence on the magnetic strength, where the line ratios decrease as the magnetic strength increases. For the 300 \rm km~s^{-1} models, the difference of the effect of magnetic field is negligible when the magnetic strength is smaller than 5 \mu G. IX Comparison with Observations Ideally, one could compare these models with observations to determine their accuracy. In practice, there are generally more free parameters than observables. The biggest problem is that there are very few observed radiative shocks whose shock speeds are known independently. An exception is the western region of the Cygnus Loop, where proper motions have been measured from HST observations by Raymond et al. 2020 . The distance to the Cygnus Loop was believed to be 735 \pm 25 pc ( Fesen et al. 2018 ) , but recent work has shown that it could be as far as 800 pc ( Ritchey et al. 2024 ) . even so, the shock speeds derived from the proper motions and distances are good to about 15%. The shock speeds range from about 90 \rm km~s^{-1} to about 160 \rm km~s^{-1} . Raymond et al. 2020 found that the average spectra agreed reasonably well with the CR code models, but with significant discrepancies for the low ionization states. A more complete comparison was carried out by Slavin & Raymond 2025 . That paper combined UV spectra from HST/COS with fluxes in optical emission lines obtained by integrating the fluxes in HST narrow band images over the COS apertures. They examined 6 positions whose shock speeds span the 90 to 160 \rm km~s^{-1} range. The goal of that work was to investigate dust destruction in shocks, so it combined the CR code model with dust destruction calculations from Jones et al. 1996 for a self-consistent computation of the ionization states, cooling and emission line spectra. The models matched the UV spectra quite well, but they predicted far too much emission in the lowest temperature lines, [O I] and [S II]. This is almost certainly because the COS aperture size is similar to the cooling length for these shocks, and the positions were chosen to be bright in [O III]. Therefore, most of the photoionization/recombination zone was missed. The HST observations in general show that the morphology of the [O III] emission is much different than that of the H \alpha or [S II] lines, largely because of the development of turbulence behind the shocks. This calls into question the applicability of 1D steady-flow models to any astrophysical shocks. However, when an observation takes in a large area, it seems that the average spectrum can be meaningfully compared with the shock models simply because both the models and the actual shocks basically just conserve matter, momentum and energy. However, it may be necessary to average the spectra over a range of shock speeds ( Vancura et al. 1992 ) . X Conclusion We compare the shock models predicted by two well-established shock codes, the CR code and MAPPINGS V . The shock flows in both codes are described by the Rankine-Hugoniot flow equation with the solutions proposed by Cox 1972 . The flow equation solvers behave identically in both codes, giving the similar solutions of shock jumps. The treatments of the radiative transfer and the photoionization are different in the two codes, leading to some uncertainty of the shock emission-line spectra. We create a set of four shock models with the shock velocity of 50 \rm km~s^{-1} , 110 \rm km~s^{-1} , 150 \rm km~s^{-1} and 300 \rm km~s^{-1} to cover the cases from slow shocks to fast shocks through intermediate-speed shocks. These models can be used as the benchmarks in future developments of shock codes. Both codes produce the entire shock structures, which are the immediate postshock region, the relaxing zone, the cooling zone and the recombination tail. The shock structures produced by CR code are slightly different from the MAPPINGS V because of the more approximate treatments of the radiative transfer and the photoionization. MAPPINGS V tracks the ionizing radiation field with around 7000 bins, producing accurate ionic population of up-to 30 elements. This refined treatment of photoionization gives more accurate predictions of the temperature of the recombination tails as well as the shock emission-lines. In these benchmark models, we see that the CR code models cool down faster than the MAPPINGS V models in the relaxing zone and the cooling zone. On the other hand, the CR code models have hotter and more extended recombination tails than the MAPPINGS V models. The significance of photoionization in shock models is amplified in predicted shock emission-line spectra. The [O iii] /H \beta ratios given by the two codes show distinguishable dependence on shock velocities. The [O iii] /H \beta ratio is subtly dependent on shock velocity in the CR code models but the MAPPINGS V models show that the [O iii] /H \beta ratio decreases more strongly with shock velocity in the faster shocks. The pronounced difference in line ratios given by the two codes highlights the intrinsic uncertainty of the interpretation of shock emission-line spectra, particularly for the extragalactic studies. The fluctuations of density and magnetic field strength caused by shock-turbulence interaction is ignored in both codes. More realistic density structures will affect the radiation field and the balance between the gas turbulent motion and magnetic pressure. Our current work already demonstrates the importance of the accurate calculation of radiative transfer and photoionization in predicting shock emission-line spectra. We strongly call for the development of the next-generation three-dimensional shock code. YFJ acknowledge the support from Prof. Yong Shi at Westlake University. Figure 1: The electron temperature of a shock model as a function of the product of the distance and gas density. The shock velocity is 110 \rm km~s^{-1} , the preshock gas density is 5 \rm cm^{-3} and the magnetic field is 5 \mu G. Four sub-structural zones: a) the immediate postshock region , b) the relaxing zone , c) the cooling zone and d) the recombination zone are labeled. Table 1: Chemical Abundance adopted from Asplund et al. 2009 Element log(X/H) H 0.00 He -1.07 C -3.57 N -4.17 O -3.31 Ne -4.07 Mg -4.40 Si -4.49 S -4.88 Ar -5.60 Ca -5.66 Fe -4.50 Table 2: Preshock Ionization States 50 \rm km~s^{-1} 110 \rm km~s^{-1} 150 \rm km~s^{-1} 300 \rm km~s^{-1} JR MV JR MV JR MV JR MV H I 0.988 0.989 0.557 0.573 0.010 0.010 0.001 0.002 He I 1.000 0.990 0.580 0.577 0.000 0.102 0.000 0.000 He II 0.000 0.010 0.420 0.418 0.670 0.854 0.000 0.114 Table 3: Physical Properties of the Immediate Postshock Region 50 \rm km~s^{-1} 110 \rm km~s^{-1} 150 \rm km~s^{-1} 300 \rm km~s^{-1} JR MV \Delta JR MV \Delta JR MV \Delta JR MV \Delta T_{e} 7.53e4 7.48e4 0.7 \% 2.43e5 2.51e5 3.2 \% 3.22e5 3.27e5 1.5 \% 1.23e6 1.24e6 0.8 \% n_{H} 17.2 17.4 1.2 \% 19.1 18.8 1.6 \% 19.3 19.5 1.0 \% 19.9 19.8 0.5 \% n_{e} 0.19 0.22 13.6 \% 8.45 8.76 3.5 \% 19.70 20.90 5.7 \% 21.40 20.00 7.0 \% Figure 2: The comparison of the electron temperatures predicted by MAPPINGS V (solid) and CR code (dashed) as a function of the distance to the shock front. In the v_{shock} =300 \rm km~s^{-1} panel, the small window shows the zoom-in cooling and recombination region between 6\times 10^{17}