A comment on “Why is the Galactic disk so cool?”, by Hamilton et al. J. A. Sellwood, James Binney Abstract Frankel et al. 2020 reported that the rate of diffusion in angular momentum by stars in the disk of the Milky Way was about ten times faster than the rate of heating, which places a stringent requirement on the nature of disk star scattering. In a recent posting, Hamilton et al. 2024 integrated orbits of test particles in a Galactic model that included transient spirals, and found that the ratio of the rates heating to radial migration in their calculations was generally larger than reported by Frankel et al. They concluded that the observed slow heating rate poses a significant challenge to dynamical models of the Milky Way and seemed to require revision of our current theories of spiral structure. Here we show that that our earlier models of radial migration we published in 2002, after correction for 3D motion, account naturally for the finding of Frankel et al. , but leave little room for additional heating by mechanisms other than transient spirals. I Introduction In a study of the dnyamical evolution of the Milky Way (MW) disk, Frankel et al. 2020 developed a parameterized model for the distribution in angular momentum ( L_{z} ), radial action ( J_{R} ), age ( \tau ) and metallicity ([Fe/H]) of low- \alpha red clump stars in the APOGEE survey. Their model took account of the complicated selection effects in the survey and restricted attention to stars at Galactic radii >4\; kpc and ages \lesssim 6\; Gyr. They used an axisymmetric model potential to compute the actions and assumed that stars of a given age and metallicity were formed on (near) circular orbits at their initial Galacto-centric radius. Their best fit model implied the rates of change of angular momentum rms \delta J_{\phi}\equiv\sqrt{\langle(L_{z}-L_{z0})^{2}\rangle}\approx 619\;\hbox{kpc\,km\,s}^{-1}(\tau/6\,{\rm Gyr})^{0.5} and of radial action rms \,\delta J_{R}\equiv\sqrt{\langle(J_{R}-J_{R0})^{2}\rangle}\approx 63\,\hbox{kpc\,km\,s}^{-1}(\tau/6\,{\rm Gyr})^{0.6} at 8 kpc. Here L_{z0} and J_{R0} are the values of these quantities they adopted at the time of formation of the star. Thus they found that the rate of heating of MW disk stars is about an order of magnitude slower than the rate of diffusion of angular momentum. This ratio of heating and diffusion rates places a tight upper bound on other heating mechanisms such as scattering by molecular clouds, infalling satellites and halo substructure, etc. . Sellwood & Binney 2002 provided a natural interpretation for the findings of Frankel et al. 2020 by showing that transient spirals caused radial migration by first trapping stars near corotation onto horseshoe orbits as the spiral amplitude grows and later, as the spiral decays, releasing them at new radii with changed L_{z} and very little increase in their random motion. In a recent posting, however, Hamilton et al. 2024 present orbit integrations in a galaxy model having transient spirals, and had difficulty reproducing the small ratio of heating to radial migration reported by Frankel et al. 2020 . They concluded: “reproducing both the observed radial migration and the small ratio of heating to migration is a highly nontrivial requirement, and poses a significant challenge to models of the Milky Way’s dynamical history, (and) theories of spiral structure”. Here we show that the original models of radial migration presented by Sellwood & Binney 2002 have, after correction for 3D motion, little difficulty in reproduing the findings of Frankel et al. 2020 , and that there is no need for drastic revisions to our understanding of spiral behavior to account for them. II Model U of SB02 Section 3 of SB02 illustrated radial migration in a controlled experiment with a razor-thin, half-mass Mestel disc ( Binney & Tremaine 2008 , §2.6.1a) having disturbance forces restricted to m=2 . Spiral structure was excited by imposing a groove in the angular-momentum distribution. Section 4 of SB02 presented a “unconstrained” model that had no initial groove and active sectoral harmonics 0\leq m\leq 4 . Unfortunately, SB02 did not measure the ratio ({\rm rms}\delta J_{R})/({\rm rms}\delta J_{\phi}) that Frankel et al. 2020 estimated from observations and HMT computed from test-particle simulations. Table 1: Numerical parameters for model U used here Grid points in (r,\phi) 230 \times 256 Grid scaling R_{i}=10 grid units Active sectoral harmonics 0\leq m\leq 8 Plummer softening length \epsilon=R_{i}/4 Number of particles 6\times 10^{6} Largest time-step 0.08R_{0}/V_{0} Radial time step zones 5 Since the results files from this 23-year old experiment are no longer available, we had to rerun it. The RH column of SB02’s Table 1 of gives the numerical parameters used for model U, except that it erroneously gives the softening length as 0.1R_{i} . This must be a typographical error, since we now find that simulations employing that value allow a strong bar to form; in order to inhibit the bar, we had to increase the softening length to 0.25R_{i} . With this one change, we were able to obtain evolution closely resembling that reported in SB02. The revised softening length and the other numerical paramaters used here are given in Table 1. Note that we have also, taken advantage of the substantial increase of cpu power over the past 23 years, to double the grid resolution and employ six times the number of particles, neither of which changed the evolution significantly. The new experiment reproduced (i) similarly mild spirals to those shown in Fig. 8 of SB02, (ii) similar changes to radial velocity dispersion and Q shown in Fig. 9, and a power spectrum like that in Fig. 10. While the L_{z} changes are not quite as large as those reported in Fig. 11 of SB02, they are qualitatively the same. Figure 1: The red plus symbols mark the values of rms \delta J_{R} /rms \delta J_{\phi} at times 100(100)1400 in our simulation. Transient spirals cause rms \delta J_{\phi} to inrease over time, but the ratio rms \delta J_{R} /rms \delta J_{\phi} remains remarkably constant at \sim 1/6 . The point labeled “Milky Way” is the value 0.1 from Frankel et al. 2020 , with a generous 30% uncertainty as estimated by HMT. Our time unit is R_{i}/V_{0} and J_{\phi} values are in units of V_{0}R_{i} . Measurements from our rerun, reported here in Fig. 1, yielded the ratio of the action changes at intervals of 100 time units as the amplitudes of the spirals rise, causing rms \delta J_{\phi} to increase monotonically over time. However, the key ratio, plotted as the ordinate, remains very nearly constant. Since we determine the actions in an axisymmetric logarithmic potential, spiral streaming motions may lead to a mild overestimate of J_{R} , perhaps causing the very slight increase in the last point or two. We have run this case many times with differing softening lengths while trying to reproduce the old result, and in other experiments we have also tried eliminating m=2 force terms, and used the coarser grid and fewer particles that were employed by SB02. But in every case we obtain the same value for the ratio of the action changes, until strong non-axisymmetric features develop. The values we obtain, rms \delta J_{R} /rms \delta J_{\phi}\sim 1/6 , are generally slightly smaller than most of those in Fig. 5(b) of HMT but are larger than the \sim 1/10 value reported by Frankel et al. 2020 . III 3D motion Note that particles are confined to motion to a plane in both our simulations and the orbit integrations by HMT. The MW’s disk, by contrast, has finite thickness, and molecular clouds are efficient at redirecting the peculiar velocities of stars out of the plane; indeed Lacey 1984 showed that that scattering by massive cloud-like particles rearranges the principal axes of the velocity ellipsiod on a timescale that is shorter than that on which they heat the disk. Both Ida et al. 1993 and Sellwood 2008 have predicted the ratio of the vertical to radial dispersions of the disk stars near the Sun should settle to be \sigma_{z}/\sigma_{R}\simeq 0.6 , in good agreement with that observed ( Sharma 2021 , e.g. ) . Thus, some fraction of the non-circular energy created by disk heating must be redirected into vertical motion, causing rms \delta J_{R} to be overestimated when vertical motion is inhibited. Assuming cloud scattering to be the only mechanism that redirects peculiar velocities, values from the Solar neighbourhood imply that the fraction of energy redirected into vertical motion is \sigma_{z}^{2}/\sigma_{\rm tot}^{2}\sim 0.2 so a 3D disk will manifest only 80% of the in-lane heating seen in the equivalent 2D disk. The epicycle approximation implies a linear relation between E_{\rm rand}=\kappa J_{R} , so if we reduce E_{\rm rand} to account for the missing vertical component, we should correspondingly reduce rms \,\delta J_{R} by 20% also. After this reduction, the discrepancy between the observed heating rate in the MW and that expected from radial migration in 2D reported in both our Fig. 1 and in Fig. 5(b) of HMT essentially disappears. IV Conclusions The heating caused by radial migration in our simulations (Fig. 1) is little different from that reported by HMT in their Fig. 5(b), for which they adopted their preferred spiral lifetimes and pitch angles. Both methods appear to find more heating than that reported by Frankel et al. 2020 , and HMT regarded this a serious discrepancy that presented “a major challenge to theories of spiral structure”. However, the motion of particles in both these calculations was confined to a plane, and therefore neglected the \sim 20 % of random energy that is redirected by massive clouds into vertical motion, as we show above in §3. Correction for this effect essentially removes any discrepancy between the observed value for the MW estimated by Frankel et al. , our simulation shown above in Fig. 1, and the calculations reported by HMT. Note that if the observed value of rms \delta J_{R} /rms \delta J_{\phi}\simeq 0.1 is correct and is produced by radial migration, it leaves little room for other possible heating sources, such as cloud scattering, infall of subhalos and small satellites, etc. to have increased, rather than redirected, random motion of disk stars in the Milky Way over the past \sim 6\; Gyr. JAS acknowledges the continuing hospitality and support of Steward Observatory.