Frank H. Shu
Research Subjects
Frank H. Shu has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
Mike J. Cai
Institute of Astronomy and Astrophysics, Academia Sinica P.O. Box 23-141, Taipei 106, Taiwan, R.O.C.
2 shared publications- Anan Lu
Trottier Space Institute and Department of Physics, McGill University , 3600 University Street, Montreal, QC H3A 2T8 , Canada
1 shared publication - Fred C. Adams
University of Michigan
1 shared publication - Fu-Heng Liang
European Southern Observatory (ESO) , Karl-Schwarzschild-Strasse 2, D-85748 Garching
1 shared publication Kyoko Onishi
Chalmers University of Technology, Onsala Observatory Department of Space, Earth and Environment SE-439 92 Onsala
1 shared publication- Lijie Liu
Cosmic Dawn Center (DAWN) , Copenhagen
1 shared publication
Latest Publications
- Open Access
- 14 September 2026
- Published by Elsevier in SSRN Electronic Journal
The Absorption Ratio and the Mean Absolute Correlation
The Absorption Ratio (AR) of Kritzman, Li, Page and Rigobon [1]-the fraction of cross-sectional variance carried by the top-k principal components of the sample correlation matrix-is widely used as a systemic-risk indicator whose spectral construction is presumed to distinguish it from simpler cross-sectional aggregates. Using rolling daily returns of S&P 500 constituents (N ∈ {118, 483}, aspect ratio q = N/T ∈ [0.24, 1.92], July 2019-August 2026), we find corr(AR k=5 , |C ij |) ≥ 0.974 in every window, rising monotonically to 0.9965 as N grows and q shrinks. An elementary Rayleigh-quotient argument identifies the limit: in a universe with a dominant market factor, the leading eigenvector is approximately uniform, so AR k=1 → C ij up to O(1/N) corrections, and the higher-order eigenvectors contribute a subdominant term whose time series co-varies with |C ij |. The equivalence is invariant under Marchenko-Pastur eigenvalue clipping (residuals floor at ∼ 10-2 , entirely from diagonal renormalisation) and is preserved as a level shift under Ledoit-Wolf shrinkage. Where the eigenvector construction genuinely adds information beyond |C ij | is not in the top-k variance sum but in the count k MP of eigenvalues above the MP bulk, which scales with N and is a data-driven regime indicator. We recommend reporting |C ij | alongside any AR, using k = k MP rather than fixed k, and treating k MP itself as the more informative spectral quantity.
- 14 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessThe Absorption Ratio and the Mean Absolute Correlation
The Absorption Ratio (AR) of Kritzman, Li, Page and Rigobon [1]-the fraction of cross-sectional variance carried by the top-k principal components of the sample correlation matrix-is widely used as a systemic-risk indicator whose spectral construction is presumed to distinguish it from simpler cross-sectional aggregates. Using rolling daily returns of S&P 500 constituents (N ∈ {118, 483}, aspect ratio q = N/T ∈ [0.24, 1.92], July 2019-August 2026), we find corr(AR k=5 , |C ij |) ≥ 0.974 in every window, rising monotonically to 0.9965 as N grows and q shrinks. An elementary Rayleigh-quotient argument identifies the limit: in a universe with a dominant market factor, the leading eigenvector is approximately uniform, so AR k=1 → C ij up to O(1/N) corrections, and the higher-order eigenvectors contribute a subdominant term whose time series co-varies with |C ij |. The equivalence is invariant under Marchenko-Pastur eigenvalue clipping (residuals floor at ∼ 10-2 , entirely from diagonal renormalisation) and is preserved as a level shift under Ledoit-Wolf shrinkage. Where the eigenvector construction genuinely adds information beyond |C ij | is not in the top-k variance sum but in the count k MP of eigenvalues above the MP bulk, which scales with N and is a data-driven regime indicator. We recommend reporting |C ij | alongside any AR, using k = k MP rather than fixed k, and treating k MP itself as the more informative spectral quantity.
- Open Access
- 25 July 2025
- Published by Oxford University Press (OUP) in Monthly Notices of the Royal Astronomical Society
- Vol. 541 (4) , 3081-3100
WISDOM Project – XXIV. Giant molecular clouds of the spiral galaxy NGC 5064: high fraction of retrograde rotation
We present high-resolution ($0. 14\,\mathrm{ arcsec}\times 0. 12\,\mathrm{ arcsec}$ or $31\times 25$ pc$^2$) Atacama Large Millimeter/sub-millimeter Array $^{12}$CO(J = 2–1) observations of the spiral galaxy NGC 5064. Our study identifies 478 molecular clouds, of which 387 are resolved both spatially and spectrally. These clouds exhibit similarities to those of the Milky Way in terms of their sizes, molecular gas masses, velocity dispersions, velocity gradients, and Larson relations. However, the NGC 5064 clouds stand out with slightly higher gas mass surface densities, lower virial parameters ($\langle \alpha _{\rm vir}\rangle =0.48\pm 0.03$ assuming a standard conversion factor $X_{\rm CO}=2\times 10^{20}$ cm$^{-2}$ (K km s$^{-1})^{-1}$; $\alpha _{\rm vir} = 1$ for a lower conversion factor of $X_{\rm CO}=0.96\pm 0.02\times 10^{20}$ cm$^{-2}$ (K km s$^{-1})^{-1}$), and an unusually high fraction of retrograde rotation ($f_{\rm retro}\approx 67~{{\ \rm per\ cent}}$). Retrograde clouds are 18 per cent larger, 58 per cent more massive, 15 per cent more turbulent and have 17 per cent larger gas mass surface densities than prograde clouds. The velocity gradients in the clouds seem to arise from turbulence rather than cloud’s intrinsic rotation or large-scale galaxy rotation. Cloud–cloud collisions provide the most plausible explanation for the elevated retrograde fraction, though further investigation is needed to confirm this scenario. Projection effects due to the galaxy’s high inclination ($i = 70{_{.}^{\circ}} 8$) may further enhance the apparent retrograde fraction. Confirmation using less inclined systems is essential to determine whether the observed dominance of retrograde rotation reflects a genuine physical phenomenon or is significantly shaped by projection effects.- 25 July 2025
- Published by Oxford University Press (OUP) in Monthly Notices of the Royal Astronomical Society
- Vol. 541 (4) , 3081-3100
Open AccessWISDOM Project – XXIV. Giant molecular clouds of the spiral galaxy NGC 5064: high fraction of retrograde rotation
We present high-resolution ($0. 14\,\mathrm{ arcsec}\times 0. 12\,\mathrm{ arcsec}$ or $31\times 25$ pc$^2$) Atacama Large Millimeter/sub-millimeter Array $^{12}$CO(J = 2–1) observations of the spiral galaxy NGC 5064. Our study identifies 478 molecular clouds, of which 387 are resolved both spatially and spectrally. These clouds exhibit similarities to those of the Milky Way in terms of their sizes, molecular gas masses, velocity dispersions, velocity gradients, and Larson relations. However, the NGC 5064 clouds stand out with slightly higher gas mass surface densities, lower virial parameters ($\langle \alpha _{\rm vir}\rangle =0.48\pm 0.03$ assuming a standard conversion factor $X_{\rm CO}=2\times 10^{20}$ cm$^{-2}$ (K km s$^{-1})^{-1}$; $\alpha _{\rm vir} = 1$ for a lower conversion factor of $X_{\rm CO}=0.96\pm 0.02\times 10^{20}$ cm$^{-2}$ (K km s$^{-1})^{-1}$), and an unusually high fraction of retrograde rotation ($f_{\rm retro}\approx 67~{{\ \rm per\ cent}}$). Retrograde clouds are 18 per cent larger, 58 per cent more massive, 15 per cent more turbulent and have 17 per cent larger gas mass surface densities than prograde clouds. The velocity gradients in the clouds seem to arise from turbulence rather than cloud’s intrinsic rotation or large-scale galaxy rotation. Cloud–cloud collisions provide the most plausible explanation for the elevated retrograde fraction, though further investigation is needed to confirm this scenario. Projection effects due to the galaxy’s high inclination ($i = 70{_{.}^{\circ}} 8$) may further enhance the apparent retrograde fraction. Confirmation using less inclined systems is essential to determine whether the observed dominance of retrograde rotation reflects a genuine physical phenomenon or is significantly shaped by projection effects. - Open Access
- 10 December 2007
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 671 (1) , 497-517
Ambipolar Diffusion in Molecular Cloud Cores and the Gravomagneto Catastrophe
[[abstract]]This paper reexamines the problem of ambipolar diffusion as a mechanism for the production and runaway evolution of centrally condensed molecular cloud cores, a process that has been termed the gravomagneto catastrophe. Our calculation applies in the geometric limit of a highly flattened core and allows for a semianalytic treatment of the full problem, although physical fixes are required to resolve a poor representation of the central region. A noteworthy feature of the overall formulation is that the solutions for the ambipolar diffusion portion of the evolution for negative times (t < 0) match smoothly onto the collapse solutions for positive times (t > 0). The treatment shows that the resulting cores display nonzero, but submagnetosonic, inward velocities at the end of the diffusion epoch, in agreement with current observations. Another important result is the derivation of an analytic relationship between the dimensionless mass-to-flux ratio lambda(0) equivalent to f(0)(-1) of the central regions produced by runaway core condensation and the dimensionless measure of the rate of ambipolar diffusion is an element of. In conjunction with previous work showing that ambipolar diffusion takes place more quickly in the presence of turbulent fluctuations, i.e., that the effective value of is an element of can be enhanced by turbulence, the resultant theory provides a viable working hypothesis for the formation of isolated molecular cloud cores and their subsequent collapse to form stars and planetary systems.[[fileno]]2010118010009[[department]]物理
- 10 December 2007
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 671 (1) , 497-517
Open AccessAmbipolar Diffusion in Molecular Cloud Cores and the Gravomagneto Catastrophe
[[abstract]]This paper reexamines the problem of ambipolar diffusion as a mechanism for the production and runaway evolution of centrally condensed molecular cloud cores, a process that has been termed the gravomagneto catastrophe. Our calculation applies in the geometric limit of a highly flattened core and allows for a semianalytic treatment of the full problem, although physical fixes are required to resolve a poor representation of the central region. A noteworthy feature of the overall formulation is that the solutions for the ambipolar diffusion portion of the evolution for negative times (t < 0) match smoothly onto the collapse solutions for positive times (t > 0). The treatment shows that the resulting cores display nonzero, but submagnetosonic, inward velocities at the end of the diffusion epoch, in agreement with current observations. Another important result is the derivation of an analytic relationship between the dimensionless mass-to-flux ratio lambda(0) equivalent to f(0)(-1) of the central regions produced by runaway core condensation and the dimensionless measure of the rate of ambipolar diffusion is an element of. In conjunction with previous work showing that ambipolar diffusion takes place more quickly in the presence of turbulent fluctuations, i.e., that the effective value of is an element of can be enhanced by turbulence, the resultant theory provides a viable working hypothesis for the formation of isolated molecular cloud cores and their subsequent collapse to form stars and planetary systems.[[fileno]]2010118010009[[department]]物理
- Open Access
- 20 August 2004
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 611 (2) , 1054-1067
Nonaxisymmetric Neutral Modes in Relativistic Disks
We perform a linear stability analysis of the axisymmetric, relativistic, self-similar isothermal disk against non-axisymmetric perturbations. Two sets of neutral modes are discovered. The first set corresponds to marginally unstable perturbations driven by gravitational radiation, and the other signals the onset of bifurcation to non-axisymmetric equilibrium solutions to the Einstein equations.
- 20 August 2004
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 611 (2) , 1054-1067
Open AccessNonaxisymmetric Neutral Modes in Relativistic Disks
We perform a linear stability analysis of the axisymmetric, relativistic, self-similar isothermal disk against non-axisymmetric perturbations. Two sets of neutral modes are discovered. The first set corresponds to marginally unstable perturbations driven by gravitational radiation, and the other signals the onset of bifurcation to non-axisymmetric equilibrium solutions to the Einstein equations.
- Open Access
- 20 January 2003
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 583 (1) , 391-401
Relativistic Singular Isothermal Toroids
We construct self-similar, axisymmetric, time-independent solutions to Einstein's field equations for an isothermal gas with a flat rotation curve in the equatorial plane. The metric scales as $ds^2 \to a^2 ds^2$ under the transformation $r\to a r$ and $t \to a^{1-n} t$, where $n$ is a dimensionless measure of the strength of the gravitational field. The solution space forms a two-parameter family characterized by the ratios of the isothermal sound speed and the equatorial rotation speed to the speed of light. The isodensity surfaces are toroids, empty of matter along the rotation axis. Unlike the Newtonian case, the velocity field is not constant on a cylindrical radius. As the configuration rotates faster, an ergoregion develops in the form of the exterior of a cone centered about the rotation axis. The sequence of solutions terminates when frame dragging becomes infinite and the ergocone closes onto the axis. The fluid velocity of the last solution has finite value in the midplane but reaches the speed of light on the axis.- 20 January 2003
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 583 (1) , 391-401
Open AccessRelativistic Singular Isothermal Toroids
We construct self-similar, axisymmetric, time-independent solutions to Einstein's field equations for an isothermal gas with a flat rotation curve in the equatorial plane. The metric scales as $ds^2 \to a^2 ds^2$ under the transformation $r\to a r$ and $t \to a^{1-n} t$, where $n$ is a dimensionless measure of the strength of the gravitational field. The solution space forms a two-parameter family characterized by the ratios of the isothermal sound speed and the equatorial rotation speed to the speed of light. The isodensity surfaces are toroids, empty of matter along the rotation axis. Unlike the Newtonian case, the velocity field is not constant on a cylindrical radius. As the configuration rotates faster, an ergoregion develops in the form of the exterior of a cone centered about the rotation axis. The sequence of solutions terminates when frame dragging becomes infinite and the ergocone closes onto the axis. The fluid velocity of the last solution has finite value in the midplane but reaches the speed of light on the axis.