1: \documentclass[12pt,preprint]{aastex}

2: %\documentclass{aastex}

3: 

4: \begin{document}

5: %\title{Core-jet Properties and Evidence for Jet Collimation in the VLBA Imaging and Polarimetry Survey (VIPS)}

6: \title{A Statistical Description of AGN Jet Evolution from the VLBA Imaging and Polarimetry Survey (VIPS)}

7: \author{J. F. Helmboldt\altaffilmark{1}, G. B. Taylor\altaffilmark{2}, R. C. Walker\altaffilmark{3}, and R. D. Blandford\altaffilmark{4}}

8: %\email{helmbold@unm.edu}

9: 

10: \altaffiltext{1}{Naval Research Laboratory, Code 7213, 4555 Overlook Avenue SW, Washington, DC 20375-5351}

11: \altaffiltext{2}{Department of Physics and Astronomy, University of New Mexico, 800 Yale Blvd NE, Albuquerque, NM 87131, USA}

12: \altaffiltext{3}{National Radio Astronomy Observatory, P.O. Box O, Socorro, NM 87801, U.S.A.}

13: \altaffiltext{4}{KIPAC, Stanford University, PO Box 20450, MS 29, Stanford, CA 94309, USA}

14: \received{?}

15: \accepted{?}

16: %\journalid{?}{?}

17: %\articleid{0}{0}

18: 

19: %\slugcomment{Not yet submitted}

20: \begin{abstract}

21: A detailed analysis of the evolution of the properties of core-jet systems within the VLBA Imaging and Polarimetry Survey (VIPS) is presented.  %We find some indication that both fainter core and jet components tend to have higher relative amounts of polarized intensity than brighter components.  However, each of these trends is strongly influenced by selection biases and are therefore not substantially strong.  We have found some statistical evidence that jet components expand and possibly fade as they move away from their cores.  

22: We find a power-law relationship between jet intensity and width that suggests for the typical jet, little if any energy is lost as it moves away from its core.  Using VLA images at 1.5 GHz, we have found evidence that parsec-scale jets tend to be aligned with the the direction of emission on kiloparsec scales.  We also found that this alignment improves as the jets move farther from their cores on projected scales as small as $\sim$50-100 pc.  This suggests that realignment of jets on these projected scales is relatively common.  We typically find a modest amount of bending (a change in jet position angle of $\sim 5^{\circ}$) on these scales, suggesting that this realignment may typically occur relatively gradually.

23: 

24: \end{abstract}

25: 

26: \keywords{galaxies: active - surveys - catalogs - galaxies: jets - galaxies: nuclei - radio continuum: galaxies}

27: 

28: \section{Introduction}

29: While the properties of active galactic nuclei (AGN) have been studied across the entire electromagnetic spectrum, Very Long Baseline Interferometry (VLBI) in the radio regime has provided the only means to study the parsec-scale structures associated with the many AGN.  Chief among these structures are the relativistic jets of material launched by most AGN.  One of the keys to understanding these phenomena is the development of a full picture of their evolution from the smallest ($\sim$10 AU) to the largest (up to $\sim$1 Mpc) scales.  Observations of AGN in the X-ray regime with the Chandra satellite as well as radio images made with interferometers such as the Very Large Array (VLA) have helped define the typical properties of jets and associated lobe emission on large ($\sim$1 kpc) scales.  The High Energy Stereoscopic System (HESS) has provided insights into the gamma-ray emitting regions ($\sim$0.01-1 pc), but these observations are biassed toward one class of AGN, BL Lac objects \citep{aha05}.  Future observations with the Gamma-Ray Large Area Space Telescope \citep[GLAST; ][]{geh99} will provide similar insights, but for a much broader and larger sample of sources.  However, data from both HESS and GLAST provide virtually no spatial information about jets on any size scale, and require radio-regime VLBI images for proper interpretation.  In addition to this, it has been seen that the strongest gamma-ray AGN may be biassed toward radio sources with higher brightness temperatures and larger core fractions \citep{tay07}.  From this, it is clear that radio-frequency VLBI images of AGN are vital to the study of the evolution of the structure and properties of jets on parsec scales.\par

30: While many VLBI monitoring campaigns have been used to examine jets evolving over time on these size scales \citep[see][and references therein]{lis05}, they have typically been focused on modest samples of the brightest compact sources since monitoring of a relatively large sample of sources may be impractical.  This has precluded any significant statistical analysis of tendencies among a variety of core-jet systems regarding their evolution.  However, it is plausible that single epoch data for a relatively large number of sources can be used to statistically discern how the typical jet evolves as each jet within the sample is presumably in a different stage of evolution.  If a large sample with well defined selection criteria is used, the affects of potential biasses on this analysis can be properly constrained, and a reasonably accurate representation of the typical core-jet system can be constructed.  We have used the 5 GHz images obtained by the VLBA Imaging and Polarimetry Survey \citep[VIPS; ][]{hel07} for 1,127 AGN with the NRAO Very Long Baseline Array (VLBA) to demonstrate that this is possible.  In \S 2, we provide a brief description of the VIPS sample.  In \S 3, we outline the selection biasses that affect our proposed analysis and describe how they have been over-come or, at least, minimized.  In addition, we describe in \S 3 the statistical constraints we have place on the evolution of the intensity, width, and direction of the typical jet using our VIPS data.  In \S 4, we summarize the implications of these results for the evolution of the intrinsic properties of the typical jet.

31: %In addition to this, previous single epoch VLBI surveys \citep[e.g.,][]{pea88,tay96,bea02} do not have polarization data for enough core-jet systems for a thorough statistical analysis of the possible relationship of the level and orientation of polarized intensity to jet structure, properties, and evolution.  To overcome these limitations, the VLBA Imaging and Polarimetry Survey \citep[VIPS; ][]{hel07} was conducted, a 5 GHz survey of 1,127 AGN with the NRAO Very Long Baseline Array (VLBA).  The survey consists of total and polarized intensity images at 5 GHz as well as a slew of derived properties such as morphologies, sizes, mean electric vector position angles (EVPAs), etc.  While the general properties of the sample and initial results were presented by \citet{hel07}, this paper will focus on the properties of the sources identified within VIPS as having core and jet components that are resolved from one another, 602 source in all.  In particular, we will discuss how the degree of polarization relates to key jet properties (\S 3.1), the morphology of the jets (i.e., the amount of jet bending and magnitude of the opening angles; \S 3.2), and we will use the relatively large sample size to statistically explore how jet properties evolve (\S 4).

32: 

33: \section{The VIPS Core-jet Sample}

34: The VIPS source list was selected from the Cosmic Lens All-Sky Survey \citep[CLASS; ][]{mey03}, an 8.5 GHz VLA survey of $\sim$12,100 flat spectrum objects ($\alpha > -0.5$ between 4.85 GHz and a lower frequency).  VIPS sources were selected to be all CLASS sources brighter than 85 mJy at 8.5 GHz which are also within the survey area of the Sloan Digital Sky Survey \citep[SDSS; ][]{yor00} and with declinations between 15$^{\circ}$ and 65$^{\circ}$.  For 147 of these sources, data were already available at 5 GHz from the Caltech-Jordell Bank Flat spectrum survey \citep[CJF; ][]{tay96,bri03,pol03}; 20 were observed for the VIPS pilot program at 5 and 15 GHz \citep{tay05}.  There are 8 sources not observed as part of either CJF or the pilot program which have been observed at 15 GHz as part of the Monitoring of Jets in AGN with VLBA Experiments project \citep[MOJAVE; ][]{lis05}.  The remaining 958 were observed with the VLBA at 5 GHz in full polarization from January to August 2006.  The data were processed automatically using a combination of the VLBA calibration pipeline \citep{sjo05} and DIFMAP imaging scripts described in detail in \citet{tay05}.  Data for the 147 sources taken from the CJF survey were reprocessed using the same imaging scripts.  All but $<$1\% of the sources were significantly detected; about 36\% had detected polarized intensity.\par

35: Each source was broken up into components by fitting Gaussian functions to significant ($>6\sigma$) peaks in its 5 GHz image.  These components were used, along with visual inspection, to classify each source as a point-like source, a core-jet system, a compact symmetric object (CSO) candidate, or a complex source.  

36: %To morphologically classify each source, Gaussian components were fit to significant ($>6\sigma$) peaks in the 5 GHz images.  An algorithm was developed to use the locations and properties of these components to classify each source as a point-like source (PS), a short jet (SJET), a long jet (LJET), a compact symmetric object candidate (CSO), or a complex source (CPLX).  These Gaussian components were also used to measure jet lengths and position angles for SJET and LJET sources.  In addition to this, the Gaussian components were used to construct image masks so that the mean fractional polarization (the ratio of polarized to total intensity) and EVPA could be measured separately for each source, its brightest component or "core", and the regions outside the core (i.e., the jet for SJET and LJET sources).  The fractional polarization of jet components ($\sim$20\%) was found to be significantly higher than that of cores ($\sim$5\%).  Evidence was also found of a tendency for core EVPAs and jet position angles to be perpendicular \citep{hel07}.

37: In the analysis presented here, only core-jet systems with more than one Gaussian component are considered, i.e., sources where the core and jet have been resolved into separate components, 602 sources in all.\par

38: In addition to the 5 GHz VLBA data, we also have optical spectra for 559 ($\sim$50\%) of the VIPS sources.  Among the 602 VIPS core-jet systems presented here, 311 have optical spectra.  These were taken from a combination of SDSS spectra and ongoing follow-up observations being conducted with a variety of telescopes \citep[see ][]{hea08}.  The spectra have been used to derive redshifts for these sources which have been used to compute distances assuming ($\Omega_{m}$,$\Omega_{\Lambda}$,$h$)=(0.3,0.7,0.7).  Throughout this paper, all jet lengths and projected separations of jet components from their cores that are presented in physical units (i.e., pc) were computed using this assumed cosmology.

39: 

40: %\section{Jet Properties and Polarization}

41: %\subsection{Basic Jet Properties}

42: %Magnetic fields may play crucial roles in both the launch and subsequent evolution of jets.  It is therefore important to explore how the existence of polarized intensity and the degree of polarization are related to the general properties of jets within core-jet systems.  To this end, we have constructed and displayed in Fig.\ \ref{jprops1} the distributions for key properties of the VIPS core-jet systems with separate histograms for all core-jets, core-jets with detected polarized intensity within the cores, and core-jets with detected polarized intensity within the jets.  The properties for which the distributions have been computed are the jet-to-core ratio, the brightness temperature of the core, and the mean brightness temperature of the brightest jet component.  To determine the brightness temperature of the core of each source, we fit a single circular Gaussian component to the visibility data using DIFMAP with initial guesses for the center and total flux of the component taken from the Gaussian component fit to the core on the 5 GHz image.  For this analysis, we have defined the jet-to-core ratio to be the ratio of the total 5 GHz flux density attributed to jet components with projected separations from their cores of more than 25.4 pc to the total flux density of the core plus any components with separations $<25.4$ pc.  Since we have observed the VIPS sources at rest-frame frequencies of $5(1+z)$ GHz, these ratios were multiplied by a factor of $(1+z)$ to account for the fact that the spectral indexes of jets are typically steeper than those of cores and that the difference between the two is typically near unity.  We have chosen the limit of 25.4 pc because for the adopted cosmology, the angular diameter distance, $D_{A}$, peaks at a redshift of 1.605, or $D_{A}=1747.62$ Mpc.  At this distance, the largest dimension of the VIPS restoring beam, 3 mas, corresponds to 25.4 pc.  We have therefore chosen to define a limit of 25.4 pc beyond which the jet components for all VIPS sources can be resolved from their central cores (e.g., about 10\% of the identified jet components are within 3 mas of their sources' cores).  While this establishes a consistent definition for the projected size of the core, regardless of redshift, it must be noted that the jet-to-core ratio is still affected by viewing angle effects.  In other words, sources with jet-to-core ratios that are low or even zero may be sources with jets at small viewing angles where the flux density for radii $<25.4$ pc contains a larger fraction of the total jet intensity.\par

43: %For all three properties, the shapes of the distributions for all core-jets and those with known polarized cores are very similar.  K-S tests performed between these pairs of distributions were inconclusive.  Systems with confirmed polarized jets, however, seem to have higher fractions of systems with brighter jets and relatively stronger cores, but with essentially the same distribution of core brightness temperatures as all core-jets.  K-S tests put the probabilities that the distributions for jet-to-core ratio and jet brightness temperature for core-jets with confirmed polarized jets were drawn from the same parent distribution as all VIPS core-jets at $4\times10^{-4}$, and $2\times10^{-10}$, respectively.  While a K-S test derived probability of $4\times10^{-4}$ is not a remarkably strong result, the extremely low probabilities for the jet brightness temperatures implies there is a substantial difference between the distributions of these quantities for sources with detected polarized intensity and those of the entire VIPS sample.  However, since it is likely more difficult to detect polarized emission from systems with fainter jets, the fact that core-jets with detected polarized intensity in their jets are biassed toward systems with higher jet brightness temperatures may be heavily influenced by this selection effect.\par

44: %To explore whether the differences in jet-to-core ratio and brightness temperature distributions seen in Fig.\ \ref{jprops1} are real or the result of selection effects, we have plotted the fractional polarization (the mean ratio of the polarized to total intensity) for the core and jet components as functions of these properties in Fig.\ \ref{jprops2}.  Along with the data, we have also included so called "box-and-wisker" plots which show the median, upper and lower quartiles, and extrema of the distributions of fractional polarization within bins of each of the jet properties.  In Fig.\ \ref{jprops2}, the median core fractional polarization increases with jet-to-core ratio.  However, the fact that the upper quartiles within jet-to-core ratio bins are relatively constant indicates that this is likely a selection effect.  Neither the median nor the upper quartile of the jet fractional polarization seem to significantly change with jet-to-core ratio.  This indicates that while the histograms in Fig.\ \ref{jprops1} imply that polarized jets have slightly stronger cores, it appears that those with the highest fractional polarization do not necessarily have the lowest jet-to-core ratios.\par

45: %It can also be seen in Fig.\ \ref{jprops2} that there are no obvious trends between $f_{pol,core}$ and the jet brightness temperature and between $f_{pol,jet}$ and core brightness temperature.  However, $f_{pol,core}$ and $f_{pol,jet}$ do appear to be anti-correlated with core and jet brightness temperature, respectively.  Both of these trends are likely due in part to selection biasses similar to those described above.  However, the fact that in both cases, the minimum and maximum values of $f_{pol}$ both vary the same way with brightness temperature imply that there may also be real tendencies for fainter cores and jets to have larger relative degrees of polarization.  We must note that the observed degree of polarization may be reduced by the finite resolution of the VLBA beam or by the degree of Faraday rotation and that this could potentially affect these trends.  However, we have found no trend between either the jet or core brightness temperature and the physical size of the restoring beam used at the redshifts of the sources.  It is also unlikely that there is a significant trend between the amount of Faraday rotation and either brightness temperature \citep[see, e.g., ][]{zav04}.  This implies that while these effects may reduce the strength of the trends in Fig.\ \ref{jprops2}, they have not artificially caused the trends.\par

46: %It should also be noted that the core and jet components with the largest brightness temperatures may be biassed toward sources with larger Doppler factors and smaller viewing angles.  While the polarized intensity of any component will be boosted by the same Doppler factor as the total intensity, this bias may indirectly cause the observed fractional polarization to be lower for these components.  This is because, as \citet{lai80} showed, if a volume of material with a randomly oriented magnetic field is compressed, the degree of polarization is largest when the material is viewed at an angle perpendicular to the direction of the compression.  If knots of jet emission are compressed by shocks along the direction of the jet motion, those with smaller viewing angles may be observed to have much smaller degrees of polarization.  However, for jets moving with speeds near the speed of light, even if the viewing angle is small, such compressed regions will appear as if viewed from the side due to relativistic effects.  Therefore, since Doppler boosting will likely cause the sample to be biassed toward faster jets, this source of bias is probably relatively weak.  Even if this is not the case and this viewing angle bias does substantially effect the trends between $f_{pol}$ and $T_{B}$ in Fig.\ \ref{jprops2}, it still requires a relatively specific physical mechanism for the source of the detected polarized intensity.  This implies that regardless of the strength of the potential viewing angle bias, these trends provide some clue(s) about the typical physical properties of parsec-scale jets.

47: 

48: %\subsection{Jet Bending on Parsec Scales}

49: %The shapes of jets and how they relate to magnetic fields of core-jet systems are also of interest.  To address this we have derived a quantitative estimate of the amount of jet bending on parsec scales as measured using our 5 GHz images.  We have done this by developing an algorithm, detailed in Appendix A, for measuring the difference between the position angle of jet emission near the core, $PA_{in}$, and the position angle with respect to the core of the outer most jet component, $PA_{out}$, for all sources with two or more jet components, 214 sources in all.  The purpose of using an automated approach rather than a "by-eye" estimate of the amount of bending is so that a consistent, objective, and quantitative measure of the degree of jet bending could be made without being significantly affected by stray or spurious components.  The distribution for $|\mbox{PA}_{in}-\mbox{PA}_{out}|$ determined using this algorithm is displayed in Fig.\ \ref{angs}.  The distribution for sources with more than two jet components, for which the algorithm provides a better estimate of the amount of bending (see Appendix A), is also plotted separately in grey.  The peaks of the distributions appear to be between 0$^{\circ}$ and 5$^{\circ}$, implying that a modest amount of jet bending on parsec scales may be common.  However, there are instances of significant amounts of jet bending with about 20\% of sources having $|\mbox{PA}_{in}-\mbox{PA}_{out}|>20^{\circ}$ and position angle changes of up to about 90$^{\circ}$.  Using this quantitative estimate, we have found no obvious trend between the degree of jet bending and jet-to-core ratio, core or jet brightness temperature, and core or jet fractional polarization.%\par

50: %We have also developed a similarly automated procedure for estimating the opening half-angle for each core-jet system.  We first fit a line to all the components of each core-jet.  We then measure the coordinates of each jet component in a rotated coordinate system with the $x$-axis along the fitted line and with the origin located at the $x$-coordinate of the core and the $y$ value of the fitted line evaluated at the $x$-coordinate of the core.  We then measure the extent of each jet component from its center along the $y$-axis (perpendicular to the jet) using the parameters for its elliptical Gaussian fit.  This extent is then deconvolved using the extent of the restoring beam along the $y$-axis, yielding an estimate of the extent of the component, referred to as $dr$.  Following this, a half-opening angle is computed for each jet component equal to tan$^{-1} [(|y|+dr)/|x|]$ and averaged over all jet components to get a mean half-opening angle, $\Psi$, for each core-jet system.  The $\Psi$ distributions for all core-jets and for core-jets with more than one jet component are displayed in the lower panel of Fig.\ \ref{angs}.  The modes of these distributions are between 12$^{\circ}$ and 18$^{\circ}$.  Both distributions have tails extending to around 60$^{\circ}$, implying that there are some systems with substantial opening half-angles ($\sim$20\% have opening half-angles $>30^{\circ}$, roughly twice the mode of the sample).  Similar to jet bending, we have found no apparent trend between opening angle and jet length, core fraction, core or jet brightness temperature, and core or jet fractional polarization.\par

51: 

52: \section{The Evolution of Jet Properties}

53: While the VIPS data set provides information regarding each source at only one epoch, the relatively large number of sources allows us to explore statistically how the properties of jet components change as they move away from their cores.  To do this, we have computed the projected separation of each jet component from its core as well as its apparent position angle with respect to the core position for all core-jets with known redshifts.  We have also computed the mean brightness temperature for each of these components using the parameters of the Gaussian components fit to the 5 GHz VLBA images.  We note that this allows for a statistical description of how the properties of the typical jet change with projected core separation, but not necessarily with time.  This is because while there is evidence that the jet speed, $\beta$, is relatively constant with core separation for individual jets \citep[see, e.g., ][]{kel07}, the mean value of $\beta$ among our sample of core-jet systems may change.  This caveat will be discussed further in \S 4.%In addition, we have estimated the fractional polarization of each component by computing the value of $f_{pol}$ at the pixel closest to the component's center on the total intensity and polarized intensity images.  In the case where this pixel has a polarized intensity that is less than five times the noise estimated by the AIPS task COMB using the rms levels from the Stokes Q and U images and the Q and U intensities for that pixel, $f_{pol}$ was taken to be zero.

54: 

55: \subsection{Brightness Temperature and Luminosity}

56: \subsubsection{Brightness Temperature and Luminosity}

57: To examine how brightness temperature and luminosity evolve on average, we can plot the brightness temperature and luminosity of each jet component as a function of its projected separation from the core in physical units rather than using apparent separations.  Perhaps the most desirable way to do this would be to measure the separations in terms of physically meaningful scale-lengths such as the Schwarzschild radius or the core radius of the host galaxy.  However, we do not have reliable black hole mass estimates for many of our sources \citep[see][]{hel07}, and for many of our higher redshift sources, the host galaxies are unresolved in their optical images.  These facts make it difficult to use such scale-lengths for a useful number of sources.\par

58: A more straightforward approach is to use the projected separation in units of parsecs.  While this does not take into account the distributions of black hole masses or host galaxy sizes, it provides us with a reasonable means for statistically constraining the evolution of jet components for the typical core-jet system using all of our sources with known redshifts.  When examining how the component brightness temperature and luminosity vary with projected core separation, however, we must deal with sources of bias brought on by the selection criteria for VIPS sources, the sensitivity and angular resolution of the VIPS 5 GHz observations, the sources' orientations, and the effects of cosmology.  These effects are problematic for two reasons.  First, the finite angular resolution of the 5 GHz images ensures that the jet components with the smallest projected separations (in physical units) will be within the most nearby sources.  Second, the components with the largest separations will likely be within the most distant sources.  This is because relatively faint jet emission that lies beyond the  limits placed on the image field of view by time averaging and bandwidth smearing of our VLBA data will not significantly effect the fit to the visibility data obtained using the clean algorithm with DIFMAP, and it will not be included in the final image of the source.  For these two reasons, we must contend with the following five dominant sources of bias:

59: \begin{enumerate}

60: \item The VIPS 8.5 GHz integrated flux density limit of 85 mJy will cause jet components within more distant sources, which likely dominate those with the largest projected separations, to be biassed toward more luminous sources which likely have brighter jet components on average.

61: \item Cosmological effects reduce the intrinsic brightness temperature of each jet component by a factor of $(1+z)$ \citep[e.g.,][]{kel98} which, when combined with the noise level in each 5 GHz image, will bias jet components within more distant sources to higher intrinsic brightness temperatures.

62: \item Since the typical spectral index (for $F_{\nu} \propto \nu^{\alpha}$) for jet components is roughly $-0.8$, the observed brightness temperatures will be even lower for jet components within sources at higher redshifts.  Combined with cosmological effects, this implies that the brightness temperature of a typical jet component is reduced by a factor of about $(1+z)^{1.8}$.

63: \item To be able to measure the intrinsic brightness temperature of a jet component, it must be resolved, implying that for redshifts at or approaching 1.605 where the angular diameter distance peaks for our assumed cosmology, there will be a bias toward physically larger components.

64: \item Jet components within sources with smaller viewing angles will, on average, have larger Doppler factors, $\delta$, and therefore, larger observed brightness temperatures and luminosity densities and possibly smaller than average projected separations from their cores.  This bias could be made stronger if the typical jet speed is larger closer to the core.  However, most observational evidence seems to indicate that jet speeds remain relatively constant \citep[see ][and references therein]{kel07}.

65: \end{enumerate}

66: We may construct a sample of jet components that is free of the first four biases listed above by setting a maximum redshift, $z_{max}$, and then selecting those components which (1) have source 8.5 GHz luminosity densities such that they would have $F_{8.5}>85$ mJy at $z=z_{max}$, (2) have observed brightness temperatures $>5\times (3.39\times 10^{8} \overline{\sigma})(1+z_{maz})^{1.8}$ where $\overline{\sigma}$ is the typical rms of the 5 GHz images of 0.24 mJy beam$^{-1}$, and (3) have intrinsic sizes large enough for them to be resolved if they were at a redshift of 1.605 or $z_{max}$, whichever is smaller, all assuming the component viewing angles are not changed.  We have developed such a sample using $z_{max}=2.3$ which yields a limiting intrinsic brightness temperature of $9.83 \times 10^{7}$ K and a limiting 8.5 GHz luminosity density of $3.46 \times 10^{34}$ ergs s$^{-1}$ Hz$^{-1}$.  Among the components from sources with redshifts that meet these criteria, 38 have large enough intrinsic sizes to be resolved at $z=1.605$.\par

67: Unfortunately, we do not have any data that would allow us to estimate the viewing angles of these 38 components, and therefore cannot eliminate the fifth bias in the list above.  However, since the component brightness temperatures are boosted by a factor of $\delta$ and the luminosity densities are boosted by a larger factor of $\delta^{n}$, where $n \approx $2.8-3.8 (for $\alpha \approx -0.8$), we may use plots of both these quantities versus projected core separation for these components to explore this effect.  We have plotted the intrinsic brightness temperature (i.e., corrected for the effects of cosmology, angular resolution, and spectral steepness) and the rest-frame 5 GHz luminosity density, assuming $\alpha=-0.8$, for these 38 components as functions of projected core separation in the upper and lower panels of Fig.\ \ref{jetcomps}, respectively.  Points representing components from the same sources are connected with dotted lines.  These lines show all four sources with multiple components represented in the plots have brightness temperatures that decrease with increasing projected separation.  Conversely, the luminosity densities of the same components show no tendency for either a positive or negative gradient.  Mean values which are plotted within bins of projected separation also show that, for all components, the typical brightness temperature is about twice as large at the smallest projected separations ($\sim$30 pc) than it is at the larger separations ($^{>}_{\sim}$150 pc), while the typical luminosity density remains roughly constant.  From these results we must conclude that either (1) the effect of the Doppler boosting bias is minimal and the trend seen in the upper panel of Fig.\ \ref{jetcomps} reflects a real trend for the intensity of jet components to decrease as they move farther from their cores, or (2) the trend is mostly caused by the typical Doppler factor being higher among components with smaller projected separations and the intrinsic luminosities substantially increase with increasing core separation.  While our data cannot rule out either scenario or even some combination of the two, we note that both are consistent with a real trend for the physical sizes of the components to expand as they move outward.  In fact, we find a significant correlation between intrinsic jet component size and projected core separation for the 38 components plotted in Fig.\ \ref{jetcomps} with a Spearman rank correlation coefficient of 0.47.%\par the way angular resolution and cosmology can effect any trend that may exist.  First, the observed brightness temperatures must be multiplied by a factor of $(1+z)$ to correct them for cosmological effects which effectively converts them into rest-frame brightness temperatures \citep[e.g.,][]{kel98}.  This biasses jet components from more distant sources to higher brightness temperatures.  This effect is problematic for two reasons.  First, the finite angular resolution of the 5 GHz images ensures that the jet components with the smallest projected separations (in physical units) will be within the most nearby sources.  Second, the components with the largest separations will likely be within the most distant sources.  This is because relatively faint jet emission that lies beyond the  limits placed on the image field of view by time averaging and bandwidth smearing of our VLBA data will not significantly effect the fit to the visibility data obtained using the clean algorithm with DIFMAP, and it will not be included in the final image of the source.  Because of these two reasons, the cosmological effects mentioned above may introduce a bias toward higher brightness temperatures at larger projected separations.  Similarly, we must also account for the fact that the VLBA angular resolution will have a larger effect on the measured sizes of components within more distant sources, which may artificially lower their measured brightness temperatures.\par

68: %To account for these effects, we have done the following.  First, we have computed deconvolved sizes for the jet components using their Gaussian fits and the dimensions of the restoring beam used; unresolved components were excluded from this analysis.  Next, we have computed the median rms for the 5 GHz images of the core-jets which corresponds to $2.29\times 10^{9}$ K, ignoring cosmological effects.  We then used this value to compute a 5$\sigma$ detection limit as a function of redshift (i.e., $1.15\times 10^{10}(1+z)$ K).  Through trial and error, we found that we could maximize the number of components used and the range in core separations while avoiding the biases brought on by cosmological effects if we confined ourselves to components with brightness temperatures $>6.31\times 10^{10}$ K which belong to sources with z$<$4.505, 179 components in all.  The brightness temperatures of these components are plotted versus their projected core separations in the upper panel of Fig.\ \ref{jetbtemp}.  Median brightness temperatures within bins of core separation are also plotted.  Data for components from the same sources are connected with dashed lines.  The slopes of these lines were also computed to obtain estimates of the brightness temperature gradients, and the distribution of gradients is plotted in the lower panel of Fig.\ \ref{jetbtemp}.  Nearly all of these gradients ($\sim$80\%) are negative, implying that the brightness temperatures generally decrease further away from the core.  The median values within core separation bins are consistent with this for projected separations larger than about 20 pc.  However, the median brightness temperature for the smallest core separations is significantly smaller.  While this is based on data for only $\sim10$ jet components, there are a total of 49 components with core separations $<20$ pc (i.e., 39 of them are outside the redshift and temperature limits used).  Among these, 43 (88\%) are fainter than $10^{12.4}$ K, which is roughly the largest temperature within the lowest core separation bin used in Fig.\ \ref{jetbtemp}.  This implies that the peak in the median values at a projected separation of between 20 and 30 pc is most likely real and indicates that on average, the jet components become brighter at these separations, and then fade as they move farther from their cores.\par

69: %One should note that even though the observed peak in brightness temperature occurs at 20 to 30 pc, that (1) these are {\it projected} separations, (2) relatively bright components are not absent at small separations but are simply more rare, and (3) even at projected separations of 20 to 30 pc, there are many relatively faint components and that it is the {\it median} brightness temperature that is larger.  In other words, the location of the peak brightness temperature has been constrained statistically and does not necessarily apply to any one source at any particular epoch.

70: \subsubsection{Brightness Temperature and Jet Width}

71: The results presented in Fig.\ \ref{jetcomps} suggest that individual jet components tend to expand as they move outward, which is by no means a new result.  However, how the intensity of each jet varies with the dimension(s) of the jet can provide insights into the changes in internal energy that may occur as the jet moves outward.  To do this properly, we must consider not just the properties of individual components, but the intensity and width of each jet as a function of core separation.  This is made difficult, however, by the limited sensitivity of our observations which is adequate for identifying the brighter features within each jet, but not for detecting the underly, relatively continuous emission.  To overcome this limitation, we have created a combined mean image, or "stacked" image, of the 31 sources with jet components used in Fig.\ \ref{jetcomps}.  We have limited ourselves to these 31 sources because (1) they belong to a quasi-volume limited sample (i.e., ignoring viewing angle effects), (2) they each have at least one jet component that is detectable on the image out to the chosen maximum redshift of 2.3, and (3) the results presented in Fig.\ \ref{jetcomps} imply that the potential Doppler boosting bias is likely relatively weak for this sub-sample, at least for projected separations $\gtrsim 30$ pc.\par

72: We combined the images by first remaking each image with a pixel width corresponding to 2.12 pc and a restoring beam of $25.4 \times 15.3$ pc, the physical sizes of the nominal pixel size and restoring beam used for a redshift of 1.605.  The phase center for each image was also adjusted to be the position of the source's core.  These new images were then rotated so that the mean jet axis of each was oriented along the x-axis of the image after which each was multiplied by $(1+z)^{1.8}$ and divided by the angular area of its restoring beam to put it in units mJy mas$^{-2}$ at rest-frame 5 GHz.  Following this, an appropriate amount of Gaussian noise was added to each image so that its rms would be nearly equal to the image with the largest rms (1.4 mJy mas$^{-2}$) to ensure the images had roughly the same sensitivity.  The images were then averaged together and the end result can be seen in Fig.\ \ref{coaddimg}.  It can be seen that the jet in this image shows no clear signs of distinct components and predominantly consists of continuous emission.\par

73: To determine the peak intensity and width of the jet in the stacked image, Gaussian functions were fit to the image columns.  We have plotted the peak brightness temperatures from these Gaussian fits as a function of image position in the upper panel of Fig.\ \ref{coaddprof} along with the mean values for the components plotted in Fig.\ \ref{jetcomps}.  The plot shows the presence of a roughly Gaussian core, as expected, that extends to radii of about 27.5 pc.  Beyond this limit, the mean values from Fig.\ \ref{jetcomps} agree well with the results from the stacked image up to radii of about 150 pc.  Beyond this limit, the superior sensitivity of the stacked images allows for a detection of emission fainter than the $10^{7.99}$ K limit used for Fig.\ \ref{jetcomps} and we can see that the intensity continues to drop until the sensitivity limit of the image is reached at a radius of about 250 pc.  In the lower panel of Fig.\ \ref{coaddprof}, we have plotted the jet intensity versus the width obtained from the Gaussian fits to the image columns.  We note that most of the VIPS jets have a moderate amount of jet bending (see \S 3.2.1), which will cause the width about the mean jet axis on the stacked image to be slightly inflated.  We note that while jet components may deviate from the true jet axis to a larger degree at larger core separations, this is likely not true in practice for our analysis since we determined the orientation of each source's jet axis using a linear fit to the positions of its components on its 5 GHz image.  We have therefore estimated the magnitude of this effect by measuring the separations of the jet components from their mean jet axis for each of the sources contained in the stacked image that have more than one jet component, 18 sources in all.  For the 53 jet components of these 18 core-jet systems, the median absolute deviation of their jet axis separations is 3.1 pc.  We have therefore corrected the jet width measurements from the stacked image for an extra 3.1 pc of dispersion caused by jet bending.  The relationship between peak intensity and width for the jet on the stacked image is well approximated by a power-law with a slope of $-2.68\pm0.03$, which is quite similar to what was found for the jet of 3C 120 by \citet{wal87}.  The implications of this relationship will be discussed in \S 4.

74: 

75: %\subsubsection{Fractional Polarization}

76: %The trend for jet components to be fainter at larger separations from the core makes interpreting a plot of $f_{pol}$ versus projected separation from the core difficult.  As discussed above, $f_{pol}$ will be biassed toward higher values for fainter jet components, so that the trend seen in Fig.\ \ref{jetcomps} would artificially introduce a trend between $f_{pol}$ and separation from the core.  Ideally, we could avoid this bias by selecting a sub-sample of jet components based on observed brightness temperature and $f_{pol}$.  However, there are only 28 jet components with detected polarized flux density which belong to sources with known redshifts.  Selecting an even smaller sub-sample from these 28 would make it difficult to statistically constrain the evolution of $f_{pol}$ among jet components.  An alternative approach is to estimate a $f_{pol}$ excess for each component given by

77: %\begin{equation}

78: %\delta f_{pol} = \frac{f_{pol} - \overline{f_{pol}}(T_{B})}{\overline{f_{pol}}(T_{B})}

79: %\end{equation}

80: %where $\overline{f_{pol}}(T_{B})$ is the mean value of $f_{pol}$ at a particular observed brightness temperature, $T_{B}$.  To determine $\overline{f_{pol}}(T_{B})$, we have computed the mean value of log $f_{pol}$ within $T_{B}$ bins for all components with detected polarized intensity (i.e., not just those with known redshifts).  In this case, $T_{B}$ has not been corrected for cosmological effects or the angular size of the restoring beam.  The results are plotted in Fig.\ \ref{jetfpol}, and it can be seen that there is a definite inverse correlation between the mean $f_{pol}$ and the observed brightness temperature which is well approximated by a power law.  A weighted, linear least-squares fit applied to the mean values within the $T_{B}$ bins was used to determine $\overline{f_{pol}}$ at the brightness temperature of each jet component which was then used to compute $\delta f_{pol}$.  For those components with known redshifts, $\delta f_{pol}$ is plotted as a function of the projected separation from the core in the lower panel of Fig.\ \ref{jetfpol} along with mean values of $\delta f_{pol}$ within core separation bins.  Unfortunately, only two core-jets have more than one jet component with $f_{pol}>0$, and a distribution of $\delta f_{pol}$ gradients could not be measured.  However, the mean values of $\delta f_{pol}$ within bins of core separation give some indication that the fractional polarization tends to increase as components move away from the core, most dramatically at projected separations of $\sim$100 pc.  The same is true individually for the two sources with more that one jet component with detected polarized intensity (see the dashed lines in the lower panel of Fig.\ \ref{jetfpol}).

81: 

82: \subsection{Jet Direction}% on parsec and kiloparsec Scales}

83: \subsubsection{Jet Bending on Parsec Scales}

84: The projected trajectory of jets may provide important insight into how the environments of jets affect their evolution.  As a first step toward exploring this, we have derived a quantitative estimate of the amount of jet bending on parsec scales as measured using our 5 GHz images.  We have done this by developing an algorithm, detailed in Appendix A, for measuring the difference between the position angle of jet emission near the core, $PA_{in}$, and the position angle with respect to the core of the outer most jet component, $PA_{out}$, for all sources with two or more jet components, 214 sources in all.  The purpose of using an automated approach rather than a "by-eye" estimate of the amount of bending is so that a consistent, objective, and quantitative measure of the degree of jet bending could be made without being significantly affected by stray or spurious components.  The distribution for $|\mbox{PA}_{in}-\mbox{PA}_{out}|$ determined using this algorithm is displayed in Fig.\ \ref{angs}.  The distribution for sources with more than two jet components, for which the algorithm provides a better estimate of the amount of bending (see Appendix A), is also plotted separately in grey.  The peaks of the distributions appear to be between 0$^{\circ}$ and 5$^{\circ}$, implying that a modest amount of jet bending on parsec scales may be common.  However, there are instances of significant amounts of jet bending with about 20\% of sources having $|\mbox{PA}_{in}-\mbox{PA}_{out}|>20^{\circ}$ and position angle changes of up to about 90$^{\circ}$.  Using this quantitative estimate, we have found no obvious trend between the degree of jet bending and jet-to-core ratio, core or jet brightness temperature, and core or jet fractional polarization.

85: 

86: \subsubsection{Changes in Jet Direct from parsec to kiloparsec Scales}

87: To examine how the motion of parsec-scale jet components relates to larger, kiloparsec-scale structures, we have employed VLA images of VIPS sources.  All VIPS sources have 8.5 GHz VLA images since the parent catalog for VIPS is the Cosmic Lens All Sky Survey (CLASS), an 8.5 GHz VLA imaging survey.  However, for about 98\% of VIPS sources, there are also 1.5 GHz VLA images available via the VLA FIRST survey, a relatively deep 21-cm survey of the SDSS footprint with the VLA.  Visual inspection of the CLASS and FIRST images of VIPS sources revealed that the quality of the FIRST images is typically much better (i.e., they have residual sidelobes less frequently) and that the sensitivity of the FIRST survey allows for extended emission to be present more frequently in the FIRST images than in the CLASS images.  Because of this, we have elected to use images from the FIRST survey to find and measure the properties of extended emission on kiloparsec-scales for the VIPS sources.  Three examples of VIPS sources with kiloparsec-scale structure apparent in their FIRST images are shown in Fig.\ \ref{eximg}.  These examples represent three different scenarios in which (1) the parsec-scale and kiloparsec-scale structures are oriented along completely different axes (J11456+4220), (2) the parsec-scale jet appears to be changing direction toward the kiloparsec-scale structure (J07547+4832), and (3) the parsec-scale jet and kiloparsec-scale structure are oriented along nearly the same axis (J17233+3417).\par

88: To quantify the degree of alignment between the parsec and kiloparsec-scale structures, we have developed an algorithm to objectively measure the orientation of extended emission on the FIRST images similar to what was used for the VIPS 5 GHz images (see Appendix B).  In the upper panel of Fig.\ \ref{vipsfirst}, we have plotted the distribution for the absolute difference between the position angles of the parsec and kiloparsec scale jet axes.  For these position angle differences, we have not taken into account whether or not the parsec or kiloparsec-scale emission is one-sided.  We have used this approach because, while all of the parsec-scale jets appear one-sided on their 5 GHz VLBA images, it is likely that this purely the result of Doppler (de-)boosting and we cannot rule out the possibility that the jets are really double-sided.  The distribution is skewed toward lower values (skewness of $\gamma_{1}=0.44$), indicating a tendency for extended emission on these two different spatial scales to be somewhat aligned.  

89: %However, this distribution does not take into account the fact that in the FIRST images of about half of the sources, the extended emission is one-sided.  In the lower panel of Fig.\ \ref{vipsfirst}, we have plotted the difference in position angles, this time letting the values be as large as 180$^{\circ}$ for sources with one-sided kiloparsec-scale emission.  This distribution is even more noticeably skewed toward lower values, further implying a tendency for the parsec and kiloparsec scale jets to be aligned.  

90: We note that a K-S test implies that the probability that this distribution was drawn from a flat distribution is negligible.  We also note that we see no evidence of a bimodal distribution like the one found by \citet{pea88} using similar data.  The combined VIPS and FIRST data has allowed us to measure this distribution for 132 sources, as opposed to \citet{pea88} who where restricted to using 18 sources.  It is therefore plausible that the previously observed bimodal shape was predominantly the result of small number statistics.\par

91: The results in Fig.\ \ref{vipsfirst} imply that parsec and kiloparsec scale jets tend to be somewhat aligned, but they do not indicate the specific physical scales on which this alignment occurs.  To explore this, we have plotted absolute difference between the PAs of the jet components identified on each VIPS image and the PA of the extended emission on the corresponding FIRST image, $|\Delta\mbox{PA}|$, as a function of core separation in the upper panel of Fig.\ \ref{first1}.  We have also computed the gradient of $|\Delta\mbox{PA}|$ as a function of core separation for all sources with more than one jet component where a negative gradient implies the components are becoming more aligned with the parsec-scale emission as the move outward (i.e., $|\Delta\mbox{PA}|$ is shrinking) and a positive gradient implies the components are moving away from the axis of the parsec-scale emission.  Mean values of $|\Delta\mbox{PA}|$ within bins of core separation hint that $|\Delta\mbox{PA}|$ may be smaller for larger separations.  There is much stronger evidence for this in the distribution of $|\Delta\mbox{PA}|$ gradients displayed in the lower panel of Fig.\ \ref{first1}.  Whether all sources with more than one jet component are considered, or if only those with more than two jet components are used, the peak of the distribution is clearly $<0$.  In addition, for the 30 sources with more than one jet component, 19($\pm$4) have gradients $<0$, demonstrating that most sources have negative gradients.   This implies a tendency for parsec-scale jet components to be oriented closer to the direction of kiloparsec-scale structure at larger core separations within individual sources.  

92: %As with the results displayed in Fig.\ \ref{vipsfirst}, we have also recomputed $|\Delta\mbox{PA}|$ for those sources where the kiloparsec-scale structures appeared one-sided on their FIRST images, allowing $|\Delta\mbox{PA}|$ to be $>90^{\circ}$ for parsec-scale jet components that were on the opposite side of their cores as the one-sided kiloparsec-scale emission.  The results using these new values are plotted in the panels of Fig.\ \ref{first2} and are only marginally different than what can be seen in Fig.\ \ref{first1}.  In this case, the peak of the $|\Delta\mbox{PA}|$ gradient distribution is still negative and the distribution for sources with more than two jet components appears even more skewed toward negative values than it does in Fig.\ \ref{first1}.

93: 

94: \section{Discussion}

95: Using single epoch data for a large number of sources, combined with optical spectra, we have been able to statistically identify trends in the evolution of jet components.  Specifically, we have shown that the components typically expand as they move away from their cores, and that this expansion coincides with either a decrease in brightness temperature or a substantial increase in luminosity.  Relativistic hydrodynamic simulations of jets have demonstrated that the flux density of the typical jet component will remain roughly constant as it moves away from its source \citep[e.g., ][]{mid97}.  This leads us to conclude that it is likely that the potential Doppler boosting bias in Fig.\ \ref{jetcomps} is minimal and that on average, the intrinsic brightness of the components drop as their luminosites remain fairly constant.\par

96: This result implies that the trend between intensity and jet width seen in Fig.\ \ref{coaddprof} likely reflects a real physical relationship between these two quantities.  From this relationship, we may gain insights into how the typical jet loses or gains energy as it moves away from its core.  Following the analysis of \citet{wal87} of a similar trend observed for 3C 120, if the typical jet is assumed to be an optically thin cylinder, then the emissivity, $\epsilon$, is proportional to $T_{B} r^{-1}$, where $r$ is the jet width.  The minimum energy density of a medium with the observed emissivity, $\epsilon$, is given by $U_{min} \propto \epsilon^{4/7}$.  Our results then imply that for the typical jet, $U_{min} \propto r^{-2.10 \pm 0.02}$.  This in turn implies that the energy density per unit length has only a weak dependence on jet width, i.e., it is proportional to $r^{-0.1}$.  This is quite consistent with what was found by \citet{wal87} for 3C 120, implying that the relation (or lack thereof) between internal energy and jet width may be somewhat commonplace among AGN jets.  From this result, it appears that the typical jet loses little, if any, of its internal energy as it move away from its core, or that there is a reduction of the typical jet speed that nearly exactly balances any losses that occur.  However, as discussed above, the results plotted in Fig.\ \ref{jetcomps} imply that the mean Doppler factor does not significantly change with projected core separation for these sources, which in turn implies that the typical jet speed, $\beta$, does not change much either.  This is because \citet{ver94} and \citet{lis97} have shown that the most probable viewing angle for a jet source is $\theta \sim 0.6 \gamma^{-1}$, where the Lorentz factor $\gamma = (1-\beta^{2})^{-1/2}$.  The relation between the typical viewing angle and $\beta$ implies that in order for the mean Doppler factor to remain constant, the typical value of $\beta$ must also remain constant since $\delta$ is only a function of $\theta$ and $\beta$.  Consequently, we conclude that it is more likely that for these sources, little energy per unit length is lost as their jets move outward.  Combining this results with the fact that the integrated luminosity density of the typical jet component remains roughly constant (see \S 3.1.1 and Fig.\ \ref{jetcomps}) as it expands, implies that the bulk of this expansion is likely perpendicular to the jet.\par

97: In addition to this result, we have also demonstrated that the jet components on parsec scales tend be aligned with the direction of extended emission on kiloparsec scales.  We have also found that this alignment improves as the jet components move further from their cores on projected scales as small as $\sim$50-100 pc.  These results suggest that interactions with the surrounding medium may produce realignment of the jets on relatively small projected scales.  We note that our data do not provide the opportunity to thoroughly study the nature of this realignment to discern whether it consists of relatively gradual changes in jet direction, as with 3C 120 \citep{wal87}, or of more abrupt changes similar to the event recently observed for 3C 279 \citep{hom03}.  However, the fact that our sample shows evidence of relatively modest amounts of jet bending on projected scales of $\sim$100 pc suggests that the former scenario is more common than the latter.

98: 

99: \acknowledgements  The authors would like to thank the referee for useful comments and suggestions.  This research was performed while the lead author held a National Research Council Research Associateship Award at the Naval Research Laboratory.  Basic research in astronomy at the Naval Research Laboratory is supported by 6.1 base funding.  The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.

100: %

101: %events that change the direction of movement of the components to be along the direction of the larger scale emission, after which, the components begin to fade.  If this is correct, the results shown in the lower panel of Fig.\ \ref{jetfpol} suggest that these events also cause an increase in the observed amount of polarized intensity and that this increase lags behind that of the total intensity.  This result can be explained by a reduction of the Faraday depth which is possibly related to the components' expansion as they move outward.\par

102: %Such a collimation event was directly observed for the C4 component of 3C279 by \citet{hom03}.  On about 1998.2, C4 was seen to brighten and change direction so that its new line of motion was parallel to that of the larger scale emission seen previously by \citet{dep83}.  The change in direction and intensity is summarized in Fig.\ \ref{3c279}.  From this figure, it can be seen that the position angle of C4 with respect to the core changed about 5$^{\circ}$ over about 15 pc, which is similar to some of the curves plotted in Fig.\ \ref{first1} and \ref{first2}.  The brightness temperature of C4 also peaked at a projected separation of about 23 pc and subsequently faded as it moved away from the core of 3C279, which is consistent with the results shown in the upper panel of Fig.\ \ref{jetcomps}.  The results presented here indicate that events such as this one may be relatively common and that the properties of the collimation within 3C279 are somewhat typical for core-jet systems.%\par

103: 

104: %If such collimation events are common, one would expect to see their effect on the surrounding medium as well as their effect on the jets.  The collimation event observed in 3C279 was estimated by \citet{hom03} to occur at a de-projected distance of about 1 kpc from the core.  This implies that the collimation event may correspond to an interaction between the jet and the interstellar medium (ISM) at a substantial distance from the center of the host galaxy.  Such an interaction would have a distinct impact on the ISM and is likely, in some cases, to induce star formation.  Indeed, jet-induced star formation has been seen in several systems \citep[see][and references therein]{cro06}.  If jet collimation is common and the events that cause the collimation often lead to star formation, one would expect the typical signatures of active star formation to be more prevalent within systems with jets, especially within those with jet emission beyond projected separations of about 25 pc (see Fig.\ \ref{jetcomps}).  Unfortunately, typical star formation signatures such as radio continuum, far infrared, H$\alpha$, or UV luminosities or bluer optical colors can also be significantly affected by the presence of a strong AGN at these wavelength regimes, which is likely for the VIPS sources.\par

105: %An alternative to these star formation diagnostics is Balmer absorption lines, the strength of which can be used to infer the presence of an intermediate-age ($^{<}_{\sim}$1 Gyr old) stellar population.  The equivalent width of Balmer absorption lines will be significantly affected by Balmer emission lines if they are present, which is likely for the VIPS sources.  However, \citet{tre04} have developed a detailed routine for fitting linear combinations of template spectra produced using the models of \citet{bru03} that allowed them to accurately subtract the stellar continua and measure emission line properties for $\sim 10^{5}$ SDSS galaxies.  This procedure also allowed them to estimate the equivalent width of Balmer absorption lines with a correction for emission line filling, either by emission lines intrinsic to each galaxy or by sky emission lines.  We have obtained the \citet{tre04} measurements of the emission line corrected $H\delta_{A}$ spectral index \citep{wor97} for 20 of our VIPS sources matched with galaxies from the fourth data release of the SDSS.  The index H$\delta_{A}$ is a measure of the absorption line equivalent width (i.e., absorption is positive) for the H$\delta$ line.  The H$\delta$ absorption line is affected by emission line filling less than the H$\alpha$, H$\beta$, and H$\gamma$ lines, and is in a region of the spectrum where the continuum level is easier to measure than that for higher order Balmer lines where the continuum is plagued by metal absorption lines.  The $H\delta_{A}$ distribution for these 20 sources is displayed in Fig.\ \ref{hda}.  For comparison, we have included scaled distributions for elliptical galaxies and for Sc and Sd galaxies (i.e., typical star forming spiral galaxies) taken from the morphologically classified catalog of SDSS galaxies of \citet{nak03}, where the $H\delta_{A}$ values were taken from \citet{tre04}.  As expected, the typically quiescent elliptical galaxies have lower values of $H\delta_{A}$, mostly less than zero, while the Sc and Sd galaxies have $H\delta_{A}$ values ranging from roughly -1 to 6 $\mbox{\AA}$.  The $H\delta_{A}$ distribution for VIPS sources, however, spans values from -3 to 3 $\mbox{\AA}$ and is relatively flat with some hint of bimodality.\par

106: %We have also displayed in Fig.\ \ref{hda} the $H\delta_{A}$ distributions for all VIPS core-jets with projected jet lengths longer than 25 pc and for all VIPS sources with no emission beyond 25 pc (i.e., core-jets with lengths $<$25 pc and PS sources).  These distributions show that the low end of the $H\delta_{A}$ distribution for the VIPS sources which is consistent with quiescent elliptical galaxies is dominated by sources with no jet emission beyond 25 pc.  In contrast, the high end of the VIPS sources' $H\delta_{A}$ distribution is dominated by core-jets with jets that extend past 25 pc.  The distribution for these longer jets lies beyond the bulk of the distribution for elliptical galaxies and is consistent with some level of recent star formation.  The fact that the $H\delta_{A}$ values for the longer jets are smaller than about 35\% of Sc and Sd galaxies may imply that they typically have a lower relative amount of recent star formation than these galaxies.  However, the presence of an AGN could increase the optical continua of the VIPS sources relative to typical late-type spirals, which would yield lower $H\delta_{A}$ values for the same level of star formation found in Sc and Sd galaxies.  In either case, there is some evidence that the collimation events discussed above have frequently induced star formation within their host galaxies.  We stress that this conclusion is based on a small number of VIPS/SDSS galaxies and has excluded sources which are dominated in the optical by their AGN (i.e., quasars).  A more thorough analysis of the optical spectra of all VIPS sources is planned to address this question more completely.

107: 

108: \appendix

109: \section{Jet-bending}

110: To establish an objective and consistent quantitative measure of the amount of jet bending on parsec scales, we have developed the following algorithm which uses our 5 GHz images.  First, for each core-jet with more than two jet components, a parabola was fit to the positions of the core and the jet components in order to asses how much the jet bends.  For sources with only two jet components, this was formally not a fit, but rather an exact solution of the parabola that passes through the center of the core and each jet component.  There are 214 VIPS core-jet systems with two or more jet components for which this could be done, 104 of which have redshifts.  For each of these core-jets, the slope of the parabola at the position of the core was used to determine the inner position angle of the jet, PA$_{in}$.  The value of the parabola at the right ascension of the jet component farthest from the core was then used to compute the outer jet position angle relative to the position of the core, PA$_{out}$.  It was found that in about 15\% of cases, the extreme of the fitted parabola was in between the position of the core and the closest jet component.  In these cases, the difference between PA$_{in}$ and PA$_{out}$ is artificially large since the fitted parabola makes a significant turn in direction in a region devoid of any components.  Since this is not the case for a large fraction of the sources, we have elected not to abandon our parabola fitting method because of the motivations stated above.  Instead, we have chosen to take PA$_{in}$ to be the position angle of the nearest jet component relative to the core position in these few (32 out of 214) instances.  Examples of the parabolic fits can be seen in Fig.\ \ref{jbend}.

111: 

112: \section{Kiloparsec-scale Position Angles}

113: To quantitatively asses the degree of alignment between the parsec-scale extended emission and the kiloparsec-scale emission, we have used the following algorithm to measure the PA of the extended emission on the FIRST images.  For consistency, this algorithm was based on the routine used to measure the position angle on parsec-scales from the VIPS 5 GHz images \citep[see ][]{hel07}.  Using the AIPS task SAD, Gaussian components were fit to significant peaks in each FIRST image down to the 6$\sigma$ level.  As with the VIPS sources, the dominant Gaussian components within each FIRST image were identified as the brightest components whose combined flux density is $>$95\% of the flux density of all Gaussian components.  For each image with more than one component, a line was fit to the component positions and was used to measure a kiloparsec-scale position angle.  Each image with more than one component was then visually inspected to flag those sources whose morphology appeared roughly circularly symmetric and/or not oriented along a singular axis.  About 7\% of the sources with multiple components were flagged by this inspection.  For the remaining 93\% that were not flagged, we measured the absolute difference between the position angle of each parsec-scale jet component measured with respect to its core and the axis along which kiloparsec-scale structure was found in the FIRST image, referred to as $|\Delta\mbox{PA}|$.  

114: 

115: \begin{thebibliography}{}

116: %\bibitem[Adelman-McCarthy et al.\ (2006)]{ade06}  Adelman-McCarthy J. K., et al.\  2006, ApJS, 162, 38

117: \bibitem[Aharonian et al.\ (2005)]{aha05} Aharonian, F. et al.\  2005, A\&A, 441, 465

118: \bibitem[Beasley et al.\ (2002)]{bea02} Beasley, A. J., Gordon, D., Peck, A. B., Petrov, L., MacMillan, D. S., Fomalont, E. B., \& Ma, C.  2002, ApJS, 141, 13

119: \bibitem[Britzen et al.\ (2003)]{bri03} Britzen, S., Vermeulen, R., Taylor, G., Campbell, R. M., Pearson, T. J., Readhead, A. C. S., Xu, W., Browne, I. W., Henstock, D. R., \& Wilkinson, P.  2003, in {\it Future Directions in High Resolution Astronomy: A Celebration of the 10th Anniversary of the VLBA} eds. Romney, J. D. \& Reid, M. J. (Socorro, NM: NRAO), 85

120: %\bibitem[Bruzual \& Charlot (2003)]{bru03} Bruzual, G. \& Charlot, S.  2003, \mnras, 344, 1000

121: %\bibitem[Croft et al.\ (2006)]{cro06} Croft, S., et al.\ 2006, \apj, 647, 1040

122: \bibitem[de Pater \& Perley (1983)]{dep83} de Pater, I., \& Perley, R. A. 1983, \apj, 273, 64

123: %\bibitem[Fomalont \& Ma (2002)]{fom02} Fomalont, E. B., \& Ma, C.  2002, ApJS, 141, 13

124: %\bibitem[Gabuzda et al.\ (2000)]{gab00} Gabuzda, D. C.  et al.\ 2000, MNRAS, 319, 1109

125: \bibitem[Gehrels \& Michelson (1999)]{geh99} Gehrels, N. \& Michelson, P.  1999, Astroparticle Physics, 11, 277

126: \bibitem[Healey et al.\ (2008)]{hea08} Healey, S., et al.\ 2008, \apjs, in press

127: \bibitem[Helmboldt et al.\ (2007)]{hel07} Helmboldt, J. F., et al.\ 2007, ApJ, 658, 203

128: \bibitem[Homan et al.\ (2003)]{hom03} Homan, D. C., Lister, M. L., Kellermann, K. I., Cohen, M. H., Ros, E., Zensus, J. A., Kadler, M., \& Vermeulen, R. C. 2003, \apj, 589, L9

129: %\bibitem[Kaspi et al.\ (2000)]{kas00} Kaspi, S., Smith, P. S., Netzer, H., Maoz, D., Jannuzi, B. T., \& Giveon, U.  2000, \apj, 533, 631

130: \bibitem[Kellermann et al.\ (1998)]{kel98} Kellermann, K. I., Vermeulen, R. C., Zensus, J. A., \& Cohen, M. H.  1998, \aj, 115, 1295

131: \bibitem[Kellermann et al.\ (2007)]{kel07} Kellermann, K. I., et al.\  2007, \apss, 311, 23

132: \bibitem[Laing (1980)]{lai80} Laing, R. A.  1980, \mnras, 193, 439

133: \bibitem[Lister \& Marscher (1997)]{lis97} Lister, M. L. \& Marscher, A. P.  1997, \apj, 476, 572

134: \bibitem[Lister \& Homan (2005)]{lis05} Lister, M. L. \& Homan, D. C.  2005, \aj, 130, 1389

135: \bibitem[Mioduszewski et al.\ (1997)]{mid97} Mioduszewski, A. J., Hughes, P. A., \& Duncan, G. C. 1997, \apj, 476, 649

136: \bibitem[Myers et al.\ (2003)]{mey03} Myers, S. T., et al.\  2003, \mnras, 341, 1

137: \bibitem[Nakamura et al.\ (2003)]{nak03} Nakamura, O., Fukugita, M., Yasuda, N., Loveday, J., Brinkmann, J., Schneider, D. P., Shimasaku, K., \& SubbaRao, M.  2003, \aj, 125, 1682

138: \bibitem[Pearson \& Readhead (1988)]{pea88} Pearson, T. J. \& Readhead, A. C. S.  1988, ApJ, 328

139: %\bibitem[Peck \& Taylor (2000)]{pec00} Peck, A. B. \& Taylor, G. B. 2000, ApJ, 532, 90

140: \bibitem[Pollack et al.\ (2003)]{pol03} Pollack, L. K., Taylor, G. B., \& Zavala, R. T.  2003, \apj, 589, 733

141: %\bibitem[Rodriguez et al.\ (2006)]{rod06} Rodriguez, C., Taylor, G. B., Zavala, R. T., Peck, A. B., Pollack, L. K., \& Romani, R. W.  2006, 646, 49

142: %\bibitem[Richards et al.\ (2006)]{ric06} Richards, G., et al.\  2006, \aj, 131, 2766

143: %\bibitem[Stoughton et al.\ (2002)]{sto02} Stoughton, C., et al.\  2002, \aj, 123, 485

144: \bibitem[Sjouwerman et al. (2005)]{sjo05} Sjouwerman, L. O., Mioduszewski, A. J., \& Greisen, E. W.  2005, in {\it Future Directions in High Resolution Astronomy} eds. Romney, J. \& Reid, M. (San Francisco:  ASP), 613

145: %\bibitem[Taylor et al.\ (1994)]{tay94} Taylor, G. B., Vermeulen, R. C., Pearson, T. J., Readhead, A. C. S., Henstock, D. R., Browne, I. W. A., \& Wilkinson, P. N.  1994, \apjs, 95, 345

146: \bibitem[Taylor et al.\ (1996)]{tay96} Taylor, G. B., Vermeulen, R. C., Readhead, A. C. S., Pearson, T. J., Henstock, D. R., \& Wilkinson, P. N.  1996, \apjs, 107, 37

147: \bibitem[Taylor et al.\ (2005)]{tay05} Taylor, G. B., Fassnacht, C. D., Sjouwerman, L. O., Myers, S. T., Ulvestad, J. S., Walker, R. C., Fomalont, E. B., Pearson, T. J., Readhead, A. C. S., Gehrels, N., \& Michelson, P. F.  2005, \apjs, 159, 27

148: \bibitem[Taylor et al.\ (2007)]{tay07} Taylor, G. B., et al.\ 2007, \apj, in press

149: \bibitem[Vermeulen \& Cohen (1994)]{ver94} Vermeulen, R. C. \& Cohen, M. H.  1994, \apj, 430, 467

150: %\bibitem[Tremonti et al.\ (2004)]{tre04} Tremonti, C. A., et al.\  2004, \apj, 613, 898

151: %\bibitem[Worthey \& Ottaviani (1997)]{wor97} Worthey, G. \& Ottaviani, D. L.  1997, \apjs, 111, 377

152: %\bibitem[Wrobel \& Walker (1999)]{wro99} Wrobel J. M. \& Walker, R. C.  1999, in {\it Synthesis Imaging in Radio Astronomy II} eds. G. B. Taylor, C. L. Carilli, \& R. A. Perley (San Francisco:  ASP), 171

153: \bibitem[Walker et al.\ (1987)]{wal87} Walker, R. C., Benson, J. M., \& Unwin, S. C.  1987, \apj, 316, 546

154: \bibitem[York et al.\ (2000)]{yor00} York, D. et al.\  2000, \aj, 120, 1579

155: \bibitem[Zavala \& Taylor (2004)]{zav04} Zavala, R. T. \& Taylor, G. B.  2004, \apj, 612, 749

156: \end{thebibliography}{}

157: 

158: %\begin{figure}

159: %\plotone{jprops_fpol.ps}

160: %\caption{For all core-jets with more than one component, the jet-to-core ratio (upper; see Sec. 3.1 for the definition), the brightness temperature of the core (middle), and the brightness temperature of the brightest jet component (lower).  In all panels, the distributions for core-jets with detected polarized intensity within their cores and those with polarized intensity detected with their jets are plotted as dotted lines in dark grey and light grey respectively.  For the jet-to-core ratio and jet brightness temperature, the K-S test derived probabilities that the distributions for sources with polarized jets were drawn from the same parent distribution as all core-jets are $4\times 10^{-4}$ and $2\times 10^{-10}$ respectively.  The results of all other K-S tests (see text) were inconclusive.}

161: %\label{jprops1}

162: %\end{figure}

163: %\clearpage

164: 

165: %\begin{figure}

166: %\plotone{fpol_btemp_cfrac.ps}

167: %\caption{The mean fractional polarization, $f_{pol}$, for the core (upper) and the jet (lower) versus the jet-to-core ratio (left; see \S 3.1 for the definition), the brightness temperature of the core (middle), and the brightness temperature of the brightest jet component (right).  Within bins of jet-to-core ratio and brightness temperature, "box-and-wisker" representations of the $f_{pol}$ distributions are plotted as shaded boxes where the top and bottom boundaries of each box represent the upper and lower quartiles, the horizontal line within each box represents the median, and the dashed lines extend to the extrema.}

168: %\label{jprops2}

169: %\end{figure}

170: %\clearpage

171: 

172: \begin{figure}

173: \plotone{f1.eps}

174: \caption{The intrinsic brightness temperature (upper) and rest-frame 5 GHz luminosity density (lower) versus projected separation from the core for jet components with intrinsic intensities $T_{B}>10^{7.99}$ and which would be resolved at $z=1.605$ where the angular diameter distance peaks (see \S 3.1) and which have parent sources with 8.5 GHz luminosity densities $>10^{34.5}$ ergs s$^{-1}$ Hz$^{-1}$ and $z<2.3$.  In both panels components belonging to the same source are connected with a dotted line and mean values are plotted within bins of projected core separation as grey points with errorbars.}

175: \label{jetcomps}

176: \end{figure}

177: \clearpage

178: 

179: \begin{figure}

180: \plotone{f2.eps}

181: \caption{The mean rest-frame 5 GHz image for the 31 sources with components plotted in Fig.\ \ref{jetcomps} (see \S 4.1.2) with contours (in red) for intensities of 0.6$\times$(-1, 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024) mJy mas$^{-2}$.}

182: \label{coaddimg}

183: \end{figure}

184: \clearpage

185: 

186: \begin{figure}

187: \plotone{f3.eps}

188: \caption{Upper:  The peak brightness temperature from the stacked image displayed in Fig.\ \ref{coaddimg} along the jet axis as determined using Gaussian fits to the image columns.  The rms value from the image is represented by a dotted line.  The mean values for the components plotted in Fig.\ \ref{jetcomps} are plotted in grey with the adopted limiting component intensity of $10^{7.99}$ K represented by a grey dotted line.  Lower:  The brightness temperature of the jet in the stacked image versus the full width at half maximum of the jet as determined using the Gaussian fits.  These widths have been corrected for a typical dispersion of 3.1 pc about the mean jet axis that results from jet bending determined using the individual components of the sources used to make the stacked image (see \S 4.1.2).}

189: \label{coaddprof}

190: \end{figure}

191: \clearpage

192: 

193: %\begin{figure}

194: %\plotone{jcomps_fpol.ps}

195: %\caption{Upper:  For all jet components with detected polarized intensity, $f_{pol}$ as a function of observed brightness temperature with mean values for log $f_{pol}$ within $T_{B}$ bins plotted as grey points with error bars.  The grey line is a weighted linear least-squares fit to these mean values.   Lower:  For each individual jet component with detected polarized intensity, the fractional polarization excess, $\delta f_{pol}$ computed according to equation (1) using the fit plotted in the upper panel to compute the mean log $f_{pol}$ at the component's measured brightness temperature versus the projected separation from its core.  Components belonging to the same core-jet system are connected with dotted lines.  Mean values within different bins of core separation are plotted as grey points with error bars.}

196: %\label{jetfpol}

197: %\end{figure}

198: %\clearpage

199: 

200: \begin{figure}

201: %\vspace{-1in}

202: %\plottwo{bt085i/results_ps/J11456+4420.ps}{FIRST/J11456+4420_I.EPS}

203: %\vspace{-0.9in}

204: %\plottwo{bt085g/results_ps/J07547+4823.ps}{FIRST/J07547+4823_I.EPS}

205: %\vspace{-0.9in}

206: %\plottwo{bt085l/results_ps/J17233+3417.ps}{FIRST/J17233+3417_I.EPS}

207: %\vspace{-0.2in}

208: \includegraphics[scale=0.9]{f4.eps}

209: \caption{Examples of jet and lobe emission on kiloparsec (VLA 1.5 GHz; left) and parsec (VLBA 5 GHz; right) scales.  The contour levels used in each panel are 0.9$\times$(-1, 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096, 8192) mJy beam$^{-1}$.}

210: \label{eximg}

211: \end{figure}

212: \clearpage

213: 

214: \begin{figure}

215: \plotone{f5.eps}

216: \caption{The distribution for the absolute change in position angle from the core to the outer most jet component (see \S 3.2).  Only core-jets with at least two jet components that are separate from the core are used (i.e., $N_{C}>2$) and the distribution for core-jets with at least three jet components (i.e., $N_{C}>3$) for which the value of $|\mbox{PA}_{in}-\mbox{PA}_{out}|$ is more robustly determined (see Appendix A) is presented as a grey histogram.}

217: \label{angs}

218: \end{figure}

219: \clearpage

220: 

221: %\begin{figure}

222: %\plotone{vips_first.ps}

223: %\caption{The distribution of the absolute difference between the position angle of the parsec-scale jet emission measured from the VIPS images to that of the kiloparsec-scale emission measured from the FIRST images.  In the upper panel, the distribution is for the difference between the two jet axes.  In the lower panel, the distribution is for the difference in jet direction, i.e., for the sources with one-sided emission on their FIRST images ($\sim 50$\% of all sources), the change in PA is allowed to be as large as 180$^{\circ}$.}

224: %\label{vipsfirst}

225: %\end{figure}

226: 

227: \begin{figure}

228: \plotone{f6.eps}

229: \caption{The distribution of the absolute difference between the position angle of the parsec-scale jet emission measured from the VIPS images to that of the kiloparsec-scale emission measured from the FIRST images.}

230: \label{vipsfirst}

231: \end{figure}

232: 

233: \begin{figure}

234: \plotone{f7.eps}

235: \caption{Upper: For each individual jet component, the absolute difference between the position angle relative to the core position and the axis along which extended emission was found on larger scales from the FIRST survey images as a function of the projected separation from the core.  Core-jet systems with FIRST images that have only one component are not included.  Components belonging to the same core-jet system are connected with dashed lines.  Mean values within different bins of core separation are plotted as grey points with error bars.  Lower: The distribution of the gradients of dotted lines shown in the upper panel.  For each source with only 2 jet components, this was computed using the values for those two components.  For sources with more that 2 jet components, a line was fit to the values for the components.  Because of this, the distribution for sources with more than 2 jet components is represented separately as a grey histogram.  The medians of the two histograms are nearly identical and are represented by a vertical dashed line.}

236: \label{first1}

237: \end{figure}

238: \clearpage

239: 

240: %\begin{figure}

241: %\plotone{jcomps_first2.ps}

242: %\caption{The same as Fig.\ \ref{first1}, except that here, the absolute difference between the jet component position angle and the direction of extended emission in the FIRST survey images was computed taking into account the fact that about half of the FIRST images had visible extended emission on only one side (i.e., for these objects, the absolute difference between the position angles measured by the VLBA and the VLA was allowed to be $>90^{\circ}$; see \S 4.2).}

243: %\label{first2}

244: %\end{figure}

245: %\clearpage

246: 

247: %\begin{figure}

248: %\plotone{3c279_path.ps}

249: %\caption{The approximate path of the C4 component of 3C279 on the sky based on the results presented by \citet{hom03} is shown in the upper panel with the change in apparent direction and speed observed to occur at about 1998.2.  The position angle of C4 measured with respect to the core of 3C279 is plotted as a function of the projected separation from the core for ($\Omega_{m}$,$\Omega_{\Lambda}$,$h$)=(0.3,0.7,0.7) in the lower left panel.  For reference, the position angle of the larger scale emission seen with the VLA by \citet{dep83} is about $-145^{\circ}$.  The approximate brightness temperature of C4 is also plotted versus the projected separation from the core in the lower right panel.}

250: %\label{3c279}

251: %\end{figure}

252: 

253: %\begin{figure}

254: %\plotone{vips_hda.ps}

255: %\caption{The distribution of H$\delta$ absorption line equivalent widths as measured by \citet{tre04} (see \S 5) for VIPS sources matched with galaxies that have SDSS spectra (black histogram).  Among these sources, separate histograms have been made for sources with no detected emission beyond a projected core separation of 25 pc (green shaded histogram) and for core jets with projected jet lengths longer than 25 pc (purple shaded histogram).  For comparison, we have included histograms for elliptical galaxies (red histogram) and Sc and Sd galaxies (blue histogram) from the morphologically classified SDSS galaxy catalog of \citet{nak03} (see \S 5).  These two histograms have been scaled to the total number of VIPS/SDSS galaxies.}

256: %\label{hda}

257: %\end{figure}

258: 

259: \begin{figure}

260: \plotone{f8.eps}

261: \caption{Examples of the parabola-fitting procedure used to quantify the amount of jet bending described in \S 3.2.  In all panels, the contours were made using the total intensity 5 GHz VLBA images with the same levels as were used in Fig. \ref{eximg}.  Each core center is marked with a red $\times$ symbol and the jet component centers are marked with red points; the parabolic fits are plotted as blue curves.  The top panels show examples of two sources with more than three components; examples of sources with only three components are show in the bottom panels.  For the sources in the left panels, the inner position angles were measured using the slopes of the parabolic fits at the core positions.  In each of the panels on the right, the extreme of the parabolic fit lies in between the core and the nearest jet component, and the inner position angle was computed using the closest component (see \S Appendix A).}

262: \label{jbend}

263: \end{figure}

264: \clearpage

265: 

266: \end{document}

267: