INNAME Input image name (name) INCLASS Input image name (class) INSEQ 0.0 9999.0 Input image name (seq. #) INDISK 0.0 9.0 Input image disk drive # IN2NAME Noise image name (name) IN2CLASS Noise image name (class) IN2SEQ 0.0 9999.0 Noise image name (seq. #) IN2DISK 0.0 9.0 Noise image disk drive # BLC 0.0 4096.0 Bottom Left corner of fit TRC 0.0 4096.0 Top Right corner of fit OUTNAME Output image name (name) OUTCLASS Output image name (class) (NOT USED AT PRESENT) OUTSEQ -1.0 9999.0 Output image name (seq. #) OUTDISK 0.0 9.0 Output disk drive # APARM (1) begin channel (2) end channel (3) initial guess of RM (4) max expected |RM| (5) > 0 use noise image (6) blanking method (7) blanking level SEE HELP re 4, 6, 7 SCALR1 -1.0 1.0 <= 0 => Output normal map > 0 => Output error on map

RM Type: Task Use: RM is a task which calculates rotation measure and intrinsic magnetic field from a set of (at least three) position angle maps made at different frequencies. The current version can accept a maximum of 65536 frequencies. The algorithm will do an unweighted or weighted fit of the position angle to the wavelength squared. In the latter case, sigma (error) maps must also be supplied. These error maps must be made with the task COMB. RM outputs both a map of the rotation measure, and a map of the intrinsic magnetic field. Various blanking options are supplied and it is vigorously suggested that one of them be used. BLC, TRC values: BLC(1-6) : TRC(1-6) apply to axes 2 - 7 of the input cube. In other words, they are the desired OUTPUT window rather than the input window. APARM(1), APARM(2), and APARM(8) control the input window on the first (actually frequency) axis of the cube. APARM values: (1): First frequency channel to use in fit, 0 -> 1 (2): Last frequency channel to use in fit, 0 -> N (3): The initial guess for the rotation measure. Use the integrated values if nothing else is known. UNITS are RADIANS per METER squared. (4): If > 0, use a method due to F. Zhou to search a wide range of lobe ambiguities, based on APARM(4) as the maximum expected RM in absolute value. (5): Solution type. <= 0 => Unweighted fit >0 => Weight fit by errors in IN2NAME et al. (6): Blanking type. 0 => No blanking 1 => Blank both output maps if corr. coeff. < APARM(7). 2 => Blank both maps only if sigma of RM exceeds APARM(7) (rad/m.m) 3 => Blank both maps only if sigma of B > APARM(7) (degrees) 4 => Blank both maps if rms dev. per point from best fit line exceeds APARM(7) (degrees) 5 => Blank both output maps if in input error of any input map value exceeds APARM(7). THIS IS THE RECOMMENDED WAY. (7): The blanking level. See APARM(6).

To run RM, the best thing to do is to start with a cube image (from IMAGR and COMB) or build a cube image either with MCUBE (regularly spaced frequencies) or FQUBE (irregularly spaced frequencies). Then tranpose the cube with TRANSCOD = '312' to make the FREQ or FQID axis be the first axis. Then run RM. Ancient intructions to run RM say that you must follow these steps: 1) Run AXDEFINE for each p.a. map and each sigma map (if you wish weighted fits), and change the 3rd axis (frequency) to an arbitrary unit (i.e. 'PERLEYS'), whose values increase uniformly from map to map. Although the order of the frequencies is not critical, the program stands the best chance of working properly if the 3rd axis increases with frequency. It is important that the two frequencies closest in wavelength have the 3rd axes set to 1 and 2. Set the third axis of the position angle maps from 1 to N, where N is the number of frequencies, then set the position angle error maps to have the third axis run from N+1 to 2N, keeping the same frequency order as the position angle maps. (If you do not intend to weight by fit, you need not include the error maps). When running AXDEFINE, set AXINC = 1 and AXREF = 1 also. 2) Run RENAME for each map to make the name and class of each the same. Make the sequence number of each map the same as the new "frequency" value (i.e. from 1 to N for the position angle maps, and N+1 to 2N for the error maps.) 3) Run MCUBE and assemble these N (or 2N if you want a weighted fit) maps into a data cube. 4) Run TRANS and transpose the cube. Make the initial first axis ('LL') the second, the initial second ('MM') the third, and the initial third (frequency in units of 'PERLEYS' or whatever) the first. That is, the new order is '312'. 5) Run RM ! INNAME is the transposed cube. Only the OUTNAME is used, OUTCLASS is pre-set to 'ROTMES' and 'BFIELD'.