O4 4 J | DISTRIBUTION f (x) COMPUTED USING ONLY READINGS OVER 50 \— tHE BODY (c) the position of the peaks in the eomp | ™ 7 _ 50 cro CONCLUSIONS l \fo From the above, we nowdrawthe followi sional conclusions: (a) For the distribution f(z) considered hei I KE ' o = 2 oO = a I possible to obtain fairly accurate « § ' (+3 %) of total activity, net (integrated in a certain interval Az and of the shap if all the parameters involved in the prob. as collimation, ete., are properly chosen text above). This is true even if statistic in g(x’) and K(z,x#’) are of the order of 5 to WwW > E a —_ tu Yr \ I \ 1 1 l | _ I \ 1 1 \—_ ! (b The FWHMI of the kernel K(z,z’) does | to influence the calculation of the total instead it seems to affect the shape of f(z net (integrated) activity in subintervals practice, it is imposstble to recogni: changes in the shapeof f(z) that are sepe distances less than the FWHMandto « i I \ 60 Fic. 80-—Distribution {(z) computed using only readings over the body, I’. Tests on the Iterative Technique In order to establish whether the iterative technique reported elsewhere®? could lead to more accurate results in the case of the distribution f(2) considered here and in the case of the experimental parameters described above, a program of calculations was undertaken for FWHAI = 8.5 em and hk = 1 em. Theiteration was stopped when f(x) became negative somewhere within the interval (a,b), or when the error corresponding to g(x’) reached the level of the expected error. The results were deemed unsatisfactory inasmuch as (a) the approximations were not converging; (b) the amplitudes varied more than in the true knowndistribution f(2)—in other words, oscillations in the solution were not betng dampened; | | I DISTRIBUTION f (x) COMPUTED USING READINGS EXTENDING —BEYOND THE BODY r I RELATIVE AMOUNT OF ACTIVITY tained by taking readings beyond the interval (a,b) of the actual distribution, and by requiring f(x) to vanish outside this interval. The results for f(z} obtained with and without readings beyond the actual distribution are plotted respectively in Figures 80 and 81. These figures showa clear improvement, of the shape of f(z) in the interval 0-14 em when additional values of g(z)' are used. 60 REAL DISTRIBUTION ——| I | { | ! | | | { 20t— ——= —_— = Le | REAL DISTRIBUTION i ° = 40 ne | | ae F = ul > e were erroneous since they were related those apparent in g(x’); some of those g(x’) were due to statistical error. In conclusion, the results were consistent with ments made elsewhere as to the experimental c suitable for the success of iterative techniques. I | | J 0 10 20 Fig. 81—Distribution f(z) tending beyond-the body. 30 40 50 computed using read