Plot Lognormal knowing ONLY values for parameters

[Alfonso Russo]

Hello everybody.
How can i plot a lognormal distribution having only knowledge of the parameters? No data at all.

Example:
mu= 10.1611 and sigma=0.7012

Thanks to all

[John Mullahy]

Alfonso: Using —twoway function— should be an easy way to do this. Assuming you want the density, perhaps try something like this:

local m=10.1611
local s=.7012
twoway function y=(1/(`s'*sqrt(2*_pi)*x))*exp(-(ln(x)-`m')^2/(2*`s'^2)), range(.001 100000)

A shortcut is to use the —normalden(.)— function:

twoway function y=(1/(`s'*x))*normalden((ln(x)-`m')/`s'), range(.001 100000)

You can of course specify the range to be whatever is appropriate for your application, and also use the other —twoway— options that may be desirable.

[John Mullahy]

P.S. An even more concise expression uses the 3-parameter version of —normalden—

twoway function y=(1/x)*normalden(ln(x),`m',`s'), range(0 100000)

[Alfonso Russo]

Ok. Fine. Now the problem comes with the assix. I included the xscale(log) option to rescale the assix but i got something like the imagine i have attached. How can i add more thicks to the left of 50000? Despite trying different options for range(xx xxxx) it changes only the section of the distribution, and not the value on the x assix.Graph.pdf

[John Mullahy]

Alfonso: You can control the x-axis label locations and ticks using the -xlabel- option. For example, consider the code

twoway function y=(1/x)*normalden(ln(x),`m',`s'), range(1000 100000) xscale(log) xlabel(1000 10000 100000)

Of course, you can specify whatever values are helpful for you in -xlabel(…)-