myprojects-hg-reborn

Fork

I am now saving in a file the geometric standard deviation. Careful
about another thing: the graphs with the arrows identifying the
residence time for a 5m-long anaconda are not automated: the residence
time for a 5m-long anaconda has to be set by hand!!!

I added an arrox and extra legend to a plot in
read_results_smoluchowsky_single_simulation.R.

I introduced the ylim() specification in the 2D number-size distribution
figure for the paper.

I slightly modified a figure in the code 3D_R_plotting.R

I modified the code 3D_R_plotting.R, which now plots a comparison
between the simulated number-size distribution at the end of the
anaconda and for a 5m-long anaconda.

I added the code plot_mass_size.R, which calculates the normalized
mass-size distribution from the the fitted mass-size distributions.

I corrected a few typos in the figures generated by 3D_R_plotting.R.

Now lognormal_binning.R plots the experimental initial and final states
(without corrections) with the error bars together with their fittings
(without extra corrections).

I corrected a few typos in the pictures produced by
R-codes/3D_R_plotting.R. Also, the labels are now larger.

I modified the code lognormal_binning.R. It now needs an extra input
file with the experimental errors associated to the number-size
distribution and it plots the distributions at TP and EA with the
statistical errors.

I modified 3D_R_plotting.R so that it now plots the .pdf files with the
correct margins.

More modifications to mydraft.tex before adding the section on the
one-dimensional approximation written by Yannis.

Now read_results_smoluchowsky_single_simulation.R plots also the
evolution of the total-number concentration vs the analytical
approximation you would have with (1) no losses, (2) constant kernel (3)
an initial monodispersed aerosol.

I added a new experimental part (by Makis) to mydraft.tex. I shifted
the numerical method after the discussion of the time-scales and I
introduce earlier on the form of the coagulation term before
re-expressing it in a different way in the section about numerics. I
still have to include the latest suggestions by Makis (comparison of the
re-corrected state with the experimental number-size distribution) and I
can possibly add a couple of figures to show the agreement.

I added some considerations about the turbulence time scale and the
saffman & Turner kernel and re-located the part about the numerical
method.

I went on revising mydraft.tex. Now all the statistics is at the
beginning and I corrected the diagram figure. I still have to assess the
relevance of the experimental results by Lee who used the same 1D
correlation for thermophoresis.

Further modifications of the paper and inclusion of a table with the
relevant time scales.

I started changing the structure of mydraft.tex. More about fractal
dimension and statistics and I re-derived a 1D equation for the particle
dynamics.

I added the latex code mydraft.tex.

I added the file mydraft.tex and created jas1 (folder containing the
paper I am writing for the Journal of Aerosol Science).

I added the code read_errors.R which I use to show the results of the
minimization of the distance (in a least-square sense) between the
results of my smoluchowski simulation and the final experimental state.

Now lognormal_binning.R also saves the cumulative number-size and
mass-size distributions.

Now in the input file I specify the units of the particle density.

Now I fixed with the appropriate factors the calculation of the
mass-size distribution in 3D_R_plotting.R.

I added a few features (e.g. cumulative mass-size distribution function)
to lognormal_binning.R and changed several plots.

See the documentation inside the hermite.f file. I modified a fortran 77
routine ti calculate hermite polys and used the Cf2py declaration (which
is not read by the fortran compiler) to let Cf2py know which variables
are input and which are output.

I re-fixed errors_vs_time.R and the definition of the total mass;
however, nothing really changes since the log_factor and log(10) factor
I added are simply constants).

I added a test to work out the initial mass (including now all the bits
and a revised effective density).

I now am working out correctly the total mass in the fitting of the
final state at EA. However, since it all involved fixing objects like
log_factor [which is a constant on the lognormally evenly spaced grid I
am using], the results for the minimization are the same.

I added the code errors_vs_time.R. This code allows one to postprocess
the results of the smoluchowsky2.py simulations in order to find a
the fitted fractal dimension as a function of time.