I have been working through some pre-calculus problems, and wxMaxima is a fun tool to play with:
wxMaxima provides a mathematical workbook environment. Maxima is a Computer Algebra System coded in Common Lisp back in the 80s.
In the screenshot you can see the semi-circle function f(x), and the function h(x, x_2) generates the tangent line for any point on the semicircle. The helpful feature exhibited here is partial application, allowing me to specify, e.g., h(x,20), to provide the graph with a new function, the tangent line attached to that point on the semi-circle. With a little more work, I could generalize h to allow passing in any semicircle function.
I wanted to upload the wxmx file, but wordpress.com does not permit upload that file type, for reasons that are beyond me.
I really wanted a nice Git repository viewer, but didn’t want any of the free online services, for various reasons. I looked into Gogs, but it proved too difficult to install on my server. But the simpler GitList software does the job:
I want to play around with the 8sync framework, but the project is unfortunately lacking an API reference. So, I am writing example code to help me learn each of the modules, and thought I might as well share. Here is some example code for the(8sync agenda) module.
This illustrates how an Agenda can be used to interweave the execution of computations, by yielding to other computations during your loop:
scheme@(guile-user)> ,use (8sync agenda examples)
$2 = done
I wanted to move on quickly from completing the FFT interface, on to FM demodulation. However, I realized that if the rest of the project was going to practical and modular, I would need some kind of block style I/O buffer management system, like GnuRadio has. And as I started to pursue that, I realized that I would need to properly modularize my code.
Up until now, all the functions were either being dumped into the default module by the C code, or loaded from one Scheme file in the source directory. I took all the external C code and organized it in sensible module divisions, and then took all the C-based Scheme primitives and put them in two modules ((sdr primitives) and (sdr hackrf primitives)). Those two modules are created and loaded when HackRF Shell is launched, while the other modules can be compiled independently.
That all required taking the time to actually learn how the module system works, as well as understanding how to create and fill-out modules through the C interfaces. So, not much amazing progress to the outside observer, but I learned a lot. The FFT demo function is still working:
One neat thing about programming in Guile is that you can jump into various modules at any time, to play around in them or see what variables are defined in them (private and public):
HackRF Shell, copyright 2019 Christopher Howard
Initializing primitive functions.
Entering the Scheme shell.
GNU Guile 2.2.3
Copyright (C) 1995-2017 Free Software Foundation, Inc.
Guile comes with ABSOLUTELY NO WARRANTY; for details type `,show w'.
This program is free software, and you are welcome to redistribute it
under certain conditions; type `,show c' for details.
Enter `,help' for help.
scheme@(guile-user)> (set! %load-path (cons "/home/christopher.qlfiles.net/Repos/hackrf-shell" %load-path))
scheme@(guile-user)> ,m (sdr hackrf util)
scheme@(sdr hackrf util)> ,b
hackrf-sensible-defaults #<variable 56314e6aaac0 value: #<procedure hackrf-sensible-defaults (d)>>
filename #<variable 56314e6aa940 value: #<undefined>>
timed-read #<variable 56314e6aa8a0 value: #<procedure timed-read (d file-path start-h start-m end-h end-m)>>
hackrf-cb-rx-to-file #<variable 56314e6aa990 value: #<procedure hackrf-cb-rx-to-file (filename)>>
out #<variable 56314e6aa9e0 value: #<undefined>>
when-hour-min #<variable 56314e6bd660 value: #<procedure when-hour-min (h m l)>>
hackrf-cb-rx-to-stream #<variable 56314e6aaa30 value: #<procedure hackrf-cb-rx-to-stream (out)>>
cb-fft-to-gnuplot #<variable 56314e6aa8f0 value: #<procedure cb-fft-to-gnuplot (center-freq samp-rate averaging-alpha)>>
scheme@(sdr hackrf util)> hackrf-
hackrf-cb-rx-to-file hackrf-open hackrf-set-sample-rate
hackrf-cb-rx-to-stream hackrf-sensible-defaults hackrf-set-vga-gain
hackrf-close hackrf-set-baseband-filter-bandwidth hackrf-start-rx
hackrf-disable-amp hackrf-set-freq hackrf-stop-rx
scheme@(sdr hackrf util)> ,m (sdr util)
scheme@(sdr util)> ,b
bv-complex-to-mag #<variable 56314e6b4700 value: #<procedure bv-complex-to-mag (complex-float-bv n_complex_nums dest-bv)>>
freq-shifter #<variable 56314e6b4670 value: #<procedure freq-shifter (freq samprate)>>
The next step in my project will be to work out my block-style I/O buffer management system, which will require me to re-learn GOOPS, Guile’s object oriented programming extension, so that will likely still take a while.
I’m not quite sure I have all the bugs worked out yet, but I implemented simple averaging, as well and the frequency display. The latter required implementing fft-shift to move center frequency to the center. The above plot shows FM stations at 101.1, 102.5, 103.9, and 104.7 Mhz, matching local stations I know about.
These simple things took quite a while, mainly because of the busyness of other life responsibilities; but also because the process of implementation led to fixing some bugs, which led to cleaning up some code, which led to learning new things about Guile scheme and C.
In Guile scheme, I learned how to implement a procedure to auto-close unreachable ports:
Here are a few tips, some of which you might not learn unless you inspect the source code:
Population on a hex can only grow to a maximum of 499, at which point growth slows to zero. So, a full hex might look good but actually has stopped producing.
Subject to the caveat above, your hex population growth is a percentage of population already on the hex.
Migration from deep territory to the front is a slow process.
Because of the above three factors, you generally want to place new villages and towns at locations that are neither too close nor too far from the front of war. If they are too close, the bonus growth effect will be operating on hexes that have only a few troops that are passing through on their way to the front. But too far away, and the bonus will be applied to hexes that are saturated and cannot produce more population anyway.
Battle outcomes are basically just a function of which player has more troops on the hex. So, you generally don’t want to push forward your war front that part of the front is mostly full of saturated (499 troop) or nearly-saturated tiles. Pushing forward effectively generally means putting flags directly in front of your current line of flags and then removing the old ones. Then wait until those new hexes become saturated.
If you realized you have pushed your front forward too quickly and are getting overcome, you can usually save the line by just pulling the front back a little.
You may want to play with the -i 0 option to ensure basically fair starting balance between players.
It usually makes more sense, mathematically, to build a new village, than to upgrade to a town or castle. This is due to cost vs. growth percentage increase. Towns or castles don’t give any kind of actual defensive bonus, just the growth increase.
If you have a good grasp of the battle, migration, and population growth mechanics, you basically know what you need to know to win most games against the AI. However, a tricky part in some games is knowing how to manipulate the other players against each other, so that they are attacking each other more instead of you. If you have a strong, non-moving set of fronts, generally you AI will attack the other players who have weaker fronts. However, this can back-fire if you allow one opponent to become stronger than you, after slurping up weaker enemies. It is a careful balancing act that requires you to intelligently focus your resource development and front movement.
At first it is tempting to focus exclusively on capturing mines, but your overall front movement and population growth is generally the more important thing to focus on. Of course, a front-movement strategy that will eventually get you three more mines, certainly has some points going for it.