Today is a double date when the date is
written in the format MM/DD/YY or YY.MM.DD. The next such date is on February
10, 2021. This is the seventh and last special date
of this year.
We had some problem solving a Hyper Sudoku and we wondered if it could be solved without guessing. For
this purpose, I modified the program
for converting a Sudoku to an Exact Cover
problem. (On September 9, 2010 is used this
program for solving the Hardest Sudoku.) With the program, I checked that the
two Hyper Sudokus as published on page 42 of Denksport Logisch 2-4★ vakantieboek NR.38, indeed did only have
one solution and could be solved without guessing.
Steen bij de Höfte
Around noon, Conny and I walked the through
the park called Zweringbeekpark. This year October 30, a replacement for the
marke stone Steen bij de Höfte
was placed in the park at about the original location of the stone.
BindingList with filtering
A BindingList<T> in the .NET framework can be used to bind
a user-interface control with a data source that consists of a list. The idea
is that the control will follow any changes in the contents of the list and
that through the control the contents of the list can be changed. But it lacks
good possibilities to implement filtering and sorting. In my search for how to
implement this, I found the blog:
Databinding with a filter and a sort. In the end, I created my own
solution, for a one-way use where the control cannot directly change the list.
I created a class derived from BindingList<T> with an
_inner member of type BindingList<T>. It implements
an event handler on ListChanged event of _inner, which
performs the filtering (and sorting) and then fills the Items
collection again with the elements (in the sorted order). Each change in the
Items collection will raise a ListChanged event. To prevent
this, the RaiseListChangedEvents property can be set to false
to surpress these events. However, it is not possibility to raise the
ListChanged event in a derived class, because it is not defined in
the proper way. (See:
How to raise base class events in derived classes in the .NET
programming guide.) I implemented a trick by which each change to the
Items collection is delayed and that the last change is applied after
RaiseListChangedEvents has been set to true again. (After
this, I also added a call to the ResetBindings method.)
I got some question from an artist with respect to random number generation on
an micro-controller. These usually have a build in pseudorandom number generator. Although these usually produced a good
random sequence, they generate the same sequence every time the
micro-controller is started. A common solution to avoid this, is to add a
call to randomSeed in the initialization with some hardware related
random source, for example, to call analogRead on an 'open' input pin.
However, analogRead from an open pin is usually not a very good random
source because they do not have a uniform distribution or contain a strong
50/60 Herzt signal from a nearby power line. The least significant bit, are
usually a better random source. There are some tricks to combine many samples
into a better signal with some shifting and exclusive-or bitwise operations.
(See for a long discussion about this: Better Arduino Random Numbers.) Another solution is to combine the 'good'
pseudo-random source with a 'weak' random hardware source. The artist wanted
to have a one-in-eight random change expression. I suggested to use:
(random(8) == (analogRead(A0) & 7))
An even better and genertic method is to mix the two random sources. See for
example the (untested) modified random function from the Arduino
static long myRandom(int n, int pin)
* Compute x = (7^5 * x) mod (2^31 - 1)
* wihout overflowing 31 bits:
* (2^31 - 1) = 127773 * (7^5) + 2836
* From "Random number generators: good ones are hard to find",
* Park and Miller, Communications of the ACM, vol. 31, no. 10,
* October 1988, p. 1195.
static long x;
long hi, lo;
x = x ^ analogRead(pin); // Mix with 'weak' random source
/* Can't be initialized with 0, so use another value. */
if (x == 0)
x = 123459876L;
hi = x / 127773L;
lo = x % 127773L;
x = 16807L * lo - 2836L * hi;
if (x < 0)
x += 0x7fffffffL;
return x % n;
Note that for us humans it is quite difficult to recognize pseudorandom
sequences from 'hardware' random sequences. Below two images, one generated
with a short pseusorandom sequence of length 127 and the other from
www.random.org. Click on the image so see
which is which.
Yesterday, we installed the Twinkly starter
light set with 105 RGB LED's. I installed the app on my phone and played
at little with it. This evening, I wanted to try to access it through a laptop
with the help of XLED - unofficial
control Twinkly - Smart Decoration LED lights. I spend some time installing
all the depended modules (using pip3) and was able to connect to it. But I
got invalid challenge-response errors like described in issue 30. I have a TW105SEUP09 device with firmware version 2.1.8. There
is a firmware update to version 2.3.8 available. I am thinking about developing
a program in C++ that uses the Real time LED operationg mode by sending UDP packets to port 7777. For
this I do need the authentication token. I hope I can retrieve this using the
XLED Python library.
Compile on save
I came up with very simple bash script, which will compile a give C++ program every time when it is
while true; do
exetime=`stat -c %Y $1`
cpptime=`stat -c %Y $1.cpp`
if [ $cpptime -gt $exetime ]; then
g++ $1.cpp -o $1
I wrote a program to convert Futoshiki puzzles into Exact Cover problems.
I first attempted this by generating all permutations for each row and column
that are allowed by restrictions, but that did not lead to an exact cover that
could be solved easily. With this second version, I took the same approach as
the method for for sudoku puzzles.
When I was cleaning the leaves in the garden, I noticed many very small
mushrooms on the trunk of the large tree in our garden. They were on the side
of trunk that is overgrown with moss. Appearantly, that part of the trunk is
also infected with some type of fungus. I have no idea if this is a bad sign or
that it is relatively benign condition. On the right a close-up of one of the mushrooms.
Pythagorean triple date
Today is a Pythagorean triple date because
16² + 12² = 12² + 16² = 20². It
was due to a new paper article by Ionica Smeets that I heard about these kind of special dates. This makes
it the eight special date of this year.
Conny and I went searching for three marke stones along the former marke Esmarke and Lonneker, namely:
Steen bij de Stroinksbleek, Steen bij erve Kleinvoort, and
Hoge Boekelersteen. We only found the first, which is not an original
stone, but one placed on Augustus 18, 2017. It happened to be near the entrance
of a school that Andy attended for some years.
We could not find the second stone, because the location was overgrown and we
could determine its location. The third stone is at the Groote Boekel
estate. We looked around the estate and some adjecent fields and roads and
walked about 4Km. We could also not locate the last stone, which should be on
the estate. Conny did take a nice
picture at the estate which happens to include me.
Twinkly red-orange animation
In the past weeks, I made some further attempts to control the Twinkly starter light set. I managed to write a Python script that writes the authenitication token to a file and changed
the mode to rt with some small modifications to the xled library, using the following script fragment:
file = open("token.txt", "w")
I also wrote a C++ program that would send UDP packets to port 7777 as
described in the documentation about Real time LED operating mode, but I failed to get it working. So, today, I
wrote a Python script, which uses the
movie mode to send movies of one frame to animate the LEDS with a
pattern consisting of red and 30 orange lights, where every second at most
five orange lights change place. Then, while writing this, I read the
documentation about the Real time LED operation mode, I realized that I had
left the byte out specifying the number of LEDs. When I modified the test
program, it worked. I modified the program to
have a similar effect as the Python script.
VUI - 202012/01
I find it rather bizar, that due to some statement about the a new
COVID-19 variant, officially known as
202012/01, all personal traffic with the United Kingdom has been stopped.
On one hand, I am surprised that now suddenly it is possible to shutdown
flights within very short noticed, while in the past people were adviced to
go into quarantine when coming from an area with a high infection rate. On the
other hand, I am surprised because the variant was already found in the
Netherlands. It might already have been going round for over a month, because
the variant was first detected early October from a sample taken in September.
It might even be the case that the variant is responsible for the 'third' wave
here in the Netherlands. We do not know this, because here in the Netherlands
not much samples are being sequenced for the detection of new variants. So, I
feel it is like slamming a door after giving the foo ample time to slip inside.
It is clearly an overreaction caused by fear.
Moleskin daily planner
This morning, I realized that I had not yet bought a Moleskin daily planner for
next year. I ordered one from the webshop of Bookshop Broekhuis for € 19.90. This evening, I found a Moleskin daily planner for 2021 being delivered, probably by someone
from the bookshop. Made me happy!
There have no been three cases being reported of the new variant of the
COVID-19 virus in the Netherlands. Two cases
are found in Amsterdam and one in the region of Rijnmond, which is near
Rotterdam. In all cases no direct relation to the United Kingdom was found,
which means that the new variant is already circulating in the Netherlands.
It is possible that the virus is not more infectious but spreading more easily
because it makes less people ill. It might look more like the normal
coronavirus for which
most people have some immunity. If people have no or vague symptoms, they are
more likely to infect others.
Goki puzzle: no touching colours
Before, I already investigated the number of solutions for the round Goki puzzle ignoring the colour of the pieces. Someone raised
the question about the number of solutions when pieces of with the same colour
are not allowed to touch. I extended the program with two new options. One option to generate an Exact Cover and one option to check the results. The program found
256,299 unique solutions with respect to rotations and mirroring.
Oklab colour space
About a week ago, I came accors the Oklab color space. This morning, I tried to use it for generating a
colour palette, like I did before with the
HSLuv colour space. For the a and b parameters, I used the sine
and cosine functions multiplated with a saturation factor. To get a palette
looking similar to the HSLuv palette, I had to use saturation (chroma) factors
ranging from 0.07 to 0.17 and values from 0.6 to 0.9 for the L
parameters. Althought Björn
Ottosson claims that the Oklab colour space is better than the other
colour spaces. About HSV he states: Only on this list because it is widely
used. Does not meet any of the requirements except having a D65 whitepoint.
He does not mention HSLuv nor compare it with his colour space. I am also a
little worried about colours getting out of range. In the palette below there
are eight clipping errors with values being 12.5% out of range. For a more
in-depth discussion see: Oklab: A perceptual color space for image processing on Hacker News.
I extended the program with another command for some
further analysis of the round Goki puzzle.
There are 1,789,051 unique solutions to the puzzle when the colour of the
pieces are ignored. For each of these solutions, there are 288 (=2*2*2*3*2*6)
solutions when the colour of the pieces are taken into consideration. This
leads to a total of 515,246,688 unique solutions not taking into account
rotations and mirroring. Only 256,299 of these solutions have no pieces with
the same colour touching eachother. That is only about 0.05% of all solutions.
According to the new command of the program only 26,661 of the 1,789,051 unique
solutions to the puzzle when the colour of the pieces are ignored, do have
solutions where no two pieces with the same colour are touching. That is about
1,49%. I furthermore, made a break down of how many of the unique solutions
ignoring the colours have solutions with no touching colours. These results
are presented in the table below. According to this table, 898 of the
unique solutions ignoring the colours do have seven solutions where the pieces
of the same colour are not touching eachother.
407 1 | 1739 10 | 84 30
1494 2 | 4187 12 | 158 32
1170 3 | 281 14 | 1 33
3631 4 | 24 15 | 104 36
441 5 | 1194 16 | 54 40
3879 6 | 317 18 | 3 42
898 7 | 599 20 | 165 48
4117 8 | 36 21 | 5 60
99 9 | 1539 24 | 6 64
11 11 | 18 28 |
Conny and I walked along the Mander circles
near the village Mander. We search for some border poles,
and found the following:
We searched for border pole 86-IV, but could not find it.
- At 15:40, pole 86-III
- At 16:08, pole 87
- At 16:17, pole 87-I
- At 16:25, pole 87-II
Conny and I went walking to Galgenberg,
which translate as gallows hill in English. It is named so, because in the
middle ages, there was a gallows where the corpses of death row inmates were
put on display as a deterrent. Before we went there, we looked at a watermill
along the mosbeek. While walking around the area, we found the following
- At 11:55, pole 83
- At 12:35, pole 84
- At 12:52, pole 85
- At 12:59, pole 84-I
This months interesting links
| November 2020
| January 2021