G'MIC Tutorial Fragments

Software G'MIC
101

Hello Garry.
I have new issues when generating the tutorial pages here, because of the Python svghelper.py script that doesn’t want to run.
G’MIC -run reports this error:

[gmic]-0./gmd2html/*foreach#5816/*expr#5823/(...)/*local#6112/run/__run/svgtoimg/ Execute external command 'python scripts/svghelper.py --densit(...)c8214260319 images/ampersand_circ.svg' in verbose mode.
  File "scripts/svghelper.py", line 12
SyntaxError: Non-ASCII character '\xc2' in file scripts/svghelper.py on line 13, but no encoding declared; see http://python.org/dev/peps/pep-0263/ for details

Any idea on how to fix this? I’m not a Python guy :slight_smile:

1 Like
102

That is curious. Probably a replacement will be pushed shortly. I think I can fix this before the press of the day begins.

103

Ah. Did not properly handle internationalization of character encoding. Methinks this will address my tendency to regard Caen as just another borough of New York City.

So lets try Commit 8df222c0

diff --git a/tutorial/cheats/scripts/svghelper.py b/tutorial/cheats/scripts/svghelper.py
index a40ecf9e..51e1601c 100755
--- a/tutorial/cheats/scripts/svghelper.py
+++ b/tutorial/cheats/scripts/svghelper.py
@@ -1,4 +1,5 @@
 #! /usr/bin/python
+#-*- coding: utf-8 -*-
 
 import argparse
 import svgpathtools
@@ -9,8 +10,8 @@ DESC   = """
 Plot points from SVG Path.
 """
 DENHLP = """
-Plot point quantity as powers of two; 4 → 2⁴ → 16.\n
-Larger → denser. Defaults to 10 (1024)
+Plot point quantity as powers of two; 4 -> 2^4 -> 16.\n
+Larger -> denser. Defaults to 10 (1024)
 """
 OUTHLP = """
 Text file for output plots.

Runs nicely here. Hopefully there as well.

gosgood@bertha ~/git_repositories/gmic-community/tutorial/cheats $ gmic it exec_cheat.gmd gmd2html. 2 ot exec_cheat.html
[gmic]-0./ Start G'MIC interpreter.
[gmic]-0./ Input text-data file 'exec_cheat.gmd'.
[gmic]-1./ Output image [0] as text-data file 'exec_cheat.html'.
[gmic]-1./ End G'MIC interpreter.

Also patched exec_cheat.gmd help display to be consistent with the python script. Fingers crossed.

1 Like
104

Looks good, this issue seems to be solved.
I get another one anyway:

[gmic]-0./gmd2html/*foreach#5816/*expr#5823/(...)/*local#6112/run/__run/svgtoimg/ Execute external command 'python scripts/svghelper.py --densit(...)c6102473865 images/ampersand_circ.svg' in verbose mode.
Traceback (most recent call last):
  File "scripts/svghelper.py", line 5, in <module>
    import svgpathtools
ImportError: No module named svgpathtools

Probably some Python package to install with pip ?

OK, I’ve done this:

$ pip install svgpathtools

and it started downloading billions of data :smiley:

1 Like
105

Yes. pip will install svgpathtools.
Sorry to have put you to so much trouble…

106

Well, unfortunately, the pip command seems to be not enough.

$ pip install svgpathtools
Requirement already satisfied: svgpathtools in /home/tschumperle/.local/lib/python3.8/site-packages (1.6.0)
Requirement already satisfied: svgwrite in /home/tschumperle/.local/lib/python3.8/site-packages (from svgpathtools) (1.4.3)
Requirement already satisfied: numpy in /home/tschumperle/.local/lib/python3.8/site-packages (from svgpathtools) (1.24.3)
Requirement already satisfied: scipy in /home/tschumperle/.local/lib/python3.8/site-packages (from svgpathtools) (1.10.1)

But I still get the error:

[gmic]-0./gmd2html/*foreach#5816/*expr#5823/(...)/*local#6112/run/__run/svgtoimg/ Execute external command 'python scripts/svghelper.py --densit(...)c6530366674 images/ampersand_circ.svg' in verbose mode.
Traceback (most recent call last):
  File "scripts/svghelper.py", line 5, in <module>
    import svgpathtools
ImportError: No module named svgpathtools
107

Hmmm…
Methinks #! /usr/bin/python will summon the system installation of Python, which will use the system python environment by default. You’ve installed svgtools locally in your user environment.

`This probably will report no error when you type it in your own local shell:

$ python -c "import svgpathtools"

However, another user will see an error.

Perhaps the site generation script runs gmd2html, in a process not owned by you, but somebody/something else, and that process does not use python modules installed in your personal directories.

108 
$ python -c "import svgpathtools"
Traceback (most recent call last):
  File "<string>", line 1, in <module>
ImportError: No module named svgpathtools
1 Like
109

Well…
Perhaps the most expedient thing to do is to remove the need for site generation to run python. Python is called to make the front-piece drawing; that could be replaced with a static image. Ditto
for the coarse ampersand at the bottom of the page. The %% %% regions that eventually call
Python will be removed. It will then be a gmic-only setting.

Can’t do it right now. The press of the day has started. But I’ll do another commit this evening. I am sure the world will survive not having tutorial updates for another day.

1 Like
110

Wait… I suspect my system to run Python 2 when I run python , while pip is the same as pip3…

EDIT:

$ python3 -c "import svgpathtools"

works!

111

OK, so replacing python with python3 in the tutorial pages works…

1 Like
112 
gosgood@bertha ~/git_repositories/gmic-community/tutorial/cheats $ python -V
Python 3.10.10

And you? Python 2.7.x or some such?
python3 will run python 3.x explicitly, so:

python3 -c "import svgpathtools"

should run without error.

However, don’t loose sleep on this. Removing the python dependency maybe the sanest route.
5:20 AM. Time to walk the puppy…

EDIT:
Ah! That is good news.
But it is still time to walk the puppy…

1 Like
113 
$ python -V
Python 2.7.18

but

$ pip -V
pip 20.0.2 from /usr/lib/python3/dist-packages/pip (python 3.8)

Another thing well done :slight_smile:

1 Like
114

So — should I remove the Python dependency that exec_cheat.gmd has, or does explicitly calling python3 solve the issue, i. e., in scripts/svgtoimg.gmic change:

    # Command to execute:
    -cmd="python scripts/svghelper.py --density "$dens" "$nflg" --output "$fn" "$ifile
    +cmd="python3 scripts/svghelper.py --density "$dens" "$nflg" --output "$fn" "$ifile

and commit that instead?
Won’t be touching anything until about 22:00 UTC in any case.

115

You decide :slight_smile:
Pros/Cons:

  • Keeping Python allows to reduce the number of image data added to the tutorial repository, as they will be generated automatically.
  • Keeping Python introduces a new dependency to an external tool, which is prone to changes and compatibility breaks in the future (see Python2 → Python3 recently).
116

OK. I opt for removing the python dependency:

Probably commit tonight — when you should be in bed (but maybe won’t be :wink: )

1 Like
117

Pushed python removal on cheats/exec_cheat.gmd:

On branch master - remove python dependency

Thank you, @David_Tschumperle. Now it’s evening walk time for the puppy…

1 Like
118

Thanks Garry, I’ve updated the tutorial pages.
Now I’m starting to read them, with a big bowl of popcorn in my hands.
I feel like I’m in for a treat (but not because of the popcorn).

Thanks a lot for all your work.

119

Here’s something to look at. I wanted to figure out how to use merge().

rep_maximum_average:
number_of_threads:=$_cpus

{$number_of_threads*100},1,1,1,:round(lerp(u(0,11,1,1),11,t/$number_of_threads));

eval[-1] :"begin(
  avg_values_per_thread=0;
 );
 avg_values_per_thread+=i;
 end(
  merge(avg_values_per_thread,max);
  print(avg_values_per_thread/100);
 );
 "

This brief code gives me a idea of what it actually do.

120

Thanks!

Had some fun with this. However, I’m putting the topic of threads off to the tail end of the Mathematical Expression tutorials. For that audience, people coming up the curve on G’MIC, the sanest advice I can (and should) give is: “Let G’MIC choose the threading model. Period. Full Stop.”

Oh, yes, some words about threading prefixes ":", "*", ">" and "<", because people will see them lurking in code (“What are those?”), but those words will come with caveats: ‘*’ and ‘:’ may not actually speed up processing at all. Likewise, don’t use ">" or "<" unless you know your particular algorithm demands a visitation order. Not sure? Then don’t use either, as guaranteed visitation order requires single-threaded operation, which will certainly be less optimal.

But — you’re past that audience. Have all the fun with threads you want.

I rewrote your script a bit for investigative purposes:

rmerge.gmic 
 1 rep_maximum_average: -skip ${1=""},${2=""}
 2    # textype ⇒ *t*hread *ex*ecution *t*ype
 3    textype,textypetoo=${1-2}
 4    number_of_threads:=$_cpus
 5    echo "We\ have:\ "$number_of_threads
 6    echo "Our\ thread execution types\ -\ input:\ "$textype",\ eval:\ "$textypetoo
 7 
 8    {$number_of_threads*100},{$number_of_threads*100},1,3,$textype${}"P     = vector3(0);
 9                                                                      idx   = round(3*t/$number_of_threads,-1);
10                                                                      P[idx]= round(u(11*t/$number_of_threads,11,1,1),-1);
11                                                                      P
12                                                                     "
13    -name. noiseimage
14    +normalize[noiseimage] 0,255
15    -output. mthreads.png
16    -remove.
17    
18    -eval[noiseimage] $textypetoo${}"begin_t(
19                      avg_values_per_thread=0;
20                      print(t)
21                    );
22               avg_values_per_thread+=norm(I);
23               end_t(
24                       critical(print(t,avg_values_per_thread/100^2*$number_of_threads));
25                    );
26               end(
27                    merge(avg_values_per_thread,max);
28                    print('---max_average---');
29                    print(avg_values_per_thread/100^2*$number_of_threads)
30                  )
31             "
32    -split[noiseimage] y,$number_of_threads
33    repeat $!
34       local[$>]
35          -channels. {int(3*$>/$number_of_threads)}
36          -echo {$>+1}":\ Average:\ "{ia}",\ Variance:\ "{iv}",\ Minimum:\ "{im}",\ Maximum:\ "{iM}",\ Sum:\ "{is}
37       -done
38    -done
39    -remove

Line 18: To make per-thread instances of avg_values_per_thread you need to use begin_t() instead of just begin() — that unadorned version runs once in thread 0; you get only one instance of avg_values_per_thread, one that only thread 0 populates. You have no other instances from which to choose a maximum in your call to merge(). begin_t() does per-thread set-up. Variables defined in that call are local to the thread. Now you have per-thread instances of avg_values_per_thread from which you can merge values.

Line 23: Likewise, use end_t() to do things just before the thread exits — like print out the bucket. See next remark.

Line 24: critical() ensures that its expression argument runs atomically — that it doesn’t get split across different threads of execution. I used it because otherwise, the printing of the thread id could occur earlier or later than its associated datum — there would be no way to disambiguate the printout otherwise.

Overall: It generates a diagnostic image, mthreads.png. The lower thread ids populated the red channel, middle thread id’s green, high thread id’s blue. Gives a quick visual confirmation that G’MIC assigns threads to sections of the pixel buffer in sequential order. Otherwise the colors would scatter all along the height of the image.

mthreads_Alice_8
mthreads_Alice_8800×800 222 KB

Dump of `rmerge.gmic` output 
gosgood@bertha /media/lydia/mthreads $ gmic rmerge.gmic rep_maximum_average '*','*'
[gmic]-0./ Start G'MIC interpreter.
[gmic]-0./ Input custom command file 'rmerge.gmic' (1 new, total: 4596).
We have: 40
Our thread execution types - input: *, eval: *
[gmic_math_parser] t = (uninitialized) (mem[17]: scalar)
[gmic_math_parser] t = (uninitialized) (mem[17]: scalar)
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = (uninitialized) (mem[42]: scalar)
[gmic_math_parser] '---max_average---' = (uninitialized) (mem[44]: vector17)
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = (uninitialized) (mem[62]: scalar)
[gmic_math_parser] t = 4
[gmic_math_parser] t = 33
[gmic_math_parser] t = 11
[gmic_math_parser] t = 2
[gmic_math_parser] t = 23
[gmic_math_parser] t = 35
[gmic_math_parser] t = 21
[gmic_math_parser] t = 39
[gmic_math_parser] t = 13
[gmic_math_parser] t = 29
[gmic_math_parser] t = 7
[gmic_math_parser] t = 5
[gmic_math_parser] t = 27
[gmic_math_parser] t = 0
[gmic_math_parser] t = 19
[gmic_math_parser] t = 9
[gmic_math_parser] t = 31
[gmic_math_parser] t = 3
[gmic_math_parser] t = 25
[gmic_math_parser] t = 37
[gmic_math_parser] t = 15
[gmic_math_parser] t = 17
[gmic_math_parser] t = 36
[gmic_math_parser] t = 24
[gmic_math_parser] t = 26
[gmic_math_parser] t = 14
[gmic_math_parser] t = 18
[gmic_math_parser] t = 28
[gmic_math_parser] t = 10
[gmic_math_parser] t = 38
[gmic_math_parser] t = 6
[gmic_math_parser] t = 1
[gmic_math_parser] t = 20
[gmic_math_parser] t = 12
[gmic_math_parser] t = 32
[gmic_math_parser] t = 30
[gmic_math_parser] t = 8
[gmic_math_parser] t = 34
[gmic_math_parser] t = 22
[gmic_math_parser] t = 16
[gmic_math_parser] t = 15

[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 30332.853667387008
[gmic_math_parser] t = 2
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 31002.581126758574
[gmic_math_parser] t = 23
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 46199.4686227684
[gmic_math_parser] t = 39
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 51480.988836662291
[gmic_math_parser] t = 13
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 35193.848779037406
[gmic_math_parser] t = 37
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 47521.786290954587
[gmic_math_parser] t = 20
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 40270.378002334597
[gmic_math_parser] t = 1
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 29007.301944291117
[gmic_math_parser] t = 24
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 48178.509871242524
[gmic_math_parser] t = 18
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 36297.851349411962
[gmic_math_parser] t = 8
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 42899.125388774875
[gmic_math_parser] t = 32
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 37610.50768864822
[gmic_math_parser] t = 17
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 34294.77197567463
[gmic_math_parser] t = 34
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 41575.587466009136
[gmic_math_parser] t = 12
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 50819.777767223357
[gmic_math_parser] t = 36
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 45535.799253311161
[gmic_math_parser] t = 14
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 28353.230868729115
[gmic_math_parser] t = 26
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 43341.428201053262
[gmic_math_parser] t = 38
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 49498.611350254061
[gmic_math_parser] t = 6
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 38946.776838678357
[gmic_math_parser] t = 16
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 32343.903074274062
[gmic_math_parser] t = 4
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 34964.318173113825
[gmic_math_parser] t = 30
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 33646.90011281109
[gmic_math_parser] t = 10
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 46861.159746437072
[gmic_math_parser] t = 28
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 29678.561232178214
[gmic_math_parser] t = 29
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 31666.907326784611
[gmic_math_parser] t = 7
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 40924.416886343002
[gmic_math_parser] t = 11
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 48839.458113040928
[gmic_math_parser] t = 31
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 35636.565902988434
[gmic_math_parser] t = 9
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 44877.175599760056
[gmic_math_parser] t = 3
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 33000.286287980081
[gmic_math_parser] t = 25
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 50159.828297389984
[gmic_math_parser] t = 35
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 43562.835854822159
[gmic_math_parser] t = 5
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 36957.409204721451
[gmic_math_parser] t = 27
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 27695.330300394893
[gmic_math_parser] t = 33
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 39615.845659919738
[gmic_math_parser] t = 19
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 38288.510858316426
[gmic_math_parser] t = 21
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 42232.88877002144
[gmic_math_parser] t = 0
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 27046.000944480351
[gmic_math_parser] t = 22
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 44219.414029459003
[gmic_math_parser] '---max_average---' = [ 45,45,45,109,97,120,95,97,118,101,114,97,103,101,45,45,45 ] (size: 17)
[gmic_math_parser] avg_values_per_thread/100^2*$number_of_threads = 51480.988836662291
1: Average: 5.5000457994219989, Variance: 10.090985759256853, Minimum: 2.612103708088398e-05, Maximum: 10.999990463256836, Sum: 2200018.319768799
2: Average: 5.6332893542196603, Variance: 9.5816783616372501, Minimum: 0.27504423260688782, Maximum: 10.999988555908203, Sum: 2253315.7416878641
3: Average: 5.7704154366585616, Variance: 9.1101382664903614, Minimum: 0.55001521110534668, Maximum: 10.999998092651367, Sum: 2308166.1746634245
4: Average: 5.9058766468685864, Variance: 8.6284305938681545, Minimum: 0.82500624656677246, Maximum: 10.999978065490723, Sum: 2362350.6587474346
5: Average: 6.0444772094285488, Variance: 8.1678953259097913, Minimum: 1.1000055074691772, Maximum: 10.999971389770508, Sum: 2417790.8837714195
6: Average: 6.1792098588848114, Variance: 7.7155750463362587, Minimum: 1.3750025033950806, Maximum: 10.999997138977051, Sum: 2471683.9435539246
7: Average: 6.3254137017980216, Variance: 7.286496033518346, Minimum: 1.6500345468521118, Maximum: 10.999953269958496, Sum: 2530165.4807192087
8: Average: 6.4534099814340475, Variance: 6.8637419200600469, Minimum: 1.9251043796539307, Maximum: 10.999983787536621, Sum: 2581363.9925736189
9: Average: 6.5977895209920403, Variance: 6.4495162883693062, Minimum: 2.2000439167022705, Maximum: 10.999987602233887, Sum: 2639115.8083968163
10: Average: 6.7406175768142935, Variance: 6.0609722856101138, Minimum: 2.4750003814697266, Maximum: 10.999974250793457, Sum: 2696247.0307257175
11: Average: 6.8765323236149554, Variance: 5.6664685828301025, Minimum: 2.7500076293945312, Maximum: 10.999983787536621, Sum: 2750612.929445982
12: Average: 7.0080290295583012, Variance: 5.3047347322104201, Minimum: 3.0250067710876465, Maximum: 10.999980926513672, Sum: 2803211.6118233204
13: Average: 7.144751217333674, Variance: 4.9398406505223793, Minimum: 3.3000013828277588, Maximum: 10.99998950958252, Sum: 2857900.4869334698
14: Average: 7.2814522379046682, Variance: 4.6009308417235362, Minimum: 3.575042724609375, Maximum: 10.999974250793457, Sum: 2912580.8951618671
15: Average: 7.426495402957797, Variance: 4.2626344751538436, Minimum: 3.8500277996063232, Maximum: 10.999984741210938, Sum: 2970598.1611831188
16: Average: 7.5649060485053061, Variance: 3.9432156546258401, Minimum: 4.1250200271606445, Maximum: 10.999997138977051, Sum: 3025962.4194021225
17: Average: 7.6947556419193743, Variance: 3.6237321491967598, Minimum: 4.4000482559204102, Maximum: 10.999964714050293, Sum: 3077902.2567677498
18: Average: 7.8387190625262262, Variance: 3.3337733210261691, Minimum: 4.6750025749206543, Maximum: 10.999994277954102, Sum: 3135487.6250104904
19: Average: 7.9730926009595393, Variance: 3.0407411731600331, Minimum: 4.9500021934509277, Maximum: 10.999984741210938, Sum: 3189237.0403838158
20: Average: 8.1185076229619977, Variance: 2.7872165787039629, Minimum: 5.2250041961669922, Maximum: 10.999999046325684, Sum: 3247403.0491847992
21: Average: 8.2501353002524382, Variance: 2.5244646324931344, Minimum: 5.5000343322753906, Maximum: 10.999993324279785, Sum: 3300054.120100975
22: Average: 8.3912606142354011, Variance: 2.2716497970423486, Minimum: 5.7750020027160645, Maximum: 10.999983787536621, Sum: 3356504.2456941605
23: Average: 8.5275903369712829, Variance: 2.0400438008999822, Minimum: 6.050015926361084, Maximum: 10.999996185302734, Sum: 3411036.1347885132
24: Average: 8.6589096027576922, Variance: 1.824979048672003, Minimum: 6.3250079154968262, Maximum: 10.99997615814209, Sum: 3463563.8411030769
25: Average: 8.798242226526737, Variance: 1.6123830444338805, Minimum: 6.6000075340270996, Maximum: 10.999978065490723, Sum: 3519296.8906106949
26: Average: 8.9384036519789696, Variance: 1.4218627859915545, Minimum: 6.8750162124633789, Maximum: 10.999991416931152, Sum: 3575361.4607915878
27: Average: 9.0753074894785879, Variance: 1.2314130959905678, Minimum: 7.1500191688537598, Maximum: 10.999990463256836, Sum: 3630122.9957914352
28: Average: 9.2130612678062924, Variance: 1.068267788592884, Minimum: 7.4250054359436035, Maximum: 10.999933242797852, Sum: 3685224.5071225166
29: Average: 9.3487650111269947, Variance: 0.90712310176046074, Minimum: 7.7000036239624023, Maximum: 10.999997138977051, Sum: 3739506.004450798
30: Average: 9.4864084709167482, Variance: 0.76263914503051577, Minimum: 7.9750089645385742, Maximum: 11, Sum: 3794563.3883666992
31: Average: 9.6247010512757303, Variance: 0.6295941643023546, Minimum: 8.2500114440917969, Maximum: 10.999995231628418, Sum: 3849880.4205102921
32: Average: 9.7635583670377724, Variance: 0.51059804363528893, Minimum: 8.5250129699707031, Maximum: 10.999996185302734, Sum: 3905423.3468151093
33: Average: 9.8999654232096663, Variance: 0.40186338801668781, Minimum: 8.8000068664550781, Maximum: 10.999999046325684, Sum: 3959986.1692838669
34: Average: 10.037592641599179, Variance: 0.30859681806450318, Minimum: 9.0750026702880859, Maximum: 10.999994277954102, Sum: 4015037.0566396713
35: Average: 10.174010604717731, Variance: 0.22740174993573165, Minimum: 9.3500022888183594, Maximum: 10.999998092651367, Sum: 4069604.2418870926
36: Average: 10.313058074333668, Variance: 0.1578218503463861, Minimum: 9.6250038146972656, Maximum: 10.999997138977051, Sum: 4125223.2297334671
37: Average: 10.449563414857387, Variance: 0.10088976044344576, Minimum: 9.9000053405761719, Maximum: 10.999998092651367, Sum: 4179825.365942955
38: Average: 10.587271196677685, Variance: 0.056689892499643783, Minimum: 10.175004959106445, Maximum: 10.999998092651367, Sum: 4234908.4786710739
39: Average: 10.725526492979526, Variance: 0.025144694210397604, Minimum: 10.450002670288086, Maximum: 11, Sum: 4290210.5971918106
40: Average: 10.862820333256721, Variance: 0.0063182894998874799, Minimum: 10.725000381469727, Maximum: 11, Sum: 4345128.1333026886
[gmic]-0./ End G'MIC interpreter.

Search for ---max_average--- in the dump. That locates the largest avg_values_per_thread that merge(avg_values_per_thread,max) picked: 51480.988836662291, accumulated by thread t = 39. No surprise there. The largest thread id operated over the last portion in the buffer with minimum pixel values of ≈10.725 → 11.000. merge() found this maximum value among all the thread-local copies of avg_values_per_thread and picked it, as expected.

Small aside:

round(lerp(u(0,11,1,1),11,t/$number_of_threads));

This seems simpler:

round(u(11*t/$number_of_threads,11,1,1),-1);

All that you want to do is to move the lower boundary of the selection interval as a function of thread_id: larger id ⇒ smaller intervals. Just make the first argument of u() such a function. Home free.

Have fun!