min activities in a building

Started by Benahmed Abdelkrim, February 28, 2026, 11:14:18 AM

Previous topic - Next topic

0 Members and 2 Guests are viewing this topic.

Print
Go Down Pages 1 ... 5 6 7 8 9 ... 22
User actions

Liviu Lalescu

#90
March 13, 2026, 08:18:28 PM
No! We should not be satisfied! I will try to squeeze some more speed, but tomorrow morning - I hope it is not too late for you.

There might be room for improvement, as I had a look now at my old code.

Liviu Lalescu

#91
March 13, 2026, 08:19:51 PM
Please send us your MTR>0 file which exposes this problem (here on the public forum).

Benahmed Abdelkrim

#92
March 13, 2026, 08:32:33 PM
Quote from: Liviu Lalescu on March 13, 2026, 08:18:28 PMNo! We should not be satisfied! I will try to squeeze some more speed, but tomorrow morning - I hope it is not too late for you.

There might be room for improvement, as I had a look now at my old code.
Very good, let's continue ...

Quote from: Liviu Lalescu on March 13, 2026, 08:19:51 PMPlease send us your MTR>0 file which exposes this problem (here on the public forum).
 
Here it is

Tried with these random seeds, the generation time is very short = 39s:
s10=1773433512, s11=62939200, s12=0,
s20=1773433512, s21=62939300, s22=0.

B.A/krim

Liviu Lalescu

#93
March 13, 2026, 08:35:16 PM
See you tomorrow early morning! Good night!  :)

Benahmed Abdelkrim

#94
March 13, 2026, 08:36:20 PM
Good night!  :)
B.A/krim

Liviu Lalescu

#95
March 14, 2026, 07:29:20 AM
The new version, fet-7.8.0-nsrt6, is released. We managed to improve speed a bit, but under 10% from nsrt5 (the seeds should remain the same, you can compare).

I think this is the final one for the moment  :)

The behavior to jump seconds is normal.

Benahmed Abdelkrim

#96
March 14, 2026, 08:21:03 AM
I did a test with both versions, NSRT5 and NSRT6, and I assure you the results were significant. The time with NSRT5 was 4 minutes and 44 seconds, while the time with NSRT6 was 32 seconds! A BIG difference! The generation speed has improved considerably! I can confidently say that this latest version is the best of all!

The test is done with the following descriptions:
MTR=3
random seeds:
s10=123, s11=123, s12=123,
s20=123, s21=123, s22=123.
The random seeds at the end of the generation are not the same.

Thank you so much for this update which has brought improvements in speed and behavior.

The file is attached below ...
B.A/krim

Liviu Lalescu

#97
March 14, 2026, 08:48:06 AM
No, dear Benahmed, it was only luck; the speed improvement is under 10%. The seeds must be identical to compare. Otherwise, you could generate many timetables and see the average.

I am tempted to add these two constraints in the official FET, but they are too specialized...

Benahmed Abdelkrim

#98
March 14, 2026, 08:57:51 AM
Quote from: Liviu Lalescu on March 14, 2026, 08:48:06 AMNo, dear Benahmed, it was only luck; the speed improvement is under 10%. The seeds must be identical to compare. Otherwise, you could generate many timetables and see the average.
yes the seeds are identical with the  both versions.

Quote from: Liviu Lalescu on March 14, 2026, 08:48:06 AMI am tempted to add these two constraints in the official FET, but they are too specialized...
 
it's a Good Idea! All FET users can take advantage of NSRT's features to create exam timetables.

B.A/krim

Liviu Lalescu

#99
March 14, 2026, 09:00:58 AM
To compare, the final seeds must be identical also.

Benahmed Abdelkrim

#100
March 14, 2026, 09:06:14 AM
Quote from: Liviu Lalescu on March 14, 2026, 09:00:58 AMTo compare, the final seeds must be identical also.
even though the two versions are not the same?!
B.A/krim

Benahmed Abdelkrim

#101
March 14, 2026, 09:13:27 AM
Another test with a slightly different file. Generation time = 1 minute. The seeds are as follows:
s10=658647871, s11=2950353718, s12=1319523356,
s20=2607661595, s21=2963827883, s22=952463516.
B.A/krim

Liviu Lalescu

#102
March 14, 2026, 09:22:06 AM
Between nsrt5 and nsrt6 we only changed a bit. By chance, we obtained the same seed for your testnsrt5. But for other files, seeds are different. You need to compare 10 or more timetables.

Liviu Lalescu

#103
March 14, 2026, 10:13:32 AM
I just arrived home. Writing from the cell phone is a bit more difficult.

I did not download your latest two files - should I? I admit I am a bit lazy to check  :)

I will add the suggestion to add the two constraints from NSRT in the official FET in the TODO and think of it. You know that I want the best for FET, and I am not sure such an addition is perfect for FET.

Are there many successful users of FET-NSRT? Is the problem solved by NSRT a common problem in Algeria?

Benahmed Abdelkrim

#104
March 14, 2026, 10:42:37 AM Last Edit: March 14, 2026, 10:46:56 AM by Benahmed Abdelkrim
Quote from: Liviu Lalescu on March 14, 2026, 10:13:32 AMI did not download your latest two files - should I? I admit I am a bit lazy to check  :)

If you want, especially the last file, it's a little different from the others, why not add it to the examples folder?

Quote from: Liviu Lalescu on March 14, 2026, 10:13:32 AMI will add the suggestion to add the two constraints from NSRT in the official FET in the TODO and think of it. You know that I want the best for FET, and I am not sure such an addition is perfect for FET.
I support this initiative; it would be very beneficial for FET and its users. The problems it solves are common in Algeria and Morocco, and even in other countries, as you can see from the reactions on social media.

I remember that users contributed to spreading NSRT through their contributions, helping to make NSRT easier to understand and use.
@YOUSSEF HOUIET
@Mubarak Mzili
... etc

This last user (Mubarak Mzili) even contributed through videos and an application to facilitate and explain the fundamentals of NSRT.

B.A/krim
Print
Go Up Pages 1 ... 5 6 7 8 9 ... 22
User actions