Log Normal and Fortran

[JohnNichols]

Jim:

I wrote a traffic generation program to model 80 years of loads on a road bridge using log normal MKL -- the samples for MKL are painfully old and tattered.  Anyway it uses a typical pattern and about 8 standard vehicles generated a few million times and I get this interesting pattern of the mass of vehicles on the bridge over time, 

see below

I had explained to the client why I thought a simple Gaussian distribution of a single pattern would generate this shape if tested 10000 times, anyway the pattern is interesting, the A and the B are also interesting.  

It is based on a 120 metre bridge and a 50 tonne standard semitrailer.  The X axis is tonnes on the bridge - this is a single lane. 

Just thought you would be interested.  I also used the @mecej4 ideas on types and that made it easy to write in Fortran.  

John

[JohnNichols]

Is it just me, or is there a deluge of new posts from new people at the moment?

Is it related to IFX? 

Apparently I am now permanently on IFX as it is selected automatically, is that normal?  

Dear Barbara:

No matter how I try I cannot get your @Bar.... etc name to pop up when I type in here, it is just not there.

Any ideas?

John

[JohnNichols]

Some of the old people may remember when I did the Fortran mugs in black.  I still use my mug.

I was thinking the other day, we should do a Santa Fortran mug in red, with Steve at the North Pole as Santa Fortran of the .eq. signs. 

[AlHill]

@JohnNichols   Too much leisure time avalable, John?

Doc (not an Intel employee or contractor)
[Maybe Windows 12 will be better]

[JohnNichols]

No, just a fun idea whilst I was waiting for a compile and run to pass. 

I had not heard from you in a while, I was wondering recently if you had passed to the great punch card in the sky, I am pleased to see you are still alive.  

Someone, like David Billingshurst, will come back and say that is not a bad idea and maybe the boat will float again.  

Got have something to do with  my spare time and I need a new coffee mug. 

You too are picking up the Dutch spelling. 

[AlHill]

@JohnNichols   "I was wondering recently if you had passed to the great punch card in the sky, I am pleased to see you are still alive. "

Now I am concerned - have you heard something I haven't?<G>  I still have some dangling chad...

Doc (not an Intel employee or contractor)
[Maybe Windows 12 will be better]

[jimdempseyatthecove]

What is the Y axis?

Jim

[JohnNichols]

Y axis is the total load on the bridge,  I am just checking the math now as I write the report.  I cannot see how it is correct, but it is interesting. 

Dearest Al:

There is an incredibly funny English comedy, where an elderly gentleman remarks that one only discusses hanging chads with one's tailor.  

Interestingly, I hold that opinion too. 

John

[jimdempseyatthecove]

So the X axis is tonnes on the bridge and the Y axis load on the bridge??? Isn't that essentially the same dimensionality?

Jim

[Barbara_P_Intel]

@JohnNichols, @Barbara_P_Intel works ok!

With the 2024.0 release, ifx is the default compiler in VS. You did see the ifort deprecation notice, didn't you?

[JohnNichols]

@Barbara_P_Intel , darn it it worked.

Only men from Texas read instructions, after 20 years of life here, I understand why they need to, but as your typical Australian male, we do not read instructions, unless we are forced to.  

As my American wife once, said, are there any rules you follow?

Rules, I said, what are rules.  She said it with some bitterness. 

[JohnNichols]

Intel® Fortran Compiler Classic (ifort) is now deprecated and will be discontinued in late 2024. Intel recommends that customers transition now to using the LLVM-based Intel® Fortran Compiler (ifx) for continued Windows* and Linux* support, new language support, new language features, and optimizations.

For more information on ifx, see the Intel® Fortran Compiler Developer Guide and Reference and the Porting Guide for ifort Users to ifx.

@Barbara_P_Intel , now I read it.  Thanks. 

Do I get a kewpie doll?

[JohnNichols]

I found out why, if you go past @BA, your name disappears and other bar's pop up. 

[AlHill]

@JohnNichols   No kewpie doll, but how about a cassowary from the Daintree area?

Doc (not an Intel employee or contractor)
[Maybe Windows 12 will be better]

[JohnNichols]

Jim:

Y axis is the count for the X axis load on the bridge.  

I have played with the masses of the trucks and the limits on each category chosen from the log normal returns and made it into a normal look now.  It was a combination of things that caused the shape, interesting really.  

Log normal are not nice to play with, very sensitive to the seed value.   I am looking for the 95% load, which is about 220 tonnes.  

I did 1 million load alternatives and 100 million and they give the same result.   I found the parabola in the log transformation interesting.  

[jimdempseyatthecove]

Bridge simulation (experienced guessing going on here) would exhibit some peculiarities not present in your charts.

Assume you have a 50T truck approaching an otherwise empty bridge. The bridge prior to truck entering the bridge is experiencing its static load (plus possibly environmental loads (wind, rain,...)).

When the front axles of the truck enters the bridge, its load transfers into the bridge, then some short time later the trailing axles enter the bridge and contribute their loads to the bridge. 

An interesting aspect of this (which should be evident in your simulation) is that the load of the axle does not immediately affect the strain into the bridge. There are factors of inertia at play, and then momentum, and the rate at which the strain propagates from the entering end bridge to the exit end of the bridge. These factors will exhibit themselves as a vibration/bounce in the bridge.

Your markers A and B in the earlier charts might be a side effect of the simulation integration step size being in phase (or 180 degrees out of phase) of some internal tuning parameter. .OR. the configuration of the test reached a harmonic of the bridge.

You should be able to investigate this by changing some parameters and then observing the changes in behavior (to see if they confirm you guess as to the cause).

Jim

[JohnNichols]

If we have a railway bridge the standard theory works, because the train and the carriages are heavy and solid, a small truck even 50t is not going to make a large movement in the bridge, often it does not out weigh the thermal motion.  Bridges move a lot - you just cannot see it.  

Acceleration wise the 14th reading is a truck, on the displacement, plotted so it fits and is a bit of a kludge, the acceleration is damped out effectively, now if I hit the bridge with a 400 tonne train - tandem at 800 tonnes and carriages with no separation, I am moving everything in a half mile radius as long as the ground is not stone. 

 

 

[JohnNichols]

There are many dangerous animals in the world, the cassowary is one on the extreme list of things to avoid. 

However, the most dangerous animals for a 3 year old in Africa are a hippo or their step father, about equal.  

We are lucky humans who have had a decent education and a safe life mostly, there are a lot who have not.  

So cassowary, nah, I prefer the desert in the NT, more fun and less dangerous.  

Fortran, does make life better for people and Intel Fortran is the best.