WP 34s - Latest Version?

[John Smitherman]

Is the current version for the physical calculator 3889?


Regards,

John

[Eric Rechlin]

Yes. I believe there have been some code changes since then, but no new build.

[John Smitherman]

Thanks Eric. I guess at some point Marcus will complete the build process. In the mean time version 3889 is stable and does what I need.


Regards,

John

[Paul Dale]

Marcus is extremely busy, I wouldn't hold out for new firmware.
The firmware has been pretty stable for a while now, so it isn't a big issue.

The only functional changes I see are:

1. Changing the Zeta function reflection point to increase accuracy for some values of its argument.
2. A fix for the orthogonal polynomials for n=1 where the returned value was incorrect.
3. A fix in the error estimate calculation for integrate which was then made irrelevant because the Romberg integration was replaced by a double exponential method.
   

I doubt any of these are overly significant to many people.


Pauli

[BarryMead]

I have a firmware build system on my Ubuntu 14.04 computer, if someone really needs the latest changes let me know and I can email you a copy of the current version of the firmware. I just built the latest 3898 trunk release of the firmware if you need this version just send me your email address in a private message and I will email it to you. The build script automatically increments the version number so the version shown by the Calculator's Version command will show as 3899 not 3898. I created a zip file called wp34s-3899-firmware.zip which contains all of the firmware flavors (with/without XTAL, with/witout IR diode, with/without Libraries etc.)

[Paul Dale]

Barry, I can add you to the sourceforge site if you want to upload images and make them available.


Pauli

[BarryMead]

(01-19-2018 09:17 AM)Paul Dale Wrote:  Barry, I can add you to the sourceforge site if you want to upload images and make them available.

Pauli

Sure, You can send me the login by PM or email. I will zip up all of the binary files and name them something like wp34s-3899-firmware.zip to keep it obviously different from the existing firmware zip file.

To assure the operational status of this build, I just flashed the calc_xtal_full.bin firmware file to one of my wp-34s calculators and all of the quick tests I have run seem to work just fine.

I am very confident in my build system, because I copied it from user "Bit" and verified that my binary files exactly matched his when testing the build system.

[Marcus von Cube]

(01-18-2018 09:47 PM)Paul Dale Wrote:  Marcus is extremely busy, I wouldn't hold out for new firmware.

Not really busy all the time but occupied with many real world problems (upcoming new job, wife's severe health problems). If someone else can reliably create new builds I'd be happy.

It's my first visit in the forum for more than a year. Sadly...

[Massimo Gnerucci]

Marcus please accept my best wishes for your wife's health.

[rprosperi]

Marcus - thanks for stopping by, congrats on new job, and add my prayers for your wife's health.

[Harald]

(01-20-2018 01:39 PM)Marcus von Cube Wrote:  

(01-18-2018 09:47 PM)Paul Dale Wrote:  Marcus is extremely busy, I wouldn't hold out for new firmware.

Not really busy all the time but occupied with many real world problems (upcoming new job, wife's severe health problems). If someone else can reliably create new builds I'd be happy.

It's my first visit in the forum for more than a year. Sadly...

Sorry to hear that Marcus, all the best for you and your wife!

Best regards,
Harald

[Marcus von Cube]

A new, untested, build is on SF.

[John Smitherman]

(01-20-2018 04:30 PM)Marcus von Cube Wrote:  A new, untested, build is on SF.

Thanks Marcus. It downloaded and installed with no issue.

All the best to you and your wife.


Regards,

John

[lrdheat]

Just out of curiosity, what motivated the switch from Romberg to double exp integration? Is it accuracy? speed? How f(x) intervals containing undefined values are handled? Rusty after many years out of the classroom...

[Paul Dale]

I don't remember for sure. I vaguely recollect that the double exponential version performed better over a number of tests and that the relevant code was donated to the project.

We're always happy to accept improvements over the existing implementation.


Pauli

[Didier Lachieze]

(01-21-2018 05:24 AM)lrdheat Wrote:  Just out of curiosity, what motivated the switch from Romberg to double exp integration?

See: [wp34s] New integration program
and: (34S) Integration (using the double-exponential method)

[Paul Dale]

Didier, thanks for the references. They mostly match what I remember happening. Of course, my memory is at fault

Essentially, it gives better results in less time. At least mostly. This is despite the expensive transcendental operations it requires.

I still think it was mostly a win.


Pauli

[emece67]

The benefits are:

• manages discontinuities of the integrand or its derivatives at the integration limits
• accepts infinities as integration limits
• is usually faster than the Romberg method (many times the speedup is dramatic)
  

as a con, the Romberg method seems to better manage such cases where the integrand or its derivatives have discontinuities inside (but not at the ends) of the integration interval.

There's another thread where I describe the basic idea around this algorithm. Beware that this description is for the tanh-sinh method, that does cope neither with semi-infinite nor infinite integration intervals. The program inside the wp34s implements the three tanh-sinh, exp-sinh and sinh-sinh methods (thus the "double exponential" name), thus coping with all three cases of the integration interval.

Although there are, of course, functions that can put this program in trouble, I've found it to be a a good/better tool for everyday work/play.

Regards.

[BarryMead]

I wanted to compile the "Emulator" for this latest 3899 version, but ran into this error
/usr/bin/ld: cannot find -lQtSerial
The loader cannot find the QtSerial library. Something changed since the compile that worked for me back at version 3886 in the QtGui build system.Does Pascal or anyone else know what is going on with the QtEmulator not compiling lately? I am trying to compile on an Ubuntu 16.04 system with all of libraries installed that compile the older 3886 version of the emulator, so I don't know why it won't compile the newer version.

[pascal_meheut]

Same thing on my Mac. I'll have a look at it.