25 Stat summations

[Matt Agajanian]

Hi all.

The 25 had eight registers. Since Σ(x^2) was calculated and stored, why was Σ(y^2) missing? Yeah, all the user had to do was x<>y x^2 STO + (register of your choice). But, why not have the 25 do that and store it in one of the remaining registers?

[rprosperi]

In this particular case, I suspect life was short and RAM was full. It would have taken one more non-general purpose register, and Σ(x^2) is used more often. If you had to choose, which would you pick?

[Steve Simpkin]

That makes sense. I also seem to recall that *all* available ROM space was used in the HP-25 code.
The feature set comparable TI-57 came out 2 years after the HP-25 and fully supported dual data point Statistics. It also squeezed its entire logic functionality into *one* integrated circuit with one capacitor and one resistor (vs 5 ICs and more discrete parts for the HP-25). A lot changed in two years!
http://www.datamath.org/Sci/MAJESTIC/JPE....htm#TI-57


Here is a brief list of advantages each model had over the other. There may have been more.

TI-57 advantages:
* Subroutine capability.
* Labels for programs (see next).
* Ability to add/delete instructions. With its use of labels, this allows you to edit a program without resorting to overwriting instructions with NOPs or overwriting large sections of your program because you cannot insert instructions on the HP-25. Even if you could, the lack of labels would likely require changing GTO instruction lines. This comes from a 10 year user of the HP-25 and is not a criticism (the HP-25 delivered far more than it started out with on the drawing board), but a long desired feature.
* Fully supports dual data point Statistics including separate mean, variance and standard deviation calculations for both X and Y data points. The HP-25 partial supports dual data statistics with mean and standard deviation for the X data points
* Can calculate the Variance in Statistics.
* Has a DSZ (Decrement and Skip if Zero) loop test.
* Has "normal" or "standard" display mode that displays only the number of digits after the decimal point that are needed (via FIX 9 or INV FIX).
* Has Xth root function.
* Has an EXC function which can exchange the number in memory N with the displayed number.
* Has 1 additional program step (50 vs 49).
* Lower selling price.

HP-25 advantages:
* All available functions and instructions are printed on the keyboard (except for register math). There are no hidden or potential obscure operations related to the TI-57's use of the INV key.
* Related to the above, register math key sequences are more obvious (STO - 1 instead of INV SUM 1).
* Has two more conditional tests.
* Has Engineering display mode.
* Has forced Scientific display mode.
* Has a % function.
* Can go to an arbitrary step (GTO nn). In a TI-57 you have to declare labels and that takes 1 additional step per label. Note labels in the TI-57 permit insertion/deletion of instructions (see above).
* Both have 8 memory registers (which some of which are used for statistics), but the TI-57 uses Reg 7 for comparisons and sometimes registers 5 & 6 when calculating complex expressions.
* Displays 10-digits when not in SCI/ENG mode (the TI-57 displays 8-digits but uses 11-digits internally)

Differences:
The HP-25 came with a 132-page Owner's Handbook and a 168-page Application Programs book. These were very professionally written and given the selling price of the HP-25 ($195 when introduced in 1975), its intended audience was primarily for technicians, engineers, scientists and graduate students. The Application Programs book includes examples of algebra, number theory, finance, navigation, numerical methods, statistics, surveying, trigonometry. analytical geometry and game programs.

The 244-page TI-57 manual, "Making Tracks into Programming A step-by-step learning guide to the power, ease and fun of using your TI programmable 57" on the other hand, was written in a more informal format. It used the simile of a train to describe how a TI-57 program moved along with branches in the program being shown as alternate "tracks" that the train can take. To me this format seems more approachable to high school students and non-technical people in general. It also includes many program examples of practical everyday math problems as well as financial, mathematics, science and game programs. The TI-57 manual also includes a brief history of calculating/computing devices and a very detailed description of the architecture and internal operation of the calculator.
https://ti57.eu.pythonanywhere.com/stati...amming.pdf

One additional undocumented advantage that the TI-57 has over the HP-25 model (not the HP-25C) is that you can turn the TI-57 display off which then draws so little power from the batteries that it is almost like having a constant memory function.
https://www.rskey.org/gene/calcgene/57c.htm

[rprosperi]

Thanks for the detailed comparison Steve, I did not realize the 57 had so many superior features. I'll admit that even if I did know, I still would have gone with the HP-25 as I did then, as the feel, quality and confidence in use were more important features for me than most of these programming improvments. Still, it does stack up well looking back on it.

Your comments about the different style manuals and TI's broader everyday use examples is also quite interesting. I don't know if they reasoned that the lower cost of the device would enable it to be in the hands of a much broader, and less technically-focused set of users, or if this was a deliberate move to intentionally position it for use by this broader base. Whatever the actual truth, I'm sure the later TI marketing revisionists claimed it was the latter, and indeed it did open up their use to a much broader (and larger!) set of users.

Thanks for enabling this different perspective.

[Matt Agajanian]

I must have missed something. I tried a set of two variable data sets on the 57.

I entered them as x x<>t, y Σ+. Then I pressed the mean and then x<>t and then the variance key and x<>t, it didn’t look like the stats for the y data points were calculated. What am I missing?

[Thomas Klemm]

(08-30-2022 02:20 AM)Matt Agajanian Wrote:  What am I missing?

Reading the manual.

On page 2.22 it states:

• Press [INV] [2nd] [\(\bar{x}\)] to calculate the mean of the "x" data points.
• Press [2nd] [\(\bar{x}\)] to calculate the mean of the "y" data points.
  

That's not intuitive on multiple levels.
But it was probably okay, once you knew it.

---

As a side note:

I was puzzled by the fact that register 7 (or t) contains:

Quote:Last x value entered, plus one.

But later it is mentioned:

Quote:Occasionally, when you're dealing with the analysis of 2 sets of related
data, you'll need to "step up" the x variable by one, for each "y"
variable you enter. (For example, for data collected by year — 1961, 1962,
1963, etc.). In this case you only need to enter the first x value, and your
calculator will automatically increment x by one for each y value
entered.

However, this feature isn't mentioned in chapter 9 (LINEAR REGRESSION), where it could be used.

[Matt Agajanian]

(08-30-2022 03:57 AM)Thomas Klemm Wrote:  

(08-30-2022 02:20 AM)Matt Agajanian Wrote:  What am I missing?

Reading the manual.

On page 2.22 it states:

• Press [INV] [2nd] [\(\bar{x}\)] to calculate the mean of the "x" data points.
• Press [2nd] [\(\bar{x}\)] to calculate the mean of the "y" data points.
  

That's not intuitive on multiple levels.
But it was probably okay, once you knew it.

It does take some mental gymnastics to condition yourself to memorize those keystrokes.

I wonder why TI did not use the same 58/59 x x<>y y Σ+ method for data entry and the same mean & standard deviation output order and the x<>t key on the 57. It seems odd that pressing the mean or variance key would display the y variable results and then INV (mean or variance) would displace the x results. I agree with you, the 57’s strategy is not intuitive. I don’t know why TI didn’t use the x<>t approach on the 57. What is equally odd is why is there a variance function on the keyboard instead of the usual standard deviation function.

[Matt Agajanian]

Massive thanks for the link to the PDF of the 57 manual.

I’ve been looking for it for over quite a few years.

I’ve been able to easily find manuals for the 50, 51, 52, 56, and 58/59.

(08-30-2022 12:23 AM)Steve Simpkin Wrote:  That makes sense. I also seem to recall that *all* available ROM space was used in the HP-25 code.
The feature set comparable TI-57 came out 2 years after the HP-25 and fully supported dual data point Statistics. It also squeezed its entire logic functionality into *one* integrated circuit with one capacitor and one resistor (vs 5 ICs and more discrete parts for the HP-25). A lot changed in two years!
http://www.datamath.org/Sci/MAJESTIC/JPE....htm#TI-57


Here is a brief list of advantages each model had over the other. There may have been more.

TI-57 advantages:
* Subroutine capability.
* Labels for programs (see next).
* Ability to add/delete instructions. With its use of labels, this allows you to edit a program without resorting to overwriting instructions with NOPs or overwriting large sections of your program because you cannot insert instructions on the HP-25. Even if you could, the lack of labels would likely require changing GTO instruction lines. This comes from a 10 year user of the HP-25 and is not a criticism (the HP-25 delivered far more than it started out with on the drawing board), but a long desired feature.
* Fully supports dual data point Statistics including separate mean, variance and standard deviation calculations for both X and Y data points. The HP-25 partial supports dual data statistics with mean and standard deviation for the X data points
* Can calculate the Variance in Statistics.
* Has a DSZ (Decrement and Skip if Zero) loop test.
* Has "normal" or "standard" display mode that displays only the number of digits after the decimal point that are needed (via FIX 9 or INV FIX).
* Has Xth root function.
* Has an EXC function which can exchange the number in memory N with the displayed number.
* Has 1 additional program step (50 vs 49).
* Lower selling price.

HP-25 advantages:
* All available functions and instructions are printed on the keyboard (except for register math). There are no hidden or potential obscure operations related to the TI-57's use of the INV key.
* Related to the above, register math key sequences are more obvious (STO - 1 instead of INV SUM 1).
* Has two more conditional tests.
* Has Engineering display mode.
* Has forced Scientific display mode.
* Has a % function.
* Can go to an arbitrary step (GTO nn). In a TI-57 you have to declare labels and that takes 1 additional step per label. Note labels in the TI-57 permit insertion/deletion of instructions (see above).
* Both have 8 memory registers (which some of which are used for statistics), but the TI-57 uses Reg 7 for comparisons and sometimes registers 5 & 6 when calculating complex expressions.
* Displays 10-digits when not in SCI/ENG mode (the TI-57 displays 8-digits but uses 11-digits internally)

Differences:
The HP-25 came with a 132-page Owner's Handbook and a 168-page Application Programs book. These were very professionally written and given the selling price of the HP-25 ($195 when introduced in 1975), its intended audience was primarily for technicians, engineers, scientists and graduate students. The Application Programs book includes examples of algebra, number theory, finance, navigation, numerical methods, statistics, surveying, trigonometry. analytical geometry and game programs.

The 244-page TI-57 manual, "Making Tracks into Programming A step-by-step learning guide to the power, ease and fun of using your TI programmable 57" on the other hand, was written in a more informal format. It used the simile of a train to describe how a TI-57 program moved along with branches in the program being shown as alternate "tracks" that the train can take. To me this format seems more approachable to high school students and non-technical people in general. It also includes many program examples of practical everyday math problems as well as financial, mathematics, science and game programs. The TI-57 manual also includes a brief history of calculating/computing devices and a very detailed description of the architecture and internal operation of the calculator.
https://ti57.eu.pythonanywhere.com/stati...amming.pdf

One additional undocumented advantage that the TI-57 has over the HP-25 model (not the HP-25C) is that you can turn the TI-57 display off which then draws so little power from the batteries that it is almost like having a constant memory function.
https://www.rskey.org/gene/calcgene/57c.htm

[Matt Agajanian]

I would think that a better comparison would be between the 57 and an HP-33E/C. Both were available at around the same time.

[Steve Simpkin]

(08-30-2022 02:13 AM)rprosperi Wrote:  Thanks for the detailed comparison Steve, I did not realize the 57 had so many superior features. I'll admit that even if I did know, I still would have gone with the HP-25 as I did then, as the feel, quality and confidence in use were more important features for me than most of these programming improvments. Still, it does stack up well looking back on it.

Your comments about the different style manuals and TI's broader everyday use examples is also quite interesting. I don't know if they reasoned that the lower cost of the device would enable it to be in the hands of a much broader, and less technically-focused set of users, or if this was a deliberate move to intentionally position it for use by this broader base. Whatever the actual truth, I'm sure the later TI marketing revisionists claimed it was the latter, and indeed it did open up their use to a much broader (and larger!) set of users.

Thanks for enabling this different perspective.

Bob, I fully agree. I bought my HP-25 in June 1977 and used it as my sole calculator for almost 10 years, through most of High School, College and 2 years at my first engineering job. By the time I replaced it with an HP-11C around early 1987, the world had long moved away from rechargeable battery LED calculators. It just did everything I needed during that time (and still works great!). After previously using algebraic calculators for a time (as it was all I could afford), transitioning to RPN seemed so natural to me that I never went back. And don't get me started on the fantastic style and quality of those early HP calculator manuals.
While I would not have traded my HP-25 for even the vastly feature-superior TI-58 (which sold for the same $100 street price as the HP-25 in late 1977), I am enjoying going through the TI manuals now and learning about how they work.

As for the differences in manual styles, I suspect TI was going for the educational market and used educators to help develop the informal documentation style. The following credit page from the TI-57 manual hints at this association as well.
   

[Steve Simpkin]

(08-30-2022 05:35 AM)Matt Agajanian Wrote:  Massive thanks for the link to the PDF of the 57 manual.
I’ve been looking for it for over quite a few years.
I’ve been able to easily find manuals for the 50, 51, 52, 56, and 58/59.

Glad to help Matt. I was struggling trying to find it as well until I finally discovered the whole title ("Making Tracks into Programming A step-by-step learning guide to the power, ease and fun of using your TI programmable 57") in a flyer for the TI-57.
Here is a color scan of the TI-57 manual. While I find the B/W version easier to read, I like being able to see the original colors used in these documents.

https://ia601901.us.archive.org/26/items...ressed.pdf


Note that TI appears to have included a less comprehensive version of the manual on some TI-57s sold, probably as a cost saving measure. This 94-page version was titled "TI Programmable 57 Introduction to Programming". While it was a full manual, it left out a lot of sections and examples that were in the original.

https://ia803406.us.archive.org/27/items..._300x4.pdf

[Steve Simpkin]

(08-30-2022 05:48 AM)Matt Agajanian Wrote:  I would think that a better comparison would be between the 57 and an HP-33E/C. Both were available at around the same time.

True, the HP-33E added subroutines, linear regression/estimate, two-variable statistics and degrees/radians conversion. It still did not have program step labels or the ability to insert/delete program steps. Of course with only 49 steps and programming that disappeared when you turned off the power (33E), editing may not have been that important on these models. It would certainly become much more important on Continuous Memory models that had more memory.

Also keep in mind that the HP-33E/C models came about a year after the TI-57. When the TI-57 was introduced around May 1977, the HP-25 was its most direct competitor and it was this model that probably shaped the TI-57 feature set.

   

[Dave Britten]

(08-30-2022 07:41 AM)Steve Simpkin Wrote:  

(08-30-2022 05:48 AM)Matt Agajanian Wrote:  I would think that a better comparison would be between the 57 and an HP-33E/C. Both were available at around the same time.

True, the HP-33E added subroutines, linear regression/estimate, two-variable statistics and degrees/radians conversion. It still did not have program step labels or the ability to insert/delete program steps. Of course with only 49 steps and programming that disappeared when you turned off the power (33E), editing may not have been that important on these models. It would certainly become much more important on Continuous Memory models that had more memory.

Also keep in mind that the HP-33E/C models came about a year after the TI-57. When the TI-57 was introduced around May 1977, the HP-25 was its most direct competitor and it was this model that probably shaped the TI-57 feature set.

As an aside, the Fairchild "Programmable Video Game" mentioned in that ad is likely the Channel F, which was the first home video game console to use ROM cartridges for software expandability. Everything prior had been simple pong/tank/etc. types of machines that could only play what was built in.