Backsolving Transpositions

Questions and answers for COW Macintosh and Windows
Post Reply
MikeCrockett
Posts: 21
Joined: Sun Aug 28, 2016 3:15 pm

Backsolving Transpositions

Post by MikeCrockett »

I'd like to suggest that the BackSolving function add a check-box option to search for possible transpositions before carrying a score up to the parent node.

There is a "Cntl-T - Add Moves that Transpose" option, but that only functions on specific positions where I suspect a transposition is possible and ask for the search. That's not an automatic function.

For accurate backsolving, transpositions should be considered as an automatic feature.
MikeAtBookup
Posts: 187
Joined: Mon Aug 01, 2016 2:02 pm

Re: Backsolving Transpositions

Post by MikeAtBookup »

I've considered making this automatic.

I imagine the madness of deleting a candidate only to have it immediately pop right back up because it, of course transposes. Every deleted candidate would do that. I have not yet thought of an elegant way around that.
-------------------
Mike Leahy
:geek: Head Geek at bookup.com
MikeCrockett
Posts: 21
Joined: Sun Aug 28, 2016 3:15 pm

Re: Backsolving Transpositions

Post by MikeCrockett »

MikeAtBookup wrote:I've considered making this automatic.

I imagine the madness of deleting a candidate only to have it immediately pop right back up because it, of course transposes. Every deleted candidate would do that. I have not yet thought of an elegant way around that.
Perhaps when a candidate is deleted, clear the evaluation as well. Then, when Backsolving a transposition, if it has no evaluation, ignore it.
MikeAtBookup
Posts: 187
Joined: Mon Aug 01, 2016 2:02 pm

Re: Backsolving Transpositions

Post by MikeAtBookup »

I assume you mean clearing the evaluation for the target position of the candidate move that was just deleted. There is no point in doing that as, once a candidate is deleted, the target position cannot be reached/seen by Backsolving.
-------------------
Mike Leahy
:geek: Head Geek at bookup.com
MikeCrockett
Posts: 21
Joined: Sun Aug 28, 2016 3:15 pm

Re: Backsolving Transpositions

Post by MikeCrockett »

MikeAtBookup wrote:I assume you mean clearing the evaluation for the target position of the candidate move that was just deleted. There is no point in doing that as, once a candidate is deleted, the target position cannot be reached/seen by Backsolving.
I guess I misunderstood your explanation that you have no "elegant way around that". My point - there would be nothing to backsolve if the transpositional move has actually been deleted. I didn't understand why a deleted move would even appear as part of a transposition sequence. However, I assumed by your explanation, that you mean the data is "logically deleted" but not "physically deleted". If that is the case, the evaluation it carries up to the parent node should be "logically deleted" as well.

To clarify - My ASK is that you add an option on the Backsolve Options panel to permit the option to insert Transpositional moves. I'm assuming only those Transpositional moves that HAVEN'T been deleted should be inserted. I am NOT asking for you to change how the CNTL-T option works.
MikeAtBookup
Posts: 187
Joined: Mon Aug 01, 2016 2:02 pm

Re: Backsolving Transpositions

Post by MikeAtBookup »

Hey Mike,

You seem to think that a "transpositional" move is different somehow from other moves. All moves are the same.

There is also nothing in the system that marks a move as having been deleted. If it's been deleted, it's just as "gone" as any other move that hasn't been entered. It would be like asking a word processing program to spellcheck only the deleted words. They... don't exist anymore.

Adding all moves that transpose could be added as a new function. I agree that having all transposing moves increases the accuracy of backsolving. The only downside is speed. Each legal move in each position would have to be made and then the target position would be searched for. That's a lot of extra searches. I'll add it to my list.
-------------------
Mike Leahy
:geek: Head Geek at bookup.com
Frankly123
Posts: 5
Joined: Wed Jan 25, 2017 10:16 pm

Re: Backsolving Transpositions

Post by Frankly123 »

On a separate but related note, specifically on transpositions alone (no backsolve), I have found that when importing an ebook that contains transpositions, that the "T" symbol for Transpose does NOT show up after the import is complete (although all the moves do show up as always).

Also, the Command "Add moves that transpose" or Ctrl-T results with the message "No moves were found that transposed to any existing positions."
This same message also occurs in the base ebook from which the import is requested and which does show the "T" when stepping through the moves.

A simple example shows this from the video "How to find transpositions" where the example is using the French Defense with both 1. e4 and 1. d4 as the first moves.
The "T" only shows up when manually entering the moves (or perhaps when loading PGN files) but NOT when importing from one ebook to another.
Is this a bug? Or am I doing something wrong? (use pgn files as a workaround for now?)

Note: the conversion of a legacy (pre-2016) ebook into the 2016 Pro version works fine - the "T" shows up now in bright red!

Thanks in advance,

Frank
Frankly123
Posts: 5
Joined: Wed Jan 25, 2017 10:16 pm

Re: Backsolving Transpositions

Post by Frankly123 »

PGNs exported from an ebook with "T" showing up for Transpositions and then imported back into a newly created ebook DO NOT seem to work using the simple example of the French Defense described below.

Is it me? What am I doing wrong?

Frank
Post Reply