Find All Matching Elements in a 2D array

[ATE-EGNE]

I'm given a 2D Array of strings where each row indicates two values are connected for example:

{A,B;
 G,'';
 D,C;
 D,A;
 E,F;
 C,B;
 H,''}

Indicates that:

• A, B, C and D are connected
• E and F are connected
• G and H are orphans

From that given input, I would like to make a VI that would output the array

{A,B,C,D;
 E,F,'','';
 G,'','','';
 H,'','',''}

I am using the open G toolkit and currently, I have this solution:

As this is a fairly critical vi I would appreciate a review to:

1. Make sure I didn't miss an edge case (It is safe to assume that the input array will always be NX2 and 'A' = '\sA'='A\s')
2. Suggest any changes I could make to improve the Vi's scope or efficiency.

[altenbach]

Sorry, I don't have any OGTK installed, so I cannot run your example. I would think a variant attribute based solution might be simpler. Have you tried?

[altenbach · Accepted solution]

---

@altenbach wrote:

I would think a variant attribute based solution might be simpler. Have you tried?

---

See if this works for you. 🙂 (No guarantees. Can probably be further simplified).

[Bob_Schor]

The following "breaks" your output (though the algorithm seems to work, it's just the Output that is strange) -- add another pair, (A, E), to your Input Array.  Before running this, predict the output (you'll probably be wrong).

Bob Schor

[mcduff]

Why is the output strange? I do not understand. A is connected to E, E is connected to F, thus A and F are connected by E.

I guess it depends on how you define a match, it appears here anything connected is considered a match.

It can be different depending on your matching criterion. Assume you are matching everything that is 10% within each other and you have the following:

A = 0.9

B = 1

C = 1.1

B & C are within 10% of each other, B is within 10% of A but A is not within 10% of B, so do they match? Much harder to match and get groups. (Sidenote: I tried to do something similar with lengths of DNA fragments and found matches with Mathematica, much more code than CA's solution 🙂  which I am still trying to understand completely.)

mcduff

[altenbach]

---

@Bob_Schor wrote:

The following "breaks" your output

---

Who is "your"?

[Bob_Schor]

---

@altenbach wrote:

---

@Bob_Schor wrote:

The following "breaks" your output

---

Who is "your"?

---

I meant the Original Poster.  I understood the format for output was strings (arrays?) listing the "connected elements", with possibly spaces at the end.  By adding an additional input element that resulted in "merging" two strings, I ended up with an output that seemed to be in the "wrong" format.  Since the Point of the Exercise, I thought, was to have a nice algorithm that worked, rather than try to show that there were no "edge cases" that the OP missed (which is hard, in general), I demonstrated (I thought) the existence of an Edge Case, proving his algorithm was faulty.  Unless I downloaded it wrong ...  or my version of LabVIEW isn't working.

Bob Schor

[ATE-EGNE]

---

@Bob_Schor wrote:

The following "breaks" your output (though the algorithm seems to work, it's just the Output that is strange) -- add another pair, (A, E), to your Input Array.  Before running this, predict the output (you'll probably be wrong).

Bob Schor

---

You are correct! I think I'll try to fix this, but I'll end up going with @altenbach's solution. I did not know you could do that kind of grouping.

[ATE-EGNE]

---

@altenbach wrote:

---

@altenbach wrote:

I would think a variant attribute based solution might be simpler. Have you tried?

---

See if this works for you. 🙂 (No guarantees. Can probably be further simplified).

---

I didn't even know you could do this! Also, if I'm not mistaken this vi would work for an N1xN2 array as long as N2max was known at compile time.

I'd be interested to know if this could be optimized further and debugged. I think I would like to use this logic as it seems to be faster and more versatile.

[ATE-EGNE]

---

@mcduff wrote:

Why is the output strange? I do not understand. A is connected to E, E is connected to F, thus A and F are connected by E.

I guess it depends on how you define a match, it appears here anything connected is considered a match.

It can be different depending on your matching criterion. Assume you are matching everything that is 10% within each other and you have the following:

A = 0.9

B = 1

C = 1.1

B & C are within 10% of each other, B is within 10% of A but A is not within 10% of B, so do they match? Much harder to match and get groups. (Sidenote: I tried to do something similar with lengths of DNA fragments and found matches with Mathematica, much more code than CA's solution 🙂  which I am still trying to understand completely.)

mcduff

---

That's an interesting problem, fortunately for me, being connected is a boolean operation. the comment about the breakage was due to the fact that adding

{'A','E'}

at the end would produce

{'','','A','B','C','D','E','F'}

 Giving me two phantom connection points