Difference between revisions of "lo'ei"

From Lojban
Jump to: navigation, search
Line 1: Line 1:
'''lo'ei'' is [[User:xorxes|xorxe's]] definition of [[lo'e]], used by him for as the definition of '''lo'e''', and used by others to refer to his usage of '''lo'e'''. ([[Adam]] have changed the original '''lo'e'''s to '''lo'ei'''s for the sake of clarity in the meta-discussion.)
+
'''lo'ei''' is [[User:xorxes|xorxe's]] definition of [[lo'e]], used by him for as the definition of '''lo'e''', and used by others to refer to his usage of '''lo'e'''. ([[Adam]] have changed the original '''lo'e'''s to '''lo'ei'''s for the sake of clarity in the meta-discussion.)
  
 
----
 
----
Line 5: Line 5:
 
''From [http://groups.yahoo.com/group/jboske/message/398 jboske 398]:''
 
''From [http://groups.yahoo.com/group/jboske/message/398 jboske 398]:''
  
I present here my definition of {[[lo'ei|lo'ei]]} (nothing new, but
+
I present here my definition of [[lo'ei]] (nothing new, but maybe better formulated). I don't think it is incompatible with And's, though the approach is different. I define it for a very specific case in a specific context, and then I generalize it.
 
 
maybe better formulated). I don't think it is incompatible
 
 
 
with And's, though the approach is different. I define
 
 
 
it for a very specific case in a specific context, and then
 
 
 
I generalize it.
 
  
 
We start from the official definition of {sisku}:
 
We start from the official definition of {sisku}:
Line 55: Line 47:
 
= tu'o ka ce'u du lo broda
 
= tu'o ka ce'u du lo broda
  
which does not in any way entail that {[[lo'ei|lo'ei]] broda} can
+
which does not in any way entail that '''[[lo'ei]] broda''' can be replaced by '''lo broda''' in other contexts.
 
 
be replaced by {lo broda} in other contexts.
 
  
 
In particular, we have:
 
In particular, we have:
Line 67: Line 57:
 
which are clearly different.
 
which are clearly different.
  
So we have defined {[[lo'ei|lo'ei]] broda} when it appears in the x2 of
+
So we have defined {[[lo'ei|lo'ei]] broda} when it appears in the x2 of '''buska'''. To generalize for any context '''brode [[lo'ei]] broda''', we need a predicate that is to {brode} as {sisku} is to {buska}.
  
{buska}. To generalize for any context {brode [[lo'ei|lo'ei]] broda}, we
+
This protopredicate is simply {kairbrode}. It takes a property in x2 instead of the x2 of brode.
 
 
need a predicate that is to {brode} as {sisku} is to {buska}.
 
 
 
This protopredicate is simply {kairbrode}. It takes a property
 
 
 
in x2 instead of the x2 of brode.
 
  
 
mu'o mi'e [[User:xorxes|xorxes]]
 
mu'o mi'e [[User:xorxes|xorxes]]
Line 81: Line 65:
 
----
 
----
  
''sisku'' is defined by the [[gimste|gimste]] as ''x1 looks for something with property x2''. In principle the most natural definition would be ''x1 looks for object x2'', but the actual definition allows searches for non-existent entities, and treats them identically to existent entities. ''buska'', as defined above, is just this second definition ''x1 looks for object x2''. ''kairbroda'' is the predicate parallel to ''sisku'' for regular predicates; e.g. ''kaircitka'' is ''x1 eats things with property x2''.
+
'''sisku''' is defined by the [[gimste]] as ''x1 looks for something with property x2''. In principle the most natural definition would be ''x1 looks for object x2'', but the actual definition allows searches for non-existent entities, and treats them identically to existent entities. '''buska''', as defined above, is just this second definition ''x1 looks for object x2''. '''kairbroda''' is the predicate parallel to '''sisku''' for regular predicates; e.g. '''kaircitka''' is ''x1 eats things with property x2''.
  
 
----
 
----
  
[[pycyn|pycyn]] has objected (at least at one point or another) to several points about this definition. The ones [[Adam|Adam]] can fathom are:
+
[[pycyn]] has objected (at least at one point or another) to several points about this definition. The ones [[Adam|Adam]] can fathom are:
 
 
# The definition requires that ''le ka ce'u du lo broda'' be the same as ''le ka ce'u du lo'ei broda''. pycyn claimed that, since ''lo broda cu ckaji le ka ce'u du lo broda'' and hence ''lo broda cu ckaji le ka ce'u du lo'ei broda'', therefore ''lo broda cu du lo'ei broda'', to which [[User:xorxes|xorxes]] strenuously objected. It is clear that ''ro da zo'u da ckaji le ka ce'u du da''; [[Adam|Adam]] think that pycyn was trying to claim that moving in the opposite direction is also possible; i.e. ''ro da ro de zo'u ganai da ckaji le ka ce'u du de gi da du de''. As I understand it, [[User:xorxes|xorxes]] objects to this because he takes properties as intensional objects, and the inner form of the sub-bridi is not important, as long as it has the same meaning.
 
 
 
**In fact, ''lo broda cu du lo'ei broda'' is true. What I object to is that this means that ''lo'ei broda'' can be substituted in general by ''lo broda''. This is as wrong as saying that because ''ro broda du lo broda'' is true (which it is) then ''ro broda'' could be substituted by ''lo broda''. --[[User:xorxes|xorxes]]
 
**To see that ''lo broda cu du lo'ei broda'' is true, we just apply the definition: ''lo broda cu du lo'ei broda'' == ''lo broda cu kairdu'o le ka ce'u broda'' == ''lo broda cu ckaji le ka ce'u broda'', since ''kairdu'o'' means "x1 is identical to something with property x2", i.e. it is just ''ckaji''.--[[User:xorxes|xorxes]]
 
 
 
# The definition ultimately has that ''broda lo'ei brode'' is defined as ''kairbroda le ka ce'u du lo'ei brode'' whereas ''broda lo brode'', in terms of ''kairbroda'', is ''da poi brode zo'u kairbroda le ka ce'u du da''. So ''lo brode'' remains bound in the main bridi, whereas ''lo'ei brode'' moves into the sub-bridi and gets bound there. In other words ''broda lo'ei brode'' should be ''ko'a goi lo'ei brode zo'u kairbroda le ka ce'u du ko'a''. [[Adam|Adam]] think that one could say either that binding ''lo'ei brode'' in the main bridi's prenex is acceptable, since ''lo'ei brod'' does not claim that there is any ''lo brode'', or that it doesn't matter where ''lo'ei brode'' is bound; since it is a singular, intensional term, it can move across term and scope boundaries without affecting the meaning.
 
  
** I think what you say is correct, but I'm not sure this was one of pc's objections. --[[User:xorxes|xorxes]]
+
# The definition requires that '''le ka ce'u du lo broda''' be the same as '''le ka ce'u du lo'ei broda'''. pycyn claimed that, since '''lo broda cu ckaji le ka ce'u du lo broda''' and hence '''lo broda cu ckaji le ka ce'u du lo'ei broda''', therefore '''lo broda cu du lo'ei broda''', to which [[User:xorxes|xorxes]] strenuously objected. It is clear that '''ro da zo'u da ckaji le ka ce'u du da'''; [[Adam]] think that pycyn was trying to claim that moving in the opposite direction is also possible; i.e. '''ro da ro de zo'u ganai da ckaji le ka ce'u du de gi da du de'''. As I understand it, [[User:xorxes|xorxes]] objects to this because he takes properties as intensional objects, and the inner form of the sub-bridi is not important, as long as it has the same meaning.
 +
#*In fact, '''lo broda cu du lo'ei broda''' is true. What I object to is that this means that ''lo'ei broda'' can be substituted in general by '''lo broda'''. This is as wrong as saying that because ''ro broda du lo broda'' is true (which it is) then ''ro broda'' could be substituted by '''lo broda'''. --[[User:xorxes|xorxes]]
 +
#*To see that '''lo broda cu du lo'ei broda''' is true, we just apply the definition: '''lo broda cu du lo'ei broda'' == ''lo broda cu kairdu'o le ka ce'u broda''' == '''lo broda cu ckaji le ka ce'u broda''', since '''kairdu'o''' means ''x1 is identical to something with property x2'', i.e. it is just '''ckaji'''.--[[User:xorxes|xorxes]]
 +
# The definition ultimately has that '''broda lo'ei brode''' is defined as '''kairbroda le ka ce'u du lo'ei brode''' whereas '''broda lo brode''', in terms of '''kairbroda''', is '''da poi brode zo'u kairbroda le ka ce'u du da'''. So '''lo brode''' remains bound in the main bridi, whereas '''lo'ei brode''' moves into the sub-bridi and gets bound there. In other words '''broda lo'ei brode''' should be '''ko'a goi lo'ei brode zo'u kairbroda le ka ce'u du ko'a'''. [[Adam|Adam]] think that one could say either that binding '''lo'ei brode''' in the main bridi's prenex is acceptable, since '''lo'ei brod''' does not claim that there is any '''lo brode''', or that it doesn't matter where '''lo'ei brode''' is bound; since it is a singular, intensional term, it can move across term and scope boundaries without affecting the meaning.
 +
#* I think what you say is correct, but I'm not sure this was one of pc's objections. --[[User:xorxes|xorxes]]
  
 
----
 
----
  
This definition of ''lo'ei'' is equivalent in meaning to the second definition of ''[[tu'o|tu'o]]'' as a quantifier given on that page (though it may or may not be exactly equivalent in the algebraic formulae that can be applied to it). In other words, in both ''broda lo'ei brode'' and ''broda tu'o brode'', there is no quantification over the set ''lo'i brode''; the (intensional) meaning of ''brode'' is added to the meaning of ''broda'' without actually going and picking out individual ''brode''s (or masses of ''brode''s, etc.)
+
This definition of '''lo'ei''' is equivalent in meaning to the second definition of '''[[tu'o]]''' as a quantifier given on that page (though it may or may not be exactly equivalent in the algebraic formulae that can be applied to it). In other words, in both '''broda lo'ei brode''' and '''broda tu'o brode''', there is no quantification over the set '''lo'i brode'''; the (intensional) meaning of '''brode'' is added to the meaning of '''broda''' without actually going and picking out individual '''brode'''s (or masses of '''brode'''s, etc.)
  
 
----
 
----
  
[[User:xorxes|xorxes]] has also defined a parallel ''[[le'ei|le'ei]]'', which is his use of ''[[le'e|le'e]]''.
+
[[User:xorxes|xorxes]] has also defined a parallel '''[[le'ei]]''', which is his use of '''[[le'e]]'''.

Revision as of 14:39, 17 August 2019

lo'ei is xorxe's definition of lo'e, used by him for as the definition of lo'e, and used by others to refer to his usage of lo'e. (Adam have changed the original lo'es to lo'eis for the sake of clarity in the meta-discussion.)


From jboske 398:

I present here my definition of lo'ei (nothing new, but maybe better formulated). I don't think it is incompatible with And's, though the approach is different. I define it for a very specific case in a specific context, and then I generalize it.

We start from the official definition of {sisku}:

sisku: x1 seeks/searches/looks for property x2 among set x3

We define a new predicate, {buska}, as follows:

DEF1:

tu'o ka ce'u goi ko'a ce'u goi ko'e ce'u goi ko'i zo'u

ko'a buska ko'e ko'i

cu du tu'o ka ce'u goi ko'a ce'u goi ko'e ce'u goi ko'i zo'u

ko'a sisku tu'o ka ce'u du ko'e kei ko'i

Now we define a particular use of {lo'ei broda} as follows:

DEF2:

buska lo'ei broda

= sisku tu'o ka ce'u broda

Notice that from DEF1 we know that:

buska lo'ei broda = sisku tu'o ka ce'u du lo'ei broda

  • This does not follow directly from DEF1, as pc pointed out during the discussion. The equivalence is still true, but it requires using DEF2 and its generalization to any other selbri. In any case, this is just an illustrative step and not a part of the definition of lo'ei. --xorxes

and from DEF2 we know that:

buska lo'ei broda = sisku tu'o ka ce'u du lo broda

So we have that:

tu'o ka ce'u du lo'ei broda

= tu'o ka ce'u du lo broda

which does not in any way entail that lo'ei broda can be replaced by lo broda in other contexts.

In particular, we have:

buska lo'ei broda = sisku tu'o ka da poi broda zo'u ce'u du da

buska lo broda = da poi broda zo'u sisku tu'o ka ce'u du da

which are clearly different.

So we have defined {lo'ei broda} when it appears in the x2 of buska. To generalize for any context brode lo'ei broda, we need a predicate that is to {brode} as {sisku} is to {buska}.

This protopredicate is simply {kairbrode}. It takes a property in x2 instead of the x2 of brode.

mu'o mi'e xorxes


sisku is defined by the gimste as x1 looks for something with property x2. In principle the most natural definition would be x1 looks for object x2, but the actual definition allows searches for non-existent entities, and treats them identically to existent entities. buska, as defined above, is just this second definition x1 looks for object x2. kairbroda is the predicate parallel to sisku for regular predicates; e.g. kaircitka is x1 eats things with property x2.


pycyn has objected (at least at one point or another) to several points about this definition. The ones Adam can fathom are:

  1. The definition requires that le ka ce'u du lo broda be the same as le ka ce'u du lo'ei broda. pycyn claimed that, since lo broda cu ckaji le ka ce'u du lo broda and hence lo broda cu ckaji le ka ce'u du lo'ei broda, therefore lo broda cu du lo'ei broda, to which xorxes strenuously objected. It is clear that ro da zo'u da ckaji le ka ce'u du da; Adam think that pycyn was trying to claim that moving in the opposite direction is also possible; i.e. ro da ro de zo'u ganai da ckaji le ka ce'u du de gi da du de. As I understand it, xorxes objects to this because he takes properties as intensional objects, and the inner form of the sub-bridi is not important, as long as it has the same meaning.
    • In fact, lo broda cu du lo'ei broda is true. What I object to is that this means that lo'ei broda can be substituted in general by lo broda. This is as wrong as saying that because ro broda du lo broda is true (which it is) then ro broda could be substituted by lo broda. --xorxes
    • To see that lo broda cu du lo'ei broda is true, we just apply the definition: lo broda cu du lo'ei broda == lo broda cu kairdu'o le ka ce'u broda == lo broda cu ckaji le ka ce'u broda, since kairdu'o means x1 is identical to something with property x2, i.e. it is just ckaji.--xorxes
  2. The definition ultimately has that broda lo'ei brode is defined as kairbroda le ka ce'u du lo'ei brode whereas broda lo brode, in terms of kairbroda, is da poi brode zo'u kairbroda le ka ce'u du da. So lo brode remains bound in the main bridi, whereas lo'ei brode moves into the sub-bridi and gets bound there. In other words broda lo'ei brode should be ko'a goi lo'ei brode zo'u kairbroda le ka ce'u du ko'a. Adam think that one could say either that binding lo'ei brode in the main bridi's prenex is acceptable, since lo'ei brod does not claim that there is any lo brode, or that it doesn't matter where lo'ei brode is bound; since it is a singular, intensional term, it can move across term and scope boundaries without affecting the meaning.
    • I think what you say is correct, but I'm not sure this was one of pc's objections. --xorxes

This definition of lo'ei' is equivalent in meaning to the second definition of tu'o as a quantifier given on that page (though it may or may not be exactly equivalent in the algebraic formulae that can be applied to it). In other words, in both broda lo'ei brode and broda tu'o brode, there is no quantification over the set lo'i brode; the (intensional) meaning of brode is added to the meaning of broda without actually going and picking out individual brodes (or masses of brodes, etc.)


xorxes has also defined a parallel le'ei, which is his use of le'e.