ó
‡oŠ\c@s?dd
lZ
ddlmZ
d„Zdef
d„ƒYZd
S(iÿÿÿÿNi
(
t
ensure_rngc

Cstt
t|ƒƒ
S(
s3 ensure that an point is hashable by a python dict (ttupletmaptfloat(
t
x((s$N:\TCAD\SLALOM\Bayes\target_space.pyt	_hashablestTargetSpacecBs×eZ
dZdd
„
Zd„Zd„Zed„ƒ
Zed„ƒ
Z	ed„ƒ
Z
ed„ƒ
Zed„ƒ
Zed	„ƒ
Z
d
„Zd„Zd„Zd
„Zd„Zd„Zd„Zd„Zd„ZRS(sÓ

    Holds the param-space coordinates (X) and target values (Y)
    Allows for constant-time appends while ensuring no duplicates are added

    Example
    -------
    >>> def target_func(p1, p2):
    >>>     return p1 + p2
    >>> pbounds = {'p1': (0, 1), 'p2': (1, 100)}
    >>> space = TargetSpace(target_func, pbounds, random_state=0)
    >>> x = space.random_points(1)[0]
    >>> y = space.register_point(x)
    >>> assert self.max_point()['max_val'] == y
    cCs®t|ƒ
|_
|
|_t|ƒ
|_tjgt|jƒd
d„ƒ

D]}|d^qIdtjƒ

|_	tj
dd|jfƒ
|_tj
ddƒ
|_
i|_dS(sp

        Parameters
        ----------
        target_func : function
            Function to be maximized.

        pbounds : dict
            Dictionary with parameters names as keys and a tuple with minimum
            and maximum values.

        random_state : int, RandomState, or None
            optionally specify a seed for a random number generator
        tkeyc

Ss|d
S(Ni((
R((s$N:\TCAD\SLALOM\Bayes\target_space.pyt<lambda>0si
tdtypetshapeiN(Rtrandom_statettarget_functsortedt_keystnptarraytitemsRt_boundstemptytdimt_paramst_targett_cache(tselfRtpboundsRtitem((s$N:\TCAD\SLALOM\Bayes\target_space.pyt__init__s	
2

cCst|
ƒ
|j
kS(
N(RR(RR((s$N:\TCAD\SLALOM\Bayes\target_space.pyt__contains__;s
c

Cs1t|j
ƒ
t|jƒ
ks$t‚
t|jƒ
S(
N(tlenRRtAssertionError(
R((s$N:\TCAD\SLALOM\Bayes\target_space.pyt__len__>s
$
c

Cst|ƒ
d
kS(Ni(
R(
R((s$N:\TCAD\SLALOM\Bayes\target_space.pyRBsc


Cs|jS(
N(
R(
R((s$N:\TCAD\SLALOM\Bayes\target_space.pytparamsFsc


Cs|jS(
N(
R(
R((s$N:\TCAD\SLALOM\Bayes\target_space.pyttargetJsc

Cs
t|j
ƒ
S(
N(RR(
R((s$N:\TCAD\SLALOM\Bayes\target_space.pyRNsc


Cs|jS(
N(
R(
R((s$N:\TCAD\SLALOM\Bayes\target_space.pytkeysRsc


Cs|jS(
N(
R(
R((s$N:\TCAD\SLALOM\Bayes\target_space.pytboundsVscCs‹y%t|
ƒ
t|j
ƒ
ks$t‚
Wn<tk
rc


td
jt|
ƒ
ƒ
dj|j
ƒ
ƒ
‚
n
Xtjg|j
D]}|
|^qtƒ
S(NsParameters' keys ({}) do s(not match the expected set of keys ({}).(tsetR"Rt
ValueErrortformatR
Rtasarray(RR R((s$N:\TCAD\SLALOM\Bayes\target_space.pytparams_to_arrayZs

%




cCs€y%t|
ƒ
t|j
ƒ
ks$t‚
WnBtk
ri


td
jt|
ƒ
ƒ
djt|j
ƒ
ƒ
ƒ
‚
n
Xtt|j
|
ƒƒ
S(Ns)Size of array ({}) is different than the s#expected number of parameters ({}).(RR"RR%R&tdicttzip(RR((s$N:\TCAD\SLALOM\Bayes\target_space.pytarray_to_paramsds

%




 cCs­ytj
|
d
tƒ

}
Wn tk
r;


|j|
ƒ
}
n
X|
jƒ}
y|
j|jksct‚
WnBtk
r¨


t	dj
t|
ƒ
ƒ
dj
t|jƒ
ƒ
ƒ
‚
n
X|
S(NR	s)Size of array ({}) is different than the s#expected number of parameters ({}).(
RR'Rt	TypeErrorR(traveltsizeRRR%R&RR"(RR((s$N:\TCAD\SLALOM\Bayes\target_space.pyt	_as_arrayns











 cCs•|j|
ƒ
}||kr3t
d
j|ƒ
ƒ
‚
n||jt|jƒƒ
<tj|j|j	ddƒgƒ
|_tj|j
|g
gƒ
|_
dS(s¾
        Append a point and its target value to the known data.

        Parameters
        ----------
        x : ndarray
            a single point, with len(x) == self.dim

        y : float
            target function value

        Raises
        ------
        KeyError:
            if the point is not unique

        Notes
        -----
        runs in ammortized constant time

        Example
        -------
        >>> pbounds = {'p1': (0, 1), 'p2': (1, 100)}
        >>> space = TargetSpace(lambda p1, p2: p1 + p2, pbounds)
        >>> len(space)
        0
        >>> x = np.array([0, 0])
        >>> y = 1
        >>> space.add_observation(x, y)
        >>> len(space)
        1
        sData point {} is not uniquei
iÿÿÿÿN(R/tKeyErrorR&RRR-RtconcatenateRtreshapeR(RR R!R((s$N:\TCAD\SLALOM\Bayes\target_space.pytregister~s!

'
cCsu|j|
ƒ
}y|j
t|ƒ
}WnHtk
rp


tt|j|ƒƒ
}
|j|
}|j||ƒ
n
X|S(
s—

        Evaulates a single point x, to obtain the value y and then records them
        as observations.

        Notes
        -----
        If x has been previously seen returns a cached value of y.

        Parameters
        ----------
        x : ndarray
            a single point, with len(x) == self.dim

        Returns
        -------
        y : float
            target function value.
        (	R/RRR0R)R*RRR3(RR RR!((s$N:\TCAD\SLALOM\Bayes\target_space.pytprobe©s






c
Csjtj
d
|jfƒ
}
xEt|jƒ
D]4\}\}}|jj||dd
ƒ
|
j|<q(W|
jƒS(sò

        Creates random points within the bounds of the space.

        Returns
        ----------
        data: ndarray
            [num x dim] array points with dimensions corresponding to `self._keys`

        Example
        -------
        >>> target_func = lambda p1, p2: p1 + p2
        >>> pbounds = {'p1': (0, 1), 'p2': (1, 100)}
        >>> space = TargetSpace(target_func, pbounds, random_state=0)
        >>> space.random_points(1)
        array([[ 55.33253689,   0.54488318]])
        i
R.(	RRRt	enumerateRRtuniformt
TR-(Rtdatatcoltlowertupper((s$N:\TCAD\SLALOM\Bayes\target_space.pyt
random_sampleÆs
"
&
c
CsayCi|jj
ƒd
6tt|j|j|jjƒƒƒ
d6}
Wntk
r\


i}
n
X|
S(s;Get maximum target value found and corresponding parametes.R!R (R!tmaxR)R*R"R targmaxR%(Rtres((s$N:\TCAD\SLALOM\Bayes\target_space.pyR=Ýs



-



c
Cseg|jD]}
t
t|j|
ƒƒ
^q
}gt|j|ƒD] \}}i|d
6|d6^qAS(s8Get all target values found and corresponding parametes.R!R (R R)R*R"R!(Rt
pR R!tparam((s$N:\TCAD\SLALOM\Bayes\target_space.pyR?ês.
cCsDx=t|j
ƒ
D],\}}||
kr|
||j|<qqWd
S(sÙ
        A method that allows changing the lower and upper searching bounds

        Parameters
        ----------
        new_bounds : dict
            A dictionary with the parameter name and its new bounds
        N(R5R"R(Rt
new_boundstrowR((s$N:\TCAD\SLALOM\Bayes\target_space.pyt
set_boundsós	

N(t__name__t
__module__t__doc__tNoneRRRtpropertyRR R!RR"R#R(R+R/R3R4R<R=R?RD(((s$N:\TCAD\SLALOM\Bayes\target_space.pyR
s&
"			
	
		+			
		(tnumpyRtutilRRtobjectR(((s$N:\TCAD\SLALOM\Bayes\target_space.pyt<module>
s