Home My Page Projects Code Snippets Project Openings diderot
Summary Activity Tracker Tasks SCM

SCM Repository

[diderot] Annotation of /branches/lamont/TODO
ViewVC logotype

Annotation of /branches/lamont/TODO

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1204 - (view) (download)
Original Path: trunk/TODO

1 : glk 1162 NOTE: GLK's approximate ranking of 8 most important tagged with
2 : jhr 1115 [GLK:1], [GLK:2], ...
3 :    
4 : glk 1156 ========================
5 :     SHORT TERM ============= (*needed* for streamlines & tractography)
6 :     ========================
7 : jhr 1115
8 : glk 1162 [GLK:3] Add sequence types (needed for evals & evecs)
9 : jhr 1115 syntax
10 :     types: ty '{' INT '}'
11 :     value construction: '{' e1 ',' … ',' en '}'
12 :     indexing: e '{' e '}'
13 :    
14 : glk 1162 [GLK:4] evals & evecs for symmetric tensor[2,2] and
15 :     tensor[3,3] (requires sequences)
16 :    
17 : glk 1156 ability to emit/track/record variables into dynamically re-sized
18 :     runtime buffer
19 : jhr 1115
20 : glk 1156 tensor fields: convolution on general tensor images
21 : jhr 1115
22 : glk 1156 ========================
23 : glk 1162 SHORT-ISH TERM ========= (to make using Diderot less annoying to
24 :     ======================== program in, and slow to execute)
25 : jhr 1115
26 : jhr 1165 value-numbering optimization [DONE, but needs more testing]
27 : jhr 1115
28 : glk 1167 Allow ".ddro" file extensions in addition to ".diderot"
29 :    
30 : glk 1204 Be able to output values of type tensor[2,2] and tensor[3,3];
31 :     (currently only scalars & vectors). Want to add some regression tests
32 :     based on this and currently can't
33 : glk 1167
34 : glk 1162 [GLK:1] Add a clamp function, which takes three arguments; either
35 :     three scalars:
36 :     clamp(lo, hi, x) = max(lo, min(hi, x))
37 :     or three vectors of the same size:
38 :     clamp(lo, hi, [x,y]) = [max(lo[0], min(hi[0], x)),
39 :     max(lo[1], min(hi[1], y))]
40 :     This would be useful in many current Diderot programs.
41 :     One question: clamp(x, lo, hi) is the argument order used in OpenCL
42 :     and other places, but clamp(lo, hi, x) is much more consistent with
43 :     lerp(lo, hi, x), hence GLK's preference
44 : jhr 1115
45 : glk 1162 [GLK:2] Proper handling of stabilize method
46 :    
47 :     allow "*" to represent "modulate": per-component multiplication of
48 :     vectors, and vectors only (not tensors of order 2 or higher). Once
49 :     sequences are implemented this should be removed: the operation is not
50 :     invariant WRT basis so it is not a legit vector computation.
51 :    
52 :     implicit type promotion of integers to reals where reals are
53 :     required (e.g. not exponentiation "^")
54 :    
55 :     [GLK:5] Save Diderot output to nrrd, instead of "mip.txt"
56 : jhr 1115 For grid of strands, save to similarly-shaped array
57 :     For list of strands, save to long 1-D (or 2-D for non-scalar output) list
58 :     For ragged things (like tractography output), will need to save both
59 :     complete list of values, as well as list of start indices and lengths
60 :     to index into complete list
61 :    
62 : glk 1162 [GLK:6] Use of Teem's "hest" command-line parser for getting
63 : jhr 1115 any input variables that are not defined in the source file
64 :    
65 : glk 1162 [GLK:7] ability to declare a field so that probe positions are
66 : glk 1120 *always* "inside"; with various ways of mapping the known image values
67 :     to non-existant index locations. One possible syntax emphasizes that
68 :     there is a index mapping function that logically precedes convolution:
69 : glk 1162 F = bspln3 ⊛ (img ◦ clamp)
70 : glk 1120 F = bspln3 ⊛ (img ◦ repeat)
71 :     F = bspln3 ⊛ (img ◦ mirror)
72 :     where "◦" or "∘" is used to indicate function composition
73 : jhr 1115
74 : glk 1162 Level of differentiability in field type should be statement about how
75 :     much differentiation the program *needs*, rather than what the kernel
76 :     *provides*. The needed differentiability can be less than or equal to
77 :     the provided differentiability.
78 :    
79 : glk 1156 Use ∇⊗ etc. syntax
80 :     syntax [DONE]
81 :     typechecking
82 :     IL and codegen
83 : jhr 1115
84 : glk 1156 Add type aliases for color types
85 :     rgb = real{3}
86 :     rgba = real{4}
87 : jhr 1115
88 :     ==============================
89 : glk 1156 MEDIUM TERM ================== (*needed* for particles)
90 : jhr 1115 ==============================
91 :    
92 :     run-time birth of strands
93 :    
94 :     "initially" supports lists
95 :    
96 :     "initially" supports lists of positions output from
97 :     different initalization Diderot program
98 :    
99 : glk 1156 Communication between strands: they have to be able to learn each
100 :     other's state (at the previous iteration). Early version of this can
101 :     have the network of neighbors be completely static (for running one
102 :     strand/pixel image computations). Later version with strands moving
103 :     through the domain will require some spatial data structure to
104 :     optimize discovery of neighbors.
105 : jhr 1115
106 : glk 1156 ============================
107 :     MEDIUM-ISH TERM ============ (to make Diderot more useful/effective)
108 :     ============================
109 : jhr 1115
110 : glk 1156 Python/ctypes interface to run-time
111 : jhr 1115
112 : glk 1156 support for Python interop and GUI
113 : jhr 1115
114 : glk 1162 Allow integer exponentiation ("^2") to apply to square matrices,
115 :     to represent repeated matrix multiplication
116 :    
117 : glk 1156 Alow X *= Y, X /= Y, X += Y, X -= Y to mean what they do in C,
118 :     provided that X*Y, X/Y, X+Y, X-Y are already supported.
119 :     Nearly every Diderot program would be simplified by this.
120 : jhr 1115
121 : glk 1156 Put small 1-D and 2-D fields, when reconstructed specifically by tent
122 :     and when differentiation is not needed, into faster texture buffers.
123 :     test/illust-vr.diderot is good example of program that uses multiple
124 :     such 1-D fields basically as lookup-table-based function evaluation
125 :    
126 :     expand trace in mid to low translation
127 :    
128 :     extend norm (|exp|) to all tensor types [DONE for vectors and matrices]
129 :    
130 :     determinant ("det") for tensor[3,3]
131 :    
132 : jhr 1115 add ":" for tensor dot product (contracts out two indices
133 :     instead of one like •), valid for all pairs of tensors with
134 :     at least two indices
135 :    
136 : glk 1156 test/uninit.diderot:
137 :     documents need for better compiler error messages when output variables
138 :     are not initialized; the current messages are very cryptic
139 : jhr 1115
140 :     want: warnings when "D" (reserved for differentiation) is declared as
141 :     a variable name (get confusing error messages now)
142 :    
143 : glk 1156 ==============================
144 :     LONG TERM ==================== (make Diderot more interesting/attractive from
145 :     ============================== a research standpoint)
146 : jhr 1115
147 : glk 1156 IL support for higher-order tensor values (matrices, etc).
148 :     tensor construction [DONE]
149 :     tensor indexing [DONE]
150 :     tensor slicing
151 :     verify that hessians work correctly [DONE]
152 : jhr 1115
153 : glk 1156 Better handling of variables that determines the scope of a variable
154 :     based on its actual use, instead of where the user defined it. So,
155 :     for example, we should lift strand-invariant variables to global
156 :     scope. Also prune out useless variables, which should include field
157 :     variables after the translation to mid-il.
158 :    
159 :     test/vr-kcomp2.diderot: Add support for code like
160 :     (F1 if x else F2)@pos
161 :     This will require duplication of the continuation of the conditional
162 :     (but we should only duplicate over the live-range of the result of the
163 :     conditional.
164 :    
165 : glk 1162 [GLK:8] Want: non-trivial field expressions & functions.
166 :     scalar fields from scalar fields F and G:
167 :     field#0(2)[] X = (sin(F) + 1.0)/2;
168 :     field#0(2)[] X = F*G;
169 :     scalar field of vector field magnitude:
170 : glk 1156 image(2)[2] Vimg = load(...);
171 :     field#0(2)[] Vlen = |Vimg ⊛ bspln3|;
172 : glk 1162 field of normalized vectors (for LIC and vector field feature extraction)
173 :     field#2(2)[2] F = ...
174 :     field#0(2)[2] V = normalize(F);
175 :     scalar field of gradient magnitude (for edge detection))
176 : glk 1156 field#2(2)[] F = Fimg ⊛ bspln3;
177 :     field#0(2)[] Gmag = |∇F|;
178 : glk 1162 scalar field of squared gradient magnitude (simpler to differentiate):
179 : glk 1156 field#2(2)[] F = Fimg ⊛ bspln3;
180 :     field#0(2)[] Gmsq = ∇F•∇F;
181 : glk 1162 There is value in having these, even if the differentiation of them is
182 :     not supported (hence the indication of "field#0" for these above)
183 : glk 1156
184 : jhr 1195 Introduce region types (syntax region(d), where d is the dimension of the
185 :     region. One useful operator would be
186 :     dom : field#k(d)[s] -> region(d)
187 :     Then the inside test could be written as
188 :     pos ∈ dom(F)
189 :     We could further extend this approach to allow geometric definitions of
190 :     regions. It might also be useful to do inside tests in world space,
191 :     instead of image space.
192 :    
193 : glk 1162 co- vs contra- index distinction
194 : glk 1156
195 : glk 1162 Permit field composition:
196 : glk 1156 field#2(3)[3] warp = bspln3 ⊛ warpData;
197 :     field#2(3)[] F = bspln3 ⊛ img;
198 :     field#2(3)[] Fwarp = F ◦ warp;
199 : glk 1162 So Fwarp(x) = F(warp(X)). Chain rule can be used for differentation.
200 :     This will be instrumental for expressing non-rigid registration
201 :     methods (but those will require co-vs-contra index distinction)
202 : glk 1156
203 : glk 1155 Allow the convolution to be specified either as a single 1D kernel
204 :     (as we have it now):
205 :     field#2(3)[] F = bspln3 ⊛ img;
206 :     or, as a tensor product of kernels, one for each axis, e.g.
207 :     field#0(3)[] F = (bspln3 ⊗ bspln3 ⊗ tent) ⊛ img;
208 :     This is especially important for things like time-varying data, or
209 :     other multi-dimensional fields where one axis of the domain is very
210 : glk 1162 different from the rest, and hence must be treated separately when
211 :     it comes to convolution. What is very unclear is how, in such cases,
212 : glk 1155 we should notate the gradient, when we only want to differentiate with
213 : glk 1162 respect to some subset of the axes. One ambitious idea would be:
214 :     field#0(3)[] Ft = (bspln3 ⊗ bspln3 ⊗ tent) ⊛ img; // 2D time-varying field
215 : glk 1204 field#0(2)[] F = lambda([x,y], Ft([x,y,42.0])) // restriction to time=42.0
216 :     vec2 grad = ∇F([x,y]); // 2D gradient
217 : glk 1155
218 : glk 1204 Tensors of order 3 (e.g. gradients of diffusion tensor fields, or
219 :     hessians of vector fields) and order 4 (e.g. Hessians of diffusion
220 :     tensor fields).
221 :    
222 : glk 1162 representation of tensor symmetry
223 : jhr 1115 (have to identify the group of index permutations that are symmetries)
224 :    
225 :     dot works on all tensors
226 :    
227 :     outer works on all tensors
228 :    
229 : glk 1204 Help for debugging Diderot programs: need to be able to uniquely
230 :     identify strands, and for particular strands that are known to behave
231 :     badly, do something like printf or other logging of their computations
232 :     and updates.
233 :    
234 :     Permit writing dimensionally general code: Have some statement of the
235 :     dimension of the world "W" (or have it be learned from one particular
236 :     field of interest), and then able to write "vec" instead of
237 :     "vec2/vec3", and perhaps "tensor[W,W]" instead of
238 :     "tensor[2,2]/tensor[3,3]"
239 :    
240 :     Traits: all things things that have boilerplate code (especially
241 :     volume rendering) should be expressed in terms of the unique
242 :     computational core. Different kinds of streamline/tractography
243 :     computation will be another example, as well as particle systems.
244 :    
245 : jhr 1115 Einstein summation notation
246 :    
247 :     "tensor comprehension" (like list comprehension)
248 :    
249 : glk 1204 Fields coming from different sources of data:
250 :     * triangular or tetrahedral meshes over 2D or 3D domains (of the
251 :     source produced by finite-element codes; these will come with their
252 :     own specialized kinds of reconstruction kernels, called "basis
253 :     functions" in this context)
254 :     * Large point clouds, with some radial basis function around each point,
255 :     which will be tuned by parameters of the point (at least one parameter
256 :     giving some notion of radius)
257 :    
258 : jhr 1115 ======================
259 :     BUGS =================
260 :     ======================
261 :    
262 :     test/zslice2.diderot:
263 :     // HEY (bug) bspln5 leads to problems ...
264 :     // uncaught exception Size [size]
265 :     // raised at c-target/c-target.sml:47.15-47.19
266 :     //field#4(3)[] F = img ⊛ bspln5;

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0